id
stringlengths 9
16
| submitter
stringlengths 3
64
⌀ | authors
stringlengths 5
6.63k
| title
stringlengths 7
245
| comments
stringlengths 1
482
⌀ | journal-ref
stringlengths 4
382
⌀ | doi
stringlengths 9
151
⌀ | report-no
stringclasses 984
values | categories
stringlengths 5
108
| license
stringclasses 9
values | abstract
stringlengths 83
3.41k
| versions
listlengths 1
20
| update_date
timestamp[s]date 2007-05-23 00:00:00
2025-04-11 00:00:00
| authors_parsed
listlengths 1
427
| prompt
stringlengths 166
3.49k
| label
stringclasses 2
values | prob
float64 0.5
0.98
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2503.05162 | Chao Zhang | Chao Zhang, Yifeng Zhou, Shuheng Wang, Wenfa Li, Degang Wang, Yi Xu,
Shaohui Jiao | EvolvingGS: High-Fidelity Streamable Volumetric Video via Evolving 3D
Gaussian Representation | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We have recently seen great progress in 3D scene reconstruction through
explicit point-based 3D Gaussian Splatting (3DGS), notable for its high quality
and fast rendering speed. However, reconstructing dynamic scenes such as
complex human performances with long durations remains challenging. Prior
efforts fall short of modeling a long-term sequence with drastic motions,
frequent topology changes or interactions with props, and resort to segmenting
the whole sequence into groups of frames that are processed independently,
which undermines temporal stability and thereby leads to an unpleasant viewing
experience and inefficient storage footprint. In view of this, we introduce
EvolvingGS, a two-stage strategy that first deforms the Gaussian model to
coarsely align with the target frame, and then refines it with minimal point
addition/subtraction, particularly in fast-changing areas. Owing to the
flexibility of the incrementally evolving representation, our method
outperforms existing approaches in terms of both per-frame and temporal quality
metrics while maintaining fast rendering through its purely explicit
representation. Moreover, by exploiting temporal coherence between successive
frames, we propose a simple yet effective compression algorithm that achieves
over 50x compression rate. Extensive experiments on both public benchmarks and
challenging custom datasets demonstrate that our method significantly advances
the state-of-the-art in dynamic scene reconstruction, particularly for extended
sequences with complex human performances.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 06:01:07 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Zhang",
"Chao",
""
],
[
"Zhou",
"Yifeng",
""
],
[
"Wang",
"Shuheng",
""
],
[
"Li",
"Wenfa",
""
],
[
"Wang",
"Degang",
""
],
[
"Xu",
"Yi",
""
],
[
"Jiao",
"Shaohui",
""
]
]
| TITLE: EvolvingGS: High-Fidelity Streamable Volumetric Video via Evolving 3D
Gaussian Representation
ABSTRACT: We have recently seen great progress in 3D scene reconstruction through
explicit point-based 3D Gaussian Splatting (3DGS), notable for its high quality
and fast rendering speed. However, reconstructing dynamic scenes such as
complex human performances with long durations remains challenging. Prior
efforts fall short of modeling a long-term sequence with drastic motions,
frequent topology changes or interactions with props, and resort to segmenting
the whole sequence into groups of frames that are processed independently,
which undermines temporal stability and thereby leads to an unpleasant viewing
experience and inefficient storage footprint. In view of this, we introduce
EvolvingGS, a two-stage strategy that first deforms the Gaussian model to
coarsely align with the target frame, and then refines it with minimal point
addition/subtraction, particularly in fast-changing areas. Owing to the
flexibility of the incrementally evolving representation, our method
outperforms existing approaches in terms of both per-frame and temporal quality
metrics while maintaining fast rendering through its purely explicit
representation. Moreover, by exploiting temporal coherence between successive
frames, we propose a simple yet effective compression algorithm that achieves
over 50x compression rate. Extensive experiments on both public benchmarks and
challenging custom datasets demonstrate that our method significantly advances
the state-of-the-art in dynamic scene reconstruction, particularly for extended
sequences with complex human performances.
| no_new_dataset | 0.945197 |
2503.05164 | Jiangtao Gong | Shanhe You, Xuewen Luo, Xinhe Liang, Jiashu Yu, Chen Zheng and
Jiangtao Gong | A Comprehensive LLM-powered Framework for Driving Intelligence
Evaluation | 8 pages, 3 figures | ICRA2025 | null | null | cs.RO cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Evaluation methods for autonomous driving are crucial for algorithm
optimization. However, due to the complexity of driving intelligence, there is
currently no comprehensive evaluation method for the level of autonomous
driving intelligence. In this paper, we propose an evaluation framework for
driving behavior intelligence in complex traffic environments, aiming to fill
this gap. We constructed a natural language evaluation dataset of human
professional drivers and passengers through naturalistic driving experiments
and post-driving behavior evaluation interviews. Based on this dataset, we
developed an LLM-powered driving evaluation framework. The effectiveness of
this framework was validated through simulated experiments in the CARLA urban
traffic simulator and further corroborated by human assessment. Our research
provides valuable insights for evaluating and designing more intelligent,
human-like autonomous driving agents. The implementation details of the
framework and detailed information about the dataset can be found at Github.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 06:03:02 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"You",
"Shanhe",
""
],
[
"Luo",
"Xuewen",
""
],
[
"Liang",
"Xinhe",
""
],
[
"Yu",
"Jiashu",
""
],
[
"Zheng",
"Chen",
""
],
[
"Gong",
"Jiangtao",
""
]
]
| TITLE: A Comprehensive LLM-powered Framework for Driving Intelligence
Evaluation
ABSTRACT: Evaluation methods for autonomous driving are crucial for algorithm
optimization. However, due to the complexity of driving intelligence, there is
currently no comprehensive evaluation method for the level of autonomous
driving intelligence. In this paper, we propose an evaluation framework for
driving behavior intelligence in complex traffic environments, aiming to fill
this gap. We constructed a natural language evaluation dataset of human
professional drivers and passengers through naturalistic driving experiments
and post-driving behavior evaluation interviews. Based on this dataset, we
developed an LLM-powered driving evaluation framework. The effectiveness of
this framework was validated through simulated experiments in the CARLA urban
traffic simulator and further corroborated by human assessment. Our research
provides valuable insights for evaluating and designing more intelligent,
human-like autonomous driving agents. The implementation details of the
framework and detailed information about the dataset can be found at Github.
| new_dataset | 0.961425 |
2503.05167 | Ruotai Li | Xinhan Zheng and Huyu Wu and Haopeng Jin and Ruotai Li | FMCHS: Advancing Traditional Chinese Medicine Herb Recommendation with
Fusion of Multiscale Correlations of Herbs and Symptoms | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Traditional Chinese medicine (TCM) exhibits remarkable therapeutic efficacy
in disease treatment and healthcare through personalized herb prescriptions.
However, current herb recommendation models inadequately capture the multiscale
relations between herbs and clinical symptoms, particularly neglecting latent
correlations at the chemical-molecular scale. To address these limitations, we
propose the Fusion of Multiscale Correlations of Herbs and Symptoms (FMCHS), an
innovative framework that synergistically integrates molecular-scale chemical
characteristics of herbs with clinical symptoms. The framework employs
multi-relational graph transformer layers to generate enriched embeddings that
preserve both structural and semantic features within herbs and symptoms.
Through systematic incorporation of herb chemical profiles into node embeddings
and implementation of attention-based feature fusion, FMCHS effectively
utilizes multiscale correlations. Comprehensive evaluations demonstrate FMCHS's
superior performance over the state-of-the-art (SOTA) baseline, achieving
relative improvements of 8.85% in Precision@5, 12.30% in Recall@5, and 10.86%
in F1@5 compared to the SOTA model on benchmark datasets. This work facilitates
the practical application of TCM in disease treatment and healthcare.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 06:14:26 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Zheng",
"Xinhan",
""
],
[
"Wu",
"Huyu",
""
],
[
"Jin",
"Haopeng",
""
],
[
"Li",
"Ruotai",
""
]
]
| TITLE: FMCHS: Advancing Traditional Chinese Medicine Herb Recommendation with
Fusion of Multiscale Correlations of Herbs and Symptoms
ABSTRACT: Traditional Chinese medicine (TCM) exhibits remarkable therapeutic efficacy
in disease treatment and healthcare through personalized herb prescriptions.
However, current herb recommendation models inadequately capture the multiscale
relations between herbs and clinical symptoms, particularly neglecting latent
correlations at the chemical-molecular scale. To address these limitations, we
propose the Fusion of Multiscale Correlations of Herbs and Symptoms (FMCHS), an
innovative framework that synergistically integrates molecular-scale chemical
characteristics of herbs with clinical symptoms. The framework employs
multi-relational graph transformer layers to generate enriched embeddings that
preserve both structural and semantic features within herbs and symptoms.
Through systematic incorporation of herb chemical profiles into node embeddings
and implementation of attention-based feature fusion, FMCHS effectively
utilizes multiscale correlations. Comprehensive evaluations demonstrate FMCHS's
superior performance over the state-of-the-art (SOTA) baseline, achieving
relative improvements of 8.85% in Precision@5, 12.30% in Recall@5, and 10.86%
in F1@5 compared to the SOTA model on benchmark datasets. This work facilitates
the practical application of TCM in disease treatment and healthcare.
| no_new_dataset | 0.949763 |
2503.05169 | Juniper Tyree | Juniper Tyree, Andreas Rupp, Petri S. Clusius, and Michael H. Boy | phepy: Visual Benchmarks and Improvements for Out-of-Distribution
Detectors | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by-sa/4.0/ | Applying machine learning to increasingly high-dimensional problems with
sparse or biased training data increases the risk that a model is used on
inputs outside its training domain. For such out-of-distribution (OOD) inputs,
the model can no longer make valid predictions, and its error is potentially
unbounded.
Testing OOD detection methods on real-world datasets is complicated by the
ambiguity around which inputs are in-distribution (ID) or OOD. We design a
benchmark for OOD detection, which includes three novel and easily-visualisable
toy examples. These simple examples provide direct and intuitive insight into
whether the detector is able to detect (1) linear and (2) non-linear concepts
and (3) identify thin ID subspaces (needles) within high-dimensional spaces
(haystacks). We use our benchmark to evaluate the performance of various
methods from the literature.
Since tactile examples of OOD inputs may benefit OOD detection, we also
review several simple methods to synthesise OOD inputs for supervised training.
We introduce two improvements, $t$-poking and OOD sample weighting, to make
supervised detectors more precise at the ID-OOD boundary. This is especially
important when conflicts between real ID and synthetic OOD sample blur the
decision boundary.
Finally, we provide recommendations for constructing and applying
out-of-distribution detectors in machine learning.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 06:25:20 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Tyree",
"Juniper",
""
],
[
"Rupp",
"Andreas",
""
],
[
"Clusius",
"Petri S.",
""
],
[
"Boy",
"Michael H.",
""
]
]
| TITLE: phepy: Visual Benchmarks and Improvements for Out-of-Distribution
Detectors
ABSTRACT: Applying machine learning to increasingly high-dimensional problems with
sparse or biased training data increases the risk that a model is used on
inputs outside its training domain. For such out-of-distribution (OOD) inputs,
the model can no longer make valid predictions, and its error is potentially
unbounded.
Testing OOD detection methods on real-world datasets is complicated by the
ambiguity around which inputs are in-distribution (ID) or OOD. We design a
benchmark for OOD detection, which includes three novel and easily-visualisable
toy examples. These simple examples provide direct and intuitive insight into
whether the detector is able to detect (1) linear and (2) non-linear concepts
and (3) identify thin ID subspaces (needles) within high-dimensional spaces
(haystacks). We use our benchmark to evaluate the performance of various
methods from the literature.
Since tactile examples of OOD inputs may benefit OOD detection, we also
review several simple methods to synthesise OOD inputs for supervised training.
We introduce two improvements, $t$-poking and OOD sample weighting, to make
supervised detectors more precise at the ID-OOD boundary. This is especially
important when conflicts between real ID and synthetic OOD sample blur the
decision boundary.
Finally, we provide recommendations for constructing and applying
out-of-distribution detectors in machine learning.
| no_new_dataset | 0.886027 |
2503.05170 | Willmer Quinones | Willmer Rafell Quinones Robles, Sakonporn Noree, Young Sin Ko, Bryan
Wong, Jongwoo Kim, Mun Yong Yi | Spatial Context-Driven Positive Pair Sampling for Enhanced
Histopathology Image Classification | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deep learning has demonstrated great promise in cancer classification from
whole-slide images (WSIs) but remains constrained by the need for extensive
annotations. Annotation-free methods, such as multiple instance learning (MIL)
and self-supervised learning (SSL), have emerged to address this challenge;
however, current SSL techniques often depend on synthetic augmentations or
temporal context, which may not adequately capture the intricate spatial
relationships inherent to histopathology. In this work, we introduce a novel
spatial context-driven positive pair sampling strategy for SSL that leverages
the natural coherence of adjacent patches in WSIs. By constructing biologically
relevant positive pairs from spatially proximate patches, our approach
harnesses inherent spatial coherence to enhance patch-level representations,
ultimately boosting slide-level classification performance. Experiments on
multiple datasets reveal that our strategy improves classification accuracy by
5\% to 10\% over the standard method, paving the way for more clinically
relevant AI models in cancer diagnosis. The code is available at
https://anonymous.4open.science/r/contextual-pairs-E72F/.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 06:31:19 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Robles",
"Willmer Rafell Quinones",
""
],
[
"Noree",
"Sakonporn",
""
],
[
"Ko",
"Young Sin",
""
],
[
"Wong",
"Bryan",
""
],
[
"Kim",
"Jongwoo",
""
],
[
"Yi",
"Mun Yong",
""
]
]
| TITLE: Spatial Context-Driven Positive Pair Sampling for Enhanced
Histopathology Image Classification
ABSTRACT: Deep learning has demonstrated great promise in cancer classification from
whole-slide images (WSIs) but remains constrained by the need for extensive
annotations. Annotation-free methods, such as multiple instance learning (MIL)
and self-supervised learning (SSL), have emerged to address this challenge;
however, current SSL techniques often depend on synthetic augmentations or
temporal context, which may not adequately capture the intricate spatial
relationships inherent to histopathology. In this work, we introduce a novel
spatial context-driven positive pair sampling strategy for SSL that leverages
the natural coherence of adjacent patches in WSIs. By constructing biologically
relevant positive pairs from spatially proximate patches, our approach
harnesses inherent spatial coherence to enhance patch-level representations,
ultimately boosting slide-level classification performance. Experiments on
multiple datasets reveal that our strategy improves classification accuracy by
5\% to 10\% over the standard method, paving the way for more clinically
relevant AI models in cancer diagnosis. The code is available at
https://anonymous.4open.science/r/contextual-pairs-E72F/.
| no_new_dataset | 0.95222 |
2503.05173 | Samson Zhou | Vincent Cohen-Addad, Shaofeng H.-C. Jiang, Qiaoyuan Yang, Yubo Zhang,
Samson Zhou | Fair Clustering in the Sliding Window Model | ICLR 2025 | null | null | null | cs.DS | http://creativecommons.org/licenses/by/4.0/ | We study streaming algorithms for proportionally fair clustering, a notion
originally suggested by Chierichetti et. al. (2017), in the sliding window
model. We show that although there exist efficient streaming algorithms in the
insertion-only model, surprisingly no algorithm can achieve finite
multiplicative ratio without violating the fairness constraint in the sliding
window. Hence, the problem of fair clustering is a rare separation between the
insertion-only streaming model and the sliding window model. On the other hand,
we show that if the fairness constraint is relaxed by a multiplicative
$(1+\varepsilon)$ factor, there exists a $(1 + \varepsilon)$-approximate
sliding window algorithm that uses $\text{poly}(k\varepsilon^{-1}\log n)$
space. This achieves essentially the best parameters (up to degree in the
polynomial) provided the aforementioned lower bound. We also implement a number
of empirical evaluations on real datasets to complement our theoretical
results.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 06:39:51 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Cohen-Addad",
"Vincent",
""
],
[
"Jiang",
"Shaofeng H. -C.",
""
],
[
"Yang",
"Qiaoyuan",
""
],
[
"Zhang",
"Yubo",
""
],
[
"Zhou",
"Samson",
""
]
]
| TITLE: Fair Clustering in the Sliding Window Model
ABSTRACT: We study streaming algorithms for proportionally fair clustering, a notion
originally suggested by Chierichetti et. al. (2017), in the sliding window
model. We show that although there exist efficient streaming algorithms in the
insertion-only model, surprisingly no algorithm can achieve finite
multiplicative ratio without violating the fairness constraint in the sliding
window. Hence, the problem of fair clustering is a rare separation between the
insertion-only streaming model and the sliding window model. On the other hand,
we show that if the fairness constraint is relaxed by a multiplicative
$(1+\varepsilon)$ factor, there exists a $(1 + \varepsilon)$-approximate
sliding window algorithm that uses $\text{poly}(k\varepsilon^{-1}\log n)$
space. This achieves essentially the best parameters (up to degree in the
polynomial) provided the aforementioned lower bound. We also implement a number
of empirical evaluations on real datasets to complement our theoretical
results.
| no_new_dataset | 0.949153 |
2503.05174 | Linqi Yang | Linqi Yang, Xiongwei Zhao, Qihao Sun, Ke Wang, Ao Chen, Peng Kang | SplatPose: Geometry-Aware 6-DoF Pose Estimation from Single RGB Image
via 3D Gaussian Splatting | Submitted to IROS 2025 | null | null | null | cs.CV cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | 6-DoF pose estimation is a fundamental task in computer vision with
wide-ranging applications in augmented reality and robotics. Existing single
RGB-based methods often compromise accuracy due to their reliance on initial
pose estimates and susceptibility to rotational ambiguity, while approaches
requiring depth sensors or multi-view setups incur significant deployment
costs. To address these limitations, we introduce SplatPose, a novel framework
that synergizes 3D Gaussian Splatting (3DGS) with a dual-branch neural
architecture to achieve high-precision pose estimation using only a single RGB
image. Central to our approach is the Dual-Attention Ray Scoring Network
(DARS-Net), which innovatively decouples positional and angular alignment
through geometry-domain attention mechanisms, explicitly modeling directional
dependencies to mitigate rotational ambiguity. Additionally, a coarse-to-fine
optimization pipeline progressively refines pose estimates by aligning dense 2D
features between query images and 3DGS-synthesized views, effectively
correcting feature misalignment and depth errors from sparse ray sampling.
Experiments on three benchmark datasets demonstrate that SplatPose achieves
state-of-the-art 6-DoF pose estimation accuracy in single RGB settings,
rivaling approaches that depend on depth or multi-view images.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 06:40:06 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Yang",
"Linqi",
""
],
[
"Zhao",
"Xiongwei",
""
],
[
"Sun",
"Qihao",
""
],
[
"Wang",
"Ke",
""
],
[
"Chen",
"Ao",
""
],
[
"Kang",
"Peng",
""
]
]
| TITLE: SplatPose: Geometry-Aware 6-DoF Pose Estimation from Single RGB Image
via 3D Gaussian Splatting
ABSTRACT: 6-DoF pose estimation is a fundamental task in computer vision with
wide-ranging applications in augmented reality and robotics. Existing single
RGB-based methods often compromise accuracy due to their reliance on initial
pose estimates and susceptibility to rotational ambiguity, while approaches
requiring depth sensors or multi-view setups incur significant deployment
costs. To address these limitations, we introduce SplatPose, a novel framework
that synergizes 3D Gaussian Splatting (3DGS) with a dual-branch neural
architecture to achieve high-precision pose estimation using only a single RGB
image. Central to our approach is the Dual-Attention Ray Scoring Network
(DARS-Net), which innovatively decouples positional and angular alignment
through geometry-domain attention mechanisms, explicitly modeling directional
dependencies to mitigate rotational ambiguity. Additionally, a coarse-to-fine
optimization pipeline progressively refines pose estimates by aligning dense 2D
features between query images and 3DGS-synthesized views, effectively
correcting feature misalignment and depth errors from sparse ray sampling.
Experiments on three benchmark datasets demonstrate that SplatPose achieves
state-of-the-art 6-DoF pose estimation accuracy in single RGB settings,
rivaling approaches that depend on depth or multi-view images.
| no_new_dataset | 0.943191 |
2503.05179 | Simon Aytes | Simon A. Aytes, Jinheon Baek, Sung Ju Hwang | Sketch-of-Thought: Efficient LLM Reasoning with Adaptive
Cognitive-Inspired Sketching | null | null | null | null | cs.CL cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent advances in large language models have demonstrated remarkable
reasoning capabilities through Chain of Thought (CoT) prompting, but often at
the cost of excessive verbosity in their intermediate outputs, which increases
computational overhead. We introduce Sketch-of-Thought (SoT), a novel prompting
framework that combines cognitive-inspired reasoning paradigms with linguistic
constraints to minimize token usage while preserving reasoning accuracy. SoT is
designed as a flexible framework that can incorporate any custom reasoning
paradigms based on cognitive science, and we instantiate it with three such
paradigms - Conceptual Chaining, Chunked Symbolism, and Expert Lexicons - each
tailored to different reasoning tasks and selected dynamically via a
lightweight routing model. Through comprehensive evaluation across 15 reasoning
datasets with multiple languages and multimodal scenarios, we demonstrate that
SoT achieves token reductions of 76% with negligible accuracy impact. In
certain domains like mathematical and multi-hop reasoning, it even improves
accuracy while using significantly fewer tokens. Our code is publicly
available: https://www.github.com/SimonAytes/SoT.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 06:57:17 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Aytes",
"Simon A.",
""
],
[
"Baek",
"Jinheon",
""
],
[
"Hwang",
"Sung Ju",
""
]
]
| TITLE: Sketch-of-Thought: Efficient LLM Reasoning with Adaptive
Cognitive-Inspired Sketching
ABSTRACT: Recent advances in large language models have demonstrated remarkable
reasoning capabilities through Chain of Thought (CoT) prompting, but often at
the cost of excessive verbosity in their intermediate outputs, which increases
computational overhead. We introduce Sketch-of-Thought (SoT), a novel prompting
framework that combines cognitive-inspired reasoning paradigms with linguistic
constraints to minimize token usage while preserving reasoning accuracy. SoT is
designed as a flexible framework that can incorporate any custom reasoning
paradigms based on cognitive science, and we instantiate it with three such
paradigms - Conceptual Chaining, Chunked Symbolism, and Expert Lexicons - each
tailored to different reasoning tasks and selected dynamically via a
lightweight routing model. Through comprehensive evaluation across 15 reasoning
datasets with multiple languages and multimodal scenarios, we demonstrate that
SoT achieves token reductions of 76% with negligible accuracy impact. In
certain domains like mathematical and multi-hop reasoning, it even improves
accuracy while using significantly fewer tokens. Our code is publicly
available: https://www.github.com/SimonAytes/SoT.
| no_new_dataset | 0.942823 |
2503.05180 | Zherui Huang | Zherui Huang, Xing Gao, Guanjie Zheng, Licheng Wen, Xuemeng Yang, Xiao
Sun | Safety-Critical Traffic Simulation with Adversarial Transfer of Driving
Intentions | Accepted by ICRA 2025 | null | null | null | cs.RO cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Traffic simulation, complementing real-world data with a long-tail
distribution, allows for effective evaluation and enhancement of the ability of
autonomous vehicles to handle accident-prone scenarios. Simulating such
safety-critical scenarios is nontrivial, however, from log data that are
typically regular scenarios, especially in consideration of dynamic adversarial
interactions between the future motions of autonomous vehicles and surrounding
traffic participants. To address it, this paper proposes an innovative and
efficient strategy, termed IntSim, that explicitly decouples the driving
intentions of surrounding actors from their motion planning for realistic and
efficient safety-critical simulation. We formulate the adversarial transfer of
driving intention as an optimization problem, facilitating extensive
exploration of diverse attack behaviors and efficient solution convergence.
Simultaneously, intention-conditioned motion planning benefits from powerful
deep models and large-scale real-world data, permitting the simulation of
realistic motion behaviors for actors. Specially, through adapting driving
intentions based on environments, IntSim facilitates the flexible realization
of dynamic adversarial interactions with autonomous vehicles. Finally,
extensive open-loop and closed-loop experiments on real-world datasets,
including nuScenes and Waymo, demonstrate that the proposed IntSim achieves
state-of-the-art performance in simulating realistic safety-critical scenarios
and further improves planners in handling such scenarios.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 06:59:27 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Huang",
"Zherui",
""
],
[
"Gao",
"Xing",
""
],
[
"Zheng",
"Guanjie",
""
],
[
"Wen",
"Licheng",
""
],
[
"Yang",
"Xuemeng",
""
],
[
"Sun",
"Xiao",
""
]
]
| TITLE: Safety-Critical Traffic Simulation with Adversarial Transfer of Driving
Intentions
ABSTRACT: Traffic simulation, complementing real-world data with a long-tail
distribution, allows for effective evaluation and enhancement of the ability of
autonomous vehicles to handle accident-prone scenarios. Simulating such
safety-critical scenarios is nontrivial, however, from log data that are
typically regular scenarios, especially in consideration of dynamic adversarial
interactions between the future motions of autonomous vehicles and surrounding
traffic participants. To address it, this paper proposes an innovative and
efficient strategy, termed IntSim, that explicitly decouples the driving
intentions of surrounding actors from their motion planning for realistic and
efficient safety-critical simulation. We formulate the adversarial transfer of
driving intention as an optimization problem, facilitating extensive
exploration of diverse attack behaviors and efficient solution convergence.
Simultaneously, intention-conditioned motion planning benefits from powerful
deep models and large-scale real-world data, permitting the simulation of
realistic motion behaviors for actors. Specially, through adapting driving
intentions based on environments, IntSim facilitates the flexible realization
of dynamic adversarial interactions with autonomous vehicles. Finally,
extensive open-loop and closed-loop experiments on real-world datasets,
including nuScenes and Waymo, demonstrate that the proposed IntSim achieves
state-of-the-art performance in simulating realistic safety-critical scenarios
and further improves planners in handling such scenarios.
| no_new_dataset | 0.945701 |
2503.05182 | Qingyuan Zhou | Qingyuan Zhou, Yuehu Gong, Weidong Yang, Jiaze Li, Yeqi Luo, Baixin
Xu, Shuhao Li, Ben Fei, Ying He | MGSR: 2D/3D Mutual-boosted Gaussian Splatting for High-fidelity Surface
Reconstruction under Various Light Conditions | 11 pages, 7 figures | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Novel view synthesis (NVS) and surface reconstruction (SR) are essential
tasks in 3D Gaussian Splatting (3D-GS). Despite recent progress, these tasks
are often addressed independently, with GS-based rendering methods struggling
under diverse light conditions and failing to produce accurate surfaces, while
GS-based reconstruction methods frequently compromise rendering quality. This
raises a central question: must rendering and reconstruction always involve a
trade-off? To address this, we propose MGSR, a 2D/3D Mutual-boosted Gaussian
splatting for Surface Reconstruction that enhances both rendering quality and
3D reconstruction accuracy. MGSR introduces two branches--one based on 2D-GS
and the other on 3D-GS. The 2D-GS branch excels in surface reconstruction,
providing precise geometry information to the 3D-GS branch. Leveraging this
geometry, the 3D-GS branch employs a geometry-guided illumination decomposition
module that captures reflected and transmitted components, enabling realistic
rendering under varied light conditions. Using the transmitted component as
supervision, the 2D-GS branch also achieves high-fidelity surface
reconstruction. Throughout the optimization process, the 2D-GS and 3D-GS
branches undergo alternating optimization, providing mutual supervision. Prior
to this, each branch completes an independent warm-up phase, with an early
stopping strategy implemented to reduce computational costs. We evaluate MGSR
on a diverse set of synthetic and real-world datasets, at both object and scene
levels, demonstrating strong performance in rendering and surface
reconstruction.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 07:06:47 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Zhou",
"Qingyuan",
""
],
[
"Gong",
"Yuehu",
""
],
[
"Yang",
"Weidong",
""
],
[
"Li",
"Jiaze",
""
],
[
"Luo",
"Yeqi",
""
],
[
"Xu",
"Baixin",
""
],
[
"Li",
"Shuhao",
""
],
[
"Fei",
"Ben",
""
],
[
"He",
"Ying",
""
]
]
| TITLE: MGSR: 2D/3D Mutual-boosted Gaussian Splatting for High-fidelity Surface
Reconstruction under Various Light Conditions
ABSTRACT: Novel view synthesis (NVS) and surface reconstruction (SR) are essential
tasks in 3D Gaussian Splatting (3D-GS). Despite recent progress, these tasks
are often addressed independently, with GS-based rendering methods struggling
under diverse light conditions and failing to produce accurate surfaces, while
GS-based reconstruction methods frequently compromise rendering quality. This
raises a central question: must rendering and reconstruction always involve a
trade-off? To address this, we propose MGSR, a 2D/3D Mutual-boosted Gaussian
splatting for Surface Reconstruction that enhances both rendering quality and
3D reconstruction accuracy. MGSR introduces two branches--one based on 2D-GS
and the other on 3D-GS. The 2D-GS branch excels in surface reconstruction,
providing precise geometry information to the 3D-GS branch. Leveraging this
geometry, the 3D-GS branch employs a geometry-guided illumination decomposition
module that captures reflected and transmitted components, enabling realistic
rendering under varied light conditions. Using the transmitted component as
supervision, the 2D-GS branch also achieves high-fidelity surface
reconstruction. Throughout the optimization process, the 2D-GS and 3D-GS
branches undergo alternating optimization, providing mutual supervision. Prior
to this, each branch completes an independent warm-up phase, with an early
stopping strategy implemented to reduce computational costs. We evaluate MGSR
on a diverse set of synthetic and real-world datasets, at both object and scene
levels, demonstrating strong performance in rendering and surface
reconstruction.
| no_new_dataset | 0.949106 |
2503.05183 | Quan Yu | Quan Yu, Yu-Hong Dai, Minru Bai | Spectral-Spatial Extraction through Layered Tensor Decomposition for
Hyperspectral Anomaly Detection | null | null | null | null | cs.CV math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Low rank tensor representation (LRTR) methods are very useful for
hyperspectral anomaly detection (HAD). To overcome the limitations that they
often overlook spectral anomaly and rely on large-scale matrix singular value
decomposition, we first apply non-negative matrix factorization (NMF) to
alleviate spectral dimensionality redundancy and extract spectral anomaly and
then employ LRTR to extract spatial anomaly while mitigating spatial
redundancy, yielding a highly efffcient layered tensor decomposition (LTD)
framework for HAD. An iterative algorithm based on proximal alternating
minimization is developed to solve the proposed LTD model, with convergence
guarantees provided. Moreover, we introduce a rank reduction strategy with
validation mechanism that adaptively reduces data size while preventing
excessive reduction. Theoretically, we rigorously establish the equivalence
between the tensor tubal rank and tensor group sparsity regularization (TGSR)
and, under mild conditions, demonstrate that the relaxed formulation of TGSR
shares the same global minimizers and optimal values as its original
counterpart. Experimental results on the Airport-Beach-Urban and MVTec datasets
demonstrate that our approach outperforms state-of-the-art methods in the HAD
task.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 07:08:14 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Yu",
"Quan",
""
],
[
"Dai",
"Yu-Hong",
""
],
[
"Bai",
"Minru",
""
]
]
| TITLE: Spectral-Spatial Extraction through Layered Tensor Decomposition for
Hyperspectral Anomaly Detection
ABSTRACT: Low rank tensor representation (LRTR) methods are very useful for
hyperspectral anomaly detection (HAD). To overcome the limitations that they
often overlook spectral anomaly and rely on large-scale matrix singular value
decomposition, we first apply non-negative matrix factorization (NMF) to
alleviate spectral dimensionality redundancy and extract spectral anomaly and
then employ LRTR to extract spatial anomaly while mitigating spatial
redundancy, yielding a highly efffcient layered tensor decomposition (LTD)
framework for HAD. An iterative algorithm based on proximal alternating
minimization is developed to solve the proposed LTD model, with convergence
guarantees provided. Moreover, we introduce a rank reduction strategy with
validation mechanism that adaptively reduces data size while preventing
excessive reduction. Theoretically, we rigorously establish the equivalence
between the tensor tubal rank and tensor group sparsity regularization (TGSR)
and, under mild conditions, demonstrate that the relaxed formulation of TGSR
shares the same global minimizers and optimal values as its original
counterpart. Experimental results on the Airport-Beach-Urban and MVTec datasets
demonstrate that our approach outperforms state-of-the-art methods in the HAD
task.
| no_new_dataset | 0.9462 |
2503.05190 | Lei Zhu | Lei Zhu, Yanyu Xu, Huazhu Fu, Xinxing Xu, Rick Siow Mong Goh, and Yong
Liu | Partially Supervised Unpaired Multi-Modal Learning for Label-Efficient
Medical Image Segmentation | Accepted to MLMI 2024 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Unpaired Multi-Modal Learning (UMML) which leverages unpaired multi-modal
data to boost model performance on each individual modality has attracted a lot
of research interests in medical image analysis. However, existing UMML methods
require multi-modal datasets to be fully labeled, which incurs tremendous
annotation cost. In this paper, we investigate the use of partially labeled
data for label-efficient unpaired multi-modal learning, which can reduce the
annotation cost by up to one half. We term the new learning paradigm as
Partially Supervised Unpaired Multi-Modal Learning (PSUMML) and propose a novel
Decomposed partial class adaptation with snapshot Ensembled Self-Training
(DEST) framework for it. Specifically, our framework consists of a compact
segmentation network with modality specific normalization layers for learning
with partially labeled unpaired multi-modal data. The key challenge in PSUMML
lies in the complex partial class distribution discrepancy due to partial class
annotation, which hinders effective knowledge transfer across modalities. We
theoretically analyze this phenomenon with a decomposition theorem and propose
a decomposed partial class adaptation technique to precisely align the
partially labeled classes across modalities to reduce the distribution
discrepancy. We further propose a snapshot ensembled self-training technique to
leverage the valuable snapshot models during training to assign pseudo-labels
to partially labeled pixels for self-training to boost model performance. We
perform extensive experiments under different scenarios of PSUMML for two
medical image segmentation tasks, namely cardiac substructure segmentation and
abdominal multi-organ segmentation. Our framework outperforms existing methods
significantly.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 07:22:42 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Zhu",
"Lei",
""
],
[
"Xu",
"Yanyu",
""
],
[
"Fu",
"Huazhu",
""
],
[
"Xu",
"Xinxing",
""
],
[
"Goh",
"Rick Siow Mong",
""
],
[
"Liu",
"Yong",
""
]
]
| TITLE: Partially Supervised Unpaired Multi-Modal Learning for Label-Efficient
Medical Image Segmentation
ABSTRACT: Unpaired Multi-Modal Learning (UMML) which leverages unpaired multi-modal
data to boost model performance on each individual modality has attracted a lot
of research interests in medical image analysis. However, existing UMML methods
require multi-modal datasets to be fully labeled, which incurs tremendous
annotation cost. In this paper, we investigate the use of partially labeled
data for label-efficient unpaired multi-modal learning, which can reduce the
annotation cost by up to one half. We term the new learning paradigm as
Partially Supervised Unpaired Multi-Modal Learning (PSUMML) and propose a novel
Decomposed partial class adaptation with snapshot Ensembled Self-Training
(DEST) framework for it. Specifically, our framework consists of a compact
segmentation network with modality specific normalization layers for learning
with partially labeled unpaired multi-modal data. The key challenge in PSUMML
lies in the complex partial class distribution discrepancy due to partial class
annotation, which hinders effective knowledge transfer across modalities. We
theoretically analyze this phenomenon with a decomposition theorem and propose
a decomposed partial class adaptation technique to precisely align the
partially labeled classes across modalities to reduce the distribution
discrepancy. We further propose a snapshot ensembled self-training technique to
leverage the valuable snapshot models during training to assign pseudo-labels
to partially labeled pixels for self-training to boost model performance. We
perform extensive experiments under different scenarios of PSUMML for two
medical image segmentation tasks, namely cardiac substructure segmentation and
abdominal multi-organ segmentation. Our framework outperforms existing methods
significantly.
| no_new_dataset | 0.949576 |
2503.05207 | Yunkai Gao | Yunkai Gao, Jiaming Guo, Fan Wu, Rui Zhang | Policy Constraint by Only Support Constraint for Offline Reinforcement
Learning | null | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Offline reinforcement learning (RL) aims to optimize a policy by using
pre-collected datasets, to maximize cumulative rewards. However, offline
reinforcement learning suffers challenges due to the distributional shift
between the learned and behavior policies, leading to errors when computing
Q-values for out-of-distribution (OOD) actions. To mitigate this issue, policy
constraint methods aim to constrain the learned policy's distribution with the
distribution of the behavior policy or confine action selection within the
support of the behavior policy. However, current policy constraint methods tend
to exhibit excessive conservatism, hindering the policy from further surpassing
the behavior policy's performance. In this work, we present Only Support
Constraint (OSC) which is derived from maximizing the total probability of
learned policy in the support of behavior policy, to address the conservatism
of policy constraint. OSC presents a regularization term that only restricts
policies to the support without imposing extra constraints on actions within
the support. Additionally, to fully harness the performance of the new policy
constraints, OSC utilizes a diffusion model to effectively characterize the
support of behavior policies. Experimental evaluations across a variety of
offline RL benchmarks demonstrate that OSC significantly enhances performance,
alleviating the challenges associated with distributional shifts and mitigating
conservatism of policy constraints. Code is available at
https://github.com/MoreanP/OSC.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 07:55:51 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Gao",
"Yunkai",
""
],
[
"Guo",
"Jiaming",
""
],
[
"Wu",
"Fan",
""
],
[
"Zhang",
"Rui",
""
]
]
| TITLE: Policy Constraint by Only Support Constraint for Offline Reinforcement
Learning
ABSTRACT: Offline reinforcement learning (RL) aims to optimize a policy by using
pre-collected datasets, to maximize cumulative rewards. However, offline
reinforcement learning suffers challenges due to the distributional shift
between the learned and behavior policies, leading to errors when computing
Q-values for out-of-distribution (OOD) actions. To mitigate this issue, policy
constraint methods aim to constrain the learned policy's distribution with the
distribution of the behavior policy or confine action selection within the
support of the behavior policy. However, current policy constraint methods tend
to exhibit excessive conservatism, hindering the policy from further surpassing
the behavior policy's performance. In this work, we present Only Support
Constraint (OSC) which is derived from maximizing the total probability of
learned policy in the support of behavior policy, to address the conservatism
of policy constraint. OSC presents a regularization term that only restricts
policies to the support without imposing extra constraints on actions within
the support. Additionally, to fully harness the performance of the new policy
constraints, OSC utilizes a diffusion model to effectively characterize the
support of behavior policies. Experimental evaluations across a variety of
offline RL benchmarks demonstrate that OSC significantly enhances performance,
alleviating the challenges associated with distributional shifts and mitigating
conservatism of policy constraints. Code is available at
https://github.com/MoreanP/OSC.
| no_new_dataset | 0.950732 |
2503.05212 | Aixin Sun | Guoxiu He, Xin Song, Aixin Sun | Knowledge Updating? No More Model Editing! Just Selective Contextual
Reasoning | null | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | As real-world knowledge evolves, the information embedded within large
language models (LLMs) can become outdated, inadequate, or erroneous. Model
editing has emerged as a prominent approach for updating LLMs' knowledge with
minimal computational costs and parameter changes. This approach typically
identifies and adjusts specific model parameters associated with newly acquired
knowledge. However, existing methods often underestimate the adverse effects
that parameter modifications can have on broadly distributed knowledge. More
critically, post-edit LLMs frequently struggle with multi-hop reasoning and
continuous knowledge updates. Although various studies have discussed these
shortcomings, there is a lack of comprehensive evaluation. In this paper, we
provide an evaluation of ten model editing methods along four dimensions:
reliability, generalization, locality, and portability. Results confirm that
all ten popular model editing methods show significant shortcomings across
multiple dimensions, suggesting model editing is less promising. We then
propose a straightforward method called Selective Contextual Reasoning (SCR),
for knowledge updating. SCR does not modify model parameters but harnesses
LLM's inherent contextual reasoning capabilities utilizing the updated
knowledge pieces. Under SCR, an LLM first assesses whether an incoming query
falls within the scope of an external knowledge base. If it does, the relevant
external knowledge texts are contextualized to enhance reasoning; otherwise,
the query is answered directly. We evaluate SCR against the ten model editing
methods on two counterfactual datasets with three backbone LLMs. Empirical
results confirm the effectiveness and efficiency of contextual reasoning for
knowledge updating.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 08:04:25 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"He",
"Guoxiu",
""
],
[
"Song",
"Xin",
""
],
[
"Sun",
"Aixin",
""
]
]
| TITLE: Knowledge Updating? No More Model Editing! Just Selective Contextual
Reasoning
ABSTRACT: As real-world knowledge evolves, the information embedded within large
language models (LLMs) can become outdated, inadequate, or erroneous. Model
editing has emerged as a prominent approach for updating LLMs' knowledge with
minimal computational costs and parameter changes. This approach typically
identifies and adjusts specific model parameters associated with newly acquired
knowledge. However, existing methods often underestimate the adverse effects
that parameter modifications can have on broadly distributed knowledge. More
critically, post-edit LLMs frequently struggle with multi-hop reasoning and
continuous knowledge updates. Although various studies have discussed these
shortcomings, there is a lack of comprehensive evaluation. In this paper, we
provide an evaluation of ten model editing methods along four dimensions:
reliability, generalization, locality, and portability. Results confirm that
all ten popular model editing methods show significant shortcomings across
multiple dimensions, suggesting model editing is less promising. We then
propose a straightforward method called Selective Contextual Reasoning (SCR),
for knowledge updating. SCR does not modify model parameters but harnesses
LLM's inherent contextual reasoning capabilities utilizing the updated
knowledge pieces. Under SCR, an LLM first assesses whether an incoming query
falls within the scope of an external knowledge base. If it does, the relevant
external knowledge texts are contextualized to enhance reasoning; otherwise,
the query is answered directly. We evaluate SCR against the ten model editing
methods on two counterfactual datasets with three backbone LLMs. Empirical
results confirm the effectiveness and efficiency of contextual reasoning for
knowledge updating.
| no_new_dataset | 0.9434 |
2503.05217 | Gulpi Pratamasunu Qorik Oktagalu | Gulpi Qorik Oktagalu Pratamasunu, Guoqing Hao and Kazuhiro Fukui | Separability Membrane: 3D Active Contour for Point Cloud Surface
Reconstruction | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | This paper proposes Separability Membrane, a robust 3D active contour for
extracting a surface from 3D point cloud object. Our approach defines the
surface of a 3D object as the boundary that maximizes the separability of point
features, such as intensity, color, or local density, between its inner and
outer regions based on Fisher's ratio. Separability Membrane identifies the
exact surface of a 3D object by maximizing class separability while controlling
the rigidity of the 3D surface model with an adaptive B-spline surface that
adjusts its properties based on the local and global separability. A key
advantage of our method is its ability to accurately reconstruct surface
boundaries even when they are ambiguous due to noise or outliers, without
requiring any training data or conversion to volumetric representation.
Evaluations on a synthetic 3D point cloud dataset and the 3DNet dataset
demonstrate the membrane's effectiveness and robustness under diverse
conditions.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 08:15:02 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Pratamasunu",
"Gulpi Qorik Oktagalu",
""
],
[
"Hao",
"Guoqing",
""
],
[
"Fukui",
"Kazuhiro",
""
]
]
| TITLE: Separability Membrane: 3D Active Contour for Point Cloud Surface
Reconstruction
ABSTRACT: This paper proposes Separability Membrane, a robust 3D active contour for
extracting a surface from 3D point cloud object. Our approach defines the
surface of a 3D object as the boundary that maximizes the separability of point
features, such as intensity, color, or local density, between its inner and
outer regions based on Fisher's ratio. Separability Membrane identifies the
exact surface of a 3D object by maximizing class separability while controlling
the rigidity of the 3D surface model with an adaptive B-spline surface that
adjusts its properties based on the local and global separability. A key
advantage of our method is its ability to accurately reconstruct surface
boundaries even when they are ambiguous due to noise or outliers, without
requiring any training data or conversion to volumetric representation.
Evaluations on a synthetic 3D point cloud dataset and the 3DNet dataset
demonstrate the membrane's effectiveness and robustness under diverse
conditions.
| no_new_dataset | 0.949201 |
2503.05223 | Yifan Liu | Yifan Liu, Yu Fang, Zhouhan Lin | DiVISe: Direct Visual-Input Speech Synthesis Preserving Speaker
Characteristics And Intelligibility | to be published in NAACL 25 | null | null | null | cs.SD cs.CV cs.LG cs.MM eess.AS | http://creativecommons.org/licenses/by/4.0/ | Video-to-speech (V2S) synthesis, the task of generating speech directly from
silent video input, is inherently more challenging than other speech synthesis
tasks due to the need to accurately reconstruct both speech content and speaker
characteristics from visual cues alone. Recently, audio-visual pre-training has
eliminated the need for additional acoustic hints in V2S, which previous
methods often relied on to ensure training convergence. However, even with
pre-training, existing methods continue to face challenges in achieving a
balance between acoustic intelligibility and the preservation of
speaker-specific characteristics. We analyzed this limitation and were
motivated to introduce DiVISe (Direct Visual-Input Speech Synthesis), an
end-to-end V2S model that predicts Mel-spectrograms directly from video frames
alone. Despite not taking any acoustic hints, DiVISe effectively preserves
speaker characteristics in the generated audio, and achieves superior
performance on both objective and subjective metrics across the LRS2 and LRS3
datasets. Our results demonstrate that DiVISe not only outperforms existing V2S
models in acoustic intelligibility but also scales more effectively with
increased data and model parameters. Code and weights can be found at
https://github.com/PussyCat0700/DiVISe.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 08:21:48 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Liu",
"Yifan",
""
],
[
"Fang",
"Yu",
""
],
[
"Lin",
"Zhouhan",
""
]
]
| TITLE: DiVISe: Direct Visual-Input Speech Synthesis Preserving Speaker
Characteristics And Intelligibility
ABSTRACT: Video-to-speech (V2S) synthesis, the task of generating speech directly from
silent video input, is inherently more challenging than other speech synthesis
tasks due to the need to accurately reconstruct both speech content and speaker
characteristics from visual cues alone. Recently, audio-visual pre-training has
eliminated the need for additional acoustic hints in V2S, which previous
methods often relied on to ensure training convergence. However, even with
pre-training, existing methods continue to face challenges in achieving a
balance between acoustic intelligibility and the preservation of
speaker-specific characteristics. We analyzed this limitation and were
motivated to introduce DiVISe (Direct Visual-Input Speech Synthesis), an
end-to-end V2S model that predicts Mel-spectrograms directly from video frames
alone. Despite not taking any acoustic hints, DiVISe effectively preserves
speaker characteristics in the generated audio, and achieves superior
performance on both objective and subjective metrics across the LRS2 and LRS3
datasets. Our results demonstrate that DiVISe not only outperforms existing V2S
models in acoustic intelligibility but also scales more effectively with
increased data and model parameters. Code and weights can be found at
https://github.com/PussyCat0700/DiVISe.
| no_new_dataset | 0.950686 |
2503.05228 | Ruoxuan Zhang | Ruoxuan Zhang, Hongxia Xie, Yi Yao, Jian-Yu Jiang-Lin, Bin Wen, Ling
Lo, Hong-Han Shuai, Yung-Hui Li, Wen-Huang Cheng | RecipeGen: A Benchmark for Real-World Recipe Image Generation | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recipe image generation is an important challenge in food computing, with
applications from culinary education to interactive recipe platforms. However,
there is currently no real-world dataset that comprehensively connects recipe
goals, sequential steps, and corresponding images. To address this, we
introduce RecipeGen, the first real-world goal-step-image benchmark for recipe
generation, featuring diverse ingredients, varied recipe steps, multiple
cooking styles, and a broad collection of food categories. Data is in
https://github.com/zhangdaxia22/RecipeGen.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 08:25:28 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Zhang",
"Ruoxuan",
""
],
[
"Xie",
"Hongxia",
""
],
[
"Yao",
"Yi",
""
],
[
"Jiang-Lin",
"Jian-Yu",
""
],
[
"Wen",
"Bin",
""
],
[
"Lo",
"Ling",
""
],
[
"Shuai",
"Hong-Han",
""
],
[
"Li",
"Yung-Hui",
""
],
[
"Cheng",
"Wen-Huang",
""
]
]
| TITLE: RecipeGen: A Benchmark for Real-World Recipe Image Generation
ABSTRACT: Recipe image generation is an important challenge in food computing, with
applications from culinary education to interactive recipe platforms. However,
there is currently no real-world dataset that comprehensively connects recipe
goals, sequential steps, and corresponding images. To address this, we
introduce RecipeGen, the first real-world goal-step-image benchmark for recipe
generation, featuring diverse ingredients, varied recipe steps, multiple
cooking styles, and a broad collection of food categories. Data is in
https://github.com/zhangdaxia22/RecipeGen.
| new_dataset | 0.9549 |
2503.05231 | Li Haonan | Shuo Jiang, Haonan Li, Ruochen Ren, Yanmin Zhou, Zhipeng Wang, Bin He | Kaiwu: A Multimodal Manipulation Dataset and Framework for Robot
Learning and Human-Robot Interaction | null | null | null | null | cs.RO cs.AI | http://creativecommons.org/licenses/by/4.0/ | Cutting-edge robot learning techniques including foundation models and
imitation learning from humans all pose huge demands on large-scale and
high-quality datasets which constitute one of the bottleneck in the general
intelligent robot fields. This paper presents the Kaiwu multimodal dataset to
address the missing real-world synchronized multimodal data problems in the
sophisticated assembling scenario,especially with dynamics information and its
fine-grained labelling. The dataset first provides an integration of
human,environment and robot data collection framework with 20 subjects and 30
interaction objects resulting in totally 11,664 instances of integrated
actions. For each of the demonstration,hand motions,operation pressures,sounds
of the assembling process,multi-view videos, high-precision motion capture
information,eye gaze with first-person videos,electromyography signals are all
recorded. Fine-grained multi-level annotation based on absolute timestamp,and
semantic segmentation labelling are performed. Kaiwu dataset aims to facilitate
robot learning,dexterous manipulation,human intention investigation and
human-robot collaboration research.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 08:28:24 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Jiang",
"Shuo",
""
],
[
"Li",
"Haonan",
""
],
[
"Ren",
"Ruochen",
""
],
[
"Zhou",
"Yanmin",
""
],
[
"Wang",
"Zhipeng",
""
],
[
"He",
"Bin",
""
]
]
| TITLE: Kaiwu: A Multimodal Manipulation Dataset and Framework for Robot
Learning and Human-Robot Interaction
ABSTRACT: Cutting-edge robot learning techniques including foundation models and
imitation learning from humans all pose huge demands on large-scale and
high-quality datasets which constitute one of the bottleneck in the general
intelligent robot fields. This paper presents the Kaiwu multimodal dataset to
address the missing real-world synchronized multimodal data problems in the
sophisticated assembling scenario,especially with dynamics information and its
fine-grained labelling. The dataset first provides an integration of
human,environment and robot data collection framework with 20 subjects and 30
interaction objects resulting in totally 11,664 instances of integrated
actions. For each of the demonstration,hand motions,operation pressures,sounds
of the assembling process,multi-view videos, high-precision motion capture
information,eye gaze with first-person videos,electromyography signals are all
recorded. Fine-grained multi-level annotation based on absolute timestamp,and
semantic segmentation labelling are performed. Kaiwu dataset aims to facilitate
robot learning,dexterous manipulation,human intention investigation and
human-robot collaboration research.
| new_dataset | 0.971966 |
2503.05236 | Yibin Wang | Yibin Wang and Yuhang Zang and Hao Li and Cheng Jin and Jiaqi Wang | Unified Reward Model for Multimodal Understanding and Generation | project page: https://codegoat24.github.io/UnifiedReward/ | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent advances in human preference alignment have significantly enhanced
multimodal generation and understanding. A key approach is training reward
models to guide preference optimization. However, existing models are often
task-specific, limiting their adaptability across diverse visual applications.
We also argue that jointly learning to assess multiple tasks may foster a
synergistic effect, where improved image understanding enhances image
generation assessment, and refined image evaluation benefits video assessment
through better frame analysis. To this end, this paper proposes UnifiedReward,
the first unified reward model for multimodal understanding and generation
assessment, enabling both pairwise ranking and pointwise scoring, which can be
employed for vision model preference alignment. Specifically, (1) we first
develop UnifiedReward on our constructed large-scale human preference dataset,
including both image and video generation/understanding tasks. (2) Then, it is
utilized to automatically construct high-quality preference pair data based on
the vision models, fine-gradually filtering their outputs through pair ranking
and point sifting. (3) Finally, these data are used for their preference
alignment through Direct Preference Optimization (DPO). Experimental results
demonstrate that joint learning to assess diverse visual tasks can lead to
substantial mutual benefits and we apply our pipeline to both image and video
understanding/generation tasks, significantly improving the performance in each
domain.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 08:36:05 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Wang",
"Yibin",
""
],
[
"Zang",
"Yuhang",
""
],
[
"Li",
"Hao",
""
],
[
"Jin",
"Cheng",
""
],
[
"Wang",
"Jiaqi",
""
]
]
| TITLE: Unified Reward Model for Multimodal Understanding and Generation
ABSTRACT: Recent advances in human preference alignment have significantly enhanced
multimodal generation and understanding. A key approach is training reward
models to guide preference optimization. However, existing models are often
task-specific, limiting their adaptability across diverse visual applications.
We also argue that jointly learning to assess multiple tasks may foster a
synergistic effect, where improved image understanding enhances image
generation assessment, and refined image evaluation benefits video assessment
through better frame analysis. To this end, this paper proposes UnifiedReward,
the first unified reward model for multimodal understanding and generation
assessment, enabling both pairwise ranking and pointwise scoring, which can be
employed for vision model preference alignment. Specifically, (1) we first
develop UnifiedReward on our constructed large-scale human preference dataset,
including both image and video generation/understanding tasks. (2) Then, it is
utilized to automatically construct high-quality preference pair data based on
the vision models, fine-gradually filtering their outputs through pair ranking
and point sifting. (3) Finally, these data are used for their preference
alignment through Direct Preference Optimization (DPO). Experimental results
demonstrate that joint learning to assess diverse visual tasks can lead to
substantial mutual benefits and we apply our pipeline to both image and video
understanding/generation tasks, significantly improving the performance in each
domain.
| new_dataset | 0.964589 |
2503.05251 | Daniele Palossi | Lorenzo Scarciglia, Antonio Paolillo, Daniele Palossi | A Map-free Deep Learning-based Framework for Gate-to-Gate Monocular
Visual Navigation aboard Miniaturized Aerial Vehicles | \c{opyright}2025 IEEE. Personal use of this material is permitted.
Permission from IEEE must be obtained for all other uses, in any current or
future media, including reprinting/republishing this material for advertising
or promotional purposes, creating new collective works, for resale or
redistribution to servers or lists, or reuse of any copyrighted component of
this work in other works | null | null | null | cs.RO cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Palm-sized autonomous nano-drones, i.e., sub-50g in weight, recently entered
the drone racing scenario, where they are tasked to avoid obstacles and
navigate as fast as possible through gates. However, in contrast with their
bigger counterparts, i.e., kg-scale drones, nano-drones expose three orders of
magnitude less onboard memory and compute power, demanding more efficient and
lightweight vision-based pipelines to win the race. This work presents a
map-free vision-based (using only a monocular camera) autonomous nano-drone
that combines a real-time deep learning gate detection front-end with a classic
yet elegant and effective visual servoing control back-end, only relying on
onboard resources. Starting from two state-of-the-art tiny deep learning
models, we adapt them for our specific task, and after a mixed
simulator-real-world training, we integrate and deploy them aboard our
nano-drone. Our best-performing pipeline costs of only 24M multiply-accumulate
operations per frame, resulting in a closed-loop control performance of 30 Hz,
while achieving a gate detection root mean square error of 1.4 pixels, on our
~20k real-world image dataset. In-field experiments highlight the capability of
our nano-drone to successfully navigate through 15 gates in 4 min, never
crashing and covering a total travel distance of ~100m, with a peak flight
speed of 1.9 m/s. Finally, to stress the generalization capability of our
system, we also test it in a never-seen-before environment, where it navigates
through gates for more than 4 min.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 09:07:07 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Scarciglia",
"Lorenzo",
""
],
[
"Paolillo",
"Antonio",
""
],
[
"Palossi",
"Daniele",
""
]
]
| TITLE: A Map-free Deep Learning-based Framework for Gate-to-Gate Monocular
Visual Navigation aboard Miniaturized Aerial Vehicles
ABSTRACT: Palm-sized autonomous nano-drones, i.e., sub-50g in weight, recently entered
the drone racing scenario, where they are tasked to avoid obstacles and
navigate as fast as possible through gates. However, in contrast with their
bigger counterparts, i.e., kg-scale drones, nano-drones expose three orders of
magnitude less onboard memory and compute power, demanding more efficient and
lightweight vision-based pipelines to win the race. This work presents a
map-free vision-based (using only a monocular camera) autonomous nano-drone
that combines a real-time deep learning gate detection front-end with a classic
yet elegant and effective visual servoing control back-end, only relying on
onboard resources. Starting from two state-of-the-art tiny deep learning
models, we adapt them for our specific task, and after a mixed
simulator-real-world training, we integrate and deploy them aboard our
nano-drone. Our best-performing pipeline costs of only 24M multiply-accumulate
operations per frame, resulting in a closed-loop control performance of 30 Hz,
while achieving a gate detection root mean square error of 1.4 pixels, on our
~20k real-world image dataset. In-field experiments highlight the capability of
our nano-drone to successfully navigate through 15 gates in 4 min, never
crashing and covering a total travel distance of ~100m, with a peak flight
speed of 1.9 m/s. Finally, to stress the generalization capability of our
system, we also test it in a never-seen-before environment, where it navigates
through gates for more than 4 min.
| no_new_dataset | 0.834811 |
2503.05255 | Yan Xia | Guanghao Zhang, Tao Zhong, Yan Xia, Zhelun Yu, Haoyuan Li, Wanggui He,
Fangxun Shu, Mushui Liu, Dong She, Yi Wang, Hao Jiang | CMMCoT: Enhancing Complex Multi-Image Comprehension via Multi-Modal
Chain-of-Thought and Memory Augmentation | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While previous multimodal slow-thinking methods have demonstrated remarkable
success in single-image understanding scenarios, their effectiveness becomes
fundamentally constrained when extended to more complex multi-image
comprehension tasks. This limitation stems from their predominant reliance on
text-based intermediate reasoning processes. While for human, when engaging in
sophisticated multi-image analysis, they typically perform two complementary
cognitive operations: (1) continuous cross-image visual comparison through
region-of-interest matching, and (2) dynamic memorization of critical visual
concepts throughout the reasoning chain. Motivated by these observations, we
propose the Complex Multi-Modal Chain-of-Thought (CMMCoT) framework, a
multi-step reasoning framework that mimics human-like "slow thinking" for
multi-image understanding. Our approach incorporates two key innovations: 1.
The construction of interleaved multimodal multi-step reasoning chains, which
utilize critical visual region tokens, extracted from intermediate reasoning
steps, as supervisory signals. This mechanism not only facilitates
comprehensive cross-modal understanding but also enhances model
interpretability. 2. The introduction of a test-time memory augmentation module
that expands the model reasoning capacity during inference while preserving
parameter efficiency. Furthermore, to facilitate research in this direction, we
have curated a novel multi-image slow-thinking dataset. Extensive experiments
demonstrate the effectiveness of our model.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 09:13:17 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Zhang",
"Guanghao",
""
],
[
"Zhong",
"Tao",
""
],
[
"Xia",
"Yan",
""
],
[
"Yu",
"Zhelun",
""
],
[
"Li",
"Haoyuan",
""
],
[
"He",
"Wanggui",
""
],
[
"Shu",
"Fangxun",
""
],
[
"Liu",
"Mushui",
""
],
[
"She",
"Dong",
""
],
[
"Wang",
"Yi",
""
],
[
"Jiang",
"Hao",
""
]
]
| TITLE: CMMCoT: Enhancing Complex Multi-Image Comprehension via Multi-Modal
Chain-of-Thought and Memory Augmentation
ABSTRACT: While previous multimodal slow-thinking methods have demonstrated remarkable
success in single-image understanding scenarios, their effectiveness becomes
fundamentally constrained when extended to more complex multi-image
comprehension tasks. This limitation stems from their predominant reliance on
text-based intermediate reasoning processes. While for human, when engaging in
sophisticated multi-image analysis, they typically perform two complementary
cognitive operations: (1) continuous cross-image visual comparison through
region-of-interest matching, and (2) dynamic memorization of critical visual
concepts throughout the reasoning chain. Motivated by these observations, we
propose the Complex Multi-Modal Chain-of-Thought (CMMCoT) framework, a
multi-step reasoning framework that mimics human-like "slow thinking" for
multi-image understanding. Our approach incorporates two key innovations: 1.
The construction of interleaved multimodal multi-step reasoning chains, which
utilize critical visual region tokens, extracted from intermediate reasoning
steps, as supervisory signals. This mechanism not only facilitates
comprehensive cross-modal understanding but also enhances model
interpretability. 2. The introduction of a test-time memory augmentation module
that expands the model reasoning capacity during inference while preserving
parameter efficiency. Furthermore, to facilitate research in this direction, we
have curated a novel multi-image slow-thinking dataset. Extensive experiments
demonstrate the effectiveness of our model.
| new_dataset | 0.864996 |
2503.05260 | Matteo Ceccarello | Matteo Ceccarello, Andrea Pietracaprina, Geppino Pucci, Francesco
Vison\`a | Fair Center Clustering in Sliding Windows | null | null | null | null | cs.DS | http://creativecommons.org/licenses/by/4.0/ | The $k$-center problem requires the selection of $k$ points (centers) from a
given metric pointset $W$ so to minimize the maximum distance of any point of
$W$ from the closest center. This paper focuses on a fair variant of the
problem, known as \emph {fair center}, where each input point belongs to some
category and each category may contribute a limited number of points to the
center set. We present the first space-efficient streaming algorithm for fair
center in general metrics, under the sliding window model. At any time $t$, the
algorithm is able to provide a solution for the current window whose quality is
almost as good as the one guaranteed by the best, polynomial-time sequential
algorithms run on the entire window, and exhibits space and time requirements
independent of the window size. Our theoretical results are backed by an
extensive set of experiments on both real-world and synthetic datasets, which
provide evidence of the practical viability of the algorithm.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 09:19:56 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Ceccarello",
"Matteo",
""
],
[
"Pietracaprina",
"Andrea",
""
],
[
"Pucci",
"Geppino",
""
],
[
"Visonà",
"Francesco",
""
]
]
| TITLE: Fair Center Clustering in Sliding Windows
ABSTRACT: The $k$-center problem requires the selection of $k$ points (centers) from a
given metric pointset $W$ so to minimize the maximum distance of any point of
$W$ from the closest center. This paper focuses on a fair variant of the
problem, known as \emph {fair center}, where each input point belongs to some
category and each category may contribute a limited number of points to the
center set. We present the first space-efficient streaming algorithm for fair
center in general metrics, under the sliding window model. At any time $t$, the
algorithm is able to provide a solution for the current window whose quality is
almost as good as the one guaranteed by the best, polynomial-time sequential
algorithms run on the entire window, and exhibits space and time requirements
independent of the window size. Our theoretical results are backed by an
extensive set of experiments on both real-world and synthetic datasets, which
provide evidence of the practical viability of the algorithm.
| no_new_dataset | 0.944893 |
2503.05268 | Francesco Cazzaro | Francesco Cazzaro, Justin Kleindienst, Sofia Marquez, Ariadna Quattoni | ZOGRASCOPE: A New Benchmark for Property Graphs | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | Natural language interfaces to knowledge graphs have become increasingly
important in recent years, enabling easy and efficient access to structured
data. In particular property graphs have seen growing adoption. However, these
kind of graphs remain relatively underrepresented in research, which has
focused in large part on RDF-style graphs. As a matter of fact there is a lack
of resources for evaluating systems on property graphs, with many existing
datasets featuring relatively simple queries. To address this gap, we introduce
ZOGRASCOPE, a benchmark designed specifically for the cypher query language.
The benchmark includes a diverse set of manually annotated queries of varying
complexity. We complement this paper with a set of experiments that test the
performance of out-of-the-box LLMs of different sizes. Our experiments show
that semantic parsing over graphs is still a challenging open problem that can
not be solved by prompting LLMs alone.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 09:33:30 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Cazzaro",
"Francesco",
""
],
[
"Kleindienst",
"Justin",
""
],
[
"Marquez",
"Sofia",
""
],
[
"Quattoni",
"Ariadna",
""
]
]
| TITLE: ZOGRASCOPE: A New Benchmark for Property Graphs
ABSTRACT: Natural language interfaces to knowledge graphs have become increasingly
important in recent years, enabling easy and efficient access to structured
data. In particular property graphs have seen growing adoption. However, these
kind of graphs remain relatively underrepresented in research, which has
focused in large part on RDF-style graphs. As a matter of fact there is a lack
of resources for evaluating systems on property graphs, with many existing
datasets featuring relatively simple queries. To address this gap, we introduce
ZOGRASCOPE, a benchmark designed specifically for the cypher query language.
The benchmark includes a diverse set of manually annotated queries of varying
complexity. We complement this paper with a set of experiments that test the
performance of out-of-the-box LLMs of different sizes. Our experiments show
that semantic parsing over graphs is still a challenging open problem that can
not be solved by prompting LLMs alone.
| new_dataset | 0.961498 |
2503.05274 | Mohammad Sajad Marvi | Sajad Marvi and Christoph Rist and Julian Schmidt and Julian Jordan
and Abhinav Valada | Evidential Uncertainty Estimation for Multi-Modal Trajectory Prediction | null | null | null | null | cs.RO cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Accurate trajectory prediction is crucial for autonomous driving, yet
uncertainty in agent behavior and perception noise makes it inherently
challenging. While multi-modal trajectory prediction models generate multiple
plausible future paths with associated probabilities, effectively quantifying
uncertainty remains an open problem. In this work, we propose a novel
multi-modal trajectory prediction approach based on evidential deep learning
that estimates both positional and mode probability uncertainty in real time.
Our approach leverages a Normal Inverse Gamma distribution for positional
uncertainty and a Dirichlet distribution for mode uncertainty. Unlike
sampling-based methods, it infers both types of uncertainty in a single forward
pass, significantly improving efficiency. Additionally, we experimented with
uncertainty-driven importance sampling to improve training efficiency by
prioritizing underrepresented high-uncertainty samples over redundant ones. We
perform extensive evaluations of our method on the Argoverse 1 and Argoverse 2
datasets, demonstrating that it provides reliable uncertainty estimates while
maintaining high trajectory prediction accuracy.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 09:46:21 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Marvi",
"Sajad",
""
],
[
"Rist",
"Christoph",
""
],
[
"Schmidt",
"Julian",
""
],
[
"Jordan",
"Julian",
""
],
[
"Valada",
"Abhinav",
""
]
]
| TITLE: Evidential Uncertainty Estimation for Multi-Modal Trajectory Prediction
ABSTRACT: Accurate trajectory prediction is crucial for autonomous driving, yet
uncertainty in agent behavior and perception noise makes it inherently
challenging. While multi-modal trajectory prediction models generate multiple
plausible future paths with associated probabilities, effectively quantifying
uncertainty remains an open problem. In this work, we propose a novel
multi-modal trajectory prediction approach based on evidential deep learning
that estimates both positional and mode probability uncertainty in real time.
Our approach leverages a Normal Inverse Gamma distribution for positional
uncertainty and a Dirichlet distribution for mode uncertainty. Unlike
sampling-based methods, it infers both types of uncertainty in a single forward
pass, significantly improving efficiency. Additionally, we experimented with
uncertainty-driven importance sampling to improve training efficiency by
prioritizing underrepresented high-uncertainty samples over redundant ones. We
perform extensive evaluations of our method on the Argoverse 1 and Argoverse 2
datasets, demonstrating that it provides reliable uncertainty estimates while
maintaining high trajectory prediction accuracy.
| no_new_dataset | 0.944944 |
2503.05283 | Souhail Hadgi | Souhail Hadgi, Luca Moschella, Andrea Santilli, Diego Gomez, Qixing
Huang, Emanuele Rodol\`a, Simone Melzi, Maks Ovsjanikov | Escaping Plato's Cave: Towards the Alignment of 3D and Text Latent
Spaces | Accepted at CVPR 2025 | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Recent works have shown that, when trained at scale, uni-modal 2D vision and
text encoders converge to learned features that share remarkable structural
properties, despite arising from different representations. However, the role
of 3D encoders with respect to other modalities remains unexplored.
Furthermore, existing 3D foundation models that leverage large datasets are
typically trained with explicit alignment objectives with respect to frozen
encoders from other representations. In this work, we investigate the
possibility of a posteriori alignment of representations obtained from
uni-modal 3D encoders compared to text-based feature spaces. We show that naive
post-training feature alignment of uni-modal text and 3D encoders results in
limited performance. We then focus on extracting subspaces of the corresponding
feature spaces and discover that by projecting learned representations onto
well-chosen lower-dimensional subspaces the quality of alignment becomes
significantly higher, leading to improved accuracy on matching and retrieval
tasks. Our analysis further sheds light on the nature of these shared
subspaces, which roughly separate between semantic and geometric data
representations. Overall, ours is the first work that helps to establish a
baseline for post-training alignment of 3D uni-modal and text feature spaces,
and helps to highlight both the shared and unique properties of 3D data
compared to other representations.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 09:51:56 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Hadgi",
"Souhail",
""
],
[
"Moschella",
"Luca",
""
],
[
"Santilli",
"Andrea",
""
],
[
"Gomez",
"Diego",
""
],
[
"Huang",
"Qixing",
""
],
[
"Rodolà",
"Emanuele",
""
],
[
"Melzi",
"Simone",
""
],
[
"Ovsjanikov",
"Maks",
""
]
]
| TITLE: Escaping Plato's Cave: Towards the Alignment of 3D and Text Latent
Spaces
ABSTRACT: Recent works have shown that, when trained at scale, uni-modal 2D vision and
text encoders converge to learned features that share remarkable structural
properties, despite arising from different representations. However, the role
of 3D encoders with respect to other modalities remains unexplored.
Furthermore, existing 3D foundation models that leverage large datasets are
typically trained with explicit alignment objectives with respect to frozen
encoders from other representations. In this work, we investigate the
possibility of a posteriori alignment of representations obtained from
uni-modal 3D encoders compared to text-based feature spaces. We show that naive
post-training feature alignment of uni-modal text and 3D encoders results in
limited performance. We then focus on extracting subspaces of the corresponding
feature spaces and discover that by projecting learned representations onto
well-chosen lower-dimensional subspaces the quality of alignment becomes
significantly higher, leading to improved accuracy on matching and retrieval
tasks. Our analysis further sheds light on the nature of these shared
subspaces, which roughly separate between semantic and geometric data
representations. Overall, ours is the first work that helps to establish a
baseline for post-training alignment of 3D uni-modal and text feature spaces,
and helps to highlight both the shared and unique properties of 3D data
compared to other representations.
| no_new_dataset | 0.94428 |
2503.05289 | Eliav Mor | Eliav Mor and Yair Carmon | An Analytical Model for Overparameterized Learning Under Class Imbalance | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | We study class-imbalanced linear classification in a high-dimensional
Gaussian mixture model. We develop a tight, closed form approximation for the
test error of several practical learning methods, including logit adjustment
and class dependent temperature. Our approximation allows us to analytically
tune and compare these methods, highlighting how and when they overcome the
pitfalls of standard cross-entropy minimization. We test our theoretical
findings on simulated data and imbalanced CIFAR10, MNIST and FashionMNIST
datasets.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 10:09:16 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Mor",
"Eliav",
""
],
[
"Carmon",
"Yair",
""
]
]
| TITLE: An Analytical Model for Overparameterized Learning Under Class Imbalance
ABSTRACT: We study class-imbalanced linear classification in a high-dimensional
Gaussian mixture model. We develop a tight, closed form approximation for the
test error of several practical learning methods, including logit adjustment
and class dependent temperature. Our approximation allows us to analytically
tune and compare these methods, highlighting how and when they overcome the
pitfalls of standard cross-entropy minimization. We test our theoretical
findings on simulated data and imbalanced CIFAR10, MNIST and FashionMNIST
datasets.
| no_new_dataset | 0.948058 |
2503.05301 | Adrian Pfisterer | Adrian Pfisterer, Xing Li, Vito Mengers, Oliver Brock | A Helping (Human) Hand in Kinematic Structure Estimation | Accepted at ICRA25; 8 pages + 7 figures; For supplementary material,
see https://www.tu.berlin/robotics/papers/helpinghands | null | null | null | cs.RO | http://creativecommons.org/licenses/by/4.0/ | Visual uncertainties such as occlusions, lack of texture, and noise present
significant challenges in obtaining accurate kinematic models for safe robotic
manipulation. We introduce a probabilistic real-time approach that leverages
the human hand as a prior to mitigate these uncertainties. By tracking the
constrained motion of the human hand during manipulation and explicitly
modeling uncertainties in visual observations, our method reliably estimates an
object's kinematic model online. We validate our approach on a novel dataset
featuring challenging objects that are occluded during manipulation and offer
limited articulations for perception. The results demonstrate that by
incorporating an appropriate prior and explicitly accounting for uncertainties,
our method produces accurate estimates, outperforming two recent baselines by
195% and 140%, respectively. Furthermore, we demonstrate that our approach's
estimates are precise enough to allow a robot to manipulate even small objects
safely.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 10:29:25 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Pfisterer",
"Adrian",
""
],
[
"Li",
"Xing",
""
],
[
"Mengers",
"Vito",
""
],
[
"Brock",
"Oliver",
""
]
]
| TITLE: A Helping (Human) Hand in Kinematic Structure Estimation
ABSTRACT: Visual uncertainties such as occlusions, lack of texture, and noise present
significant challenges in obtaining accurate kinematic models for safe robotic
manipulation. We introduce a probabilistic real-time approach that leverages
the human hand as a prior to mitigate these uncertainties. By tracking the
constrained motion of the human hand during manipulation and explicitly
modeling uncertainties in visual observations, our method reliably estimates an
object's kinematic model online. We validate our approach on a novel dataset
featuring challenging objects that are occluded during manipulation and offer
limited articulations for perception. The results demonstrate that by
incorporating an appropriate prior and explicitly accounting for uncertainties,
our method produces accurate estimates, outperforming two recent baselines by
195% and 140%, respectively. Furthermore, we demonstrate that our approach's
estimates are precise enough to allow a robot to manipulate even small objects
safely.
| new_dataset | 0.961025 |
2503.05305 | Hu Yu | Hu Yu, Hao Luo, Hangjie Yuan, Yu Rong, Feng Zhao | Frequency Autoregressive Image Generation with Continuous Tokens | null | null | null | null | cs.CV cs.AI | http://creativecommons.org/publicdomain/zero/1.0/ | Autoregressive (AR) models for image generation typically adopt a two-stage
paradigm of vector quantization and raster-scan ``next-token prediction",
inspired by its great success in language modeling. However, due to the huge
modality gap, image autoregressive models may require a systematic reevaluation
from two perspectives: tokenizer format and regression direction. In this
paper, we introduce the frequency progressive autoregressive (\textbf{FAR})
paradigm and instantiate FAR with the continuous tokenizer. Specifically, we
identify spectral dependency as the desirable regression direction for FAR,
wherein higher-frequency components build upon the lower one to progressively
construct a complete image. This design seamlessly fits the causality
requirement for autoregressive models and preserves the unique spatial locality
of image data. Besides, we delve into the integration of FAR and the continuous
tokenizer, introducing a series of techniques to address optimization
challenges and improve the efficiency of training and inference processes. We
demonstrate the efficacy of FAR through comprehensive experiments on the
ImageNet dataset and verify its potential on text-to-image generation.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 10:34:04 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Yu",
"Hu",
""
],
[
"Luo",
"Hao",
""
],
[
"Yuan",
"Hangjie",
""
],
[
"Rong",
"Yu",
""
],
[
"Zhao",
"Feng",
""
]
]
| TITLE: Frequency Autoregressive Image Generation with Continuous Tokens
ABSTRACT: Autoregressive (AR) models for image generation typically adopt a two-stage
paradigm of vector quantization and raster-scan ``next-token prediction",
inspired by its great success in language modeling. However, due to the huge
modality gap, image autoregressive models may require a systematic reevaluation
from two perspectives: tokenizer format and regression direction. In this
paper, we introduce the frequency progressive autoregressive (\textbf{FAR})
paradigm and instantiate FAR with the continuous tokenizer. Specifically, we
identify spectral dependency as the desirable regression direction for FAR,
wherein higher-frequency components build upon the lower one to progressively
construct a complete image. This design seamlessly fits the causality
requirement for autoregressive models and preserves the unique spatial locality
of image data. Besides, we delve into the integration of FAR and the continuous
tokenizer, introducing a series of techniques to address optimization
challenges and improve the efficiency of training and inference processes. We
demonstrate the efficacy of FAR through comprehensive experiments on the
ImageNet dataset and verify its potential on text-to-image generation.
| no_new_dataset | 0.949295 |
2503.05306 | Hyungkyu Kang | Hyungkyu Kang and Min-hwan Oh | Adversarial Policy Optimization for Offline Preference-based
Reinforcement Learning | null | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we study offline preference-based reinforcement learning
(PbRL), where learning is based on pre-collected preference feedback over pairs
of trajectories. While offline PbRL has demonstrated remarkable empirical
success, existing theoretical approaches face challenges in ensuring
conservatism under uncertainty, requiring computationally intractable
confidence set constructions. We address this limitation by proposing
Adversarial Preference-based Policy Optimization (APPO), a computationally
efficient algorithm for offline PbRL that guarantees sample complexity bounds
without relying on explicit confidence sets. By framing PbRL as a two-player
game between a policy and a model, our approach enforces conservatism in a
tractable manner. Using standard assumptions on function approximation and
bounded trajectory concentrability, we derive a sample complexity bound. To our
knowledge, APPO is the first offline PbRL algorithm to offer both statistical
efficiency and practical applicability. Experimental results on continuous
control tasks demonstrate that APPO effectively learns from complex datasets,
showing comparable performance with existing state-of-the-art methods.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 10:35:01 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Kang",
"Hyungkyu",
""
],
[
"Oh",
"Min-hwan",
""
]
]
| TITLE: Adversarial Policy Optimization for Offline Preference-based
Reinforcement Learning
ABSTRACT: In this paper, we study offline preference-based reinforcement learning
(PbRL), where learning is based on pre-collected preference feedback over pairs
of trajectories. While offline PbRL has demonstrated remarkable empirical
success, existing theoretical approaches face challenges in ensuring
conservatism under uncertainty, requiring computationally intractable
confidence set constructions. We address this limitation by proposing
Adversarial Preference-based Policy Optimization (APPO), a computationally
efficient algorithm for offline PbRL that guarantees sample complexity bounds
without relying on explicit confidence sets. By framing PbRL as a two-player
game between a policy and a model, our approach enforces conservatism in a
tractable manner. Using standard assumptions on function approximation and
bounded trajectory concentrability, we derive a sample complexity bound. To our
knowledge, APPO is the first offline PbRL algorithm to offer both statistical
efficiency and practical applicability. Experimental results on continuous
control tasks demonstrate that APPO effectively learns from complex datasets,
showing comparable performance with existing state-of-the-art methods.
| no_new_dataset | 0.946349 |
2503.05319 | Xinkun Wang | Xinkun Wang, Yifang Wang, Senwei Liang, Feilong Tang, Chengzhi Liu,
Ming Hu, Chao Hu, Junjun He, Zongyuan Ge, and Imran Razzak | Robust Multimodal Learning for Ophthalmic Disease Grading via
Disentangled Representation | 10pages | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | This paper discusses how ophthalmologists often rely on multimodal data to
improve diagnostic accuracy. However, complete multimodal data is rare in
real-world applications due to a lack of medical equipment and concerns about
data privacy. Traditional deep learning methods typically address these issues
by learning representations in latent space. However, the paper highlights two
key limitations of these approaches: (i) Task-irrelevant redundant information
(e.g., numerous slices) in complex modalities leads to significant redundancy
in latent space representations. (ii) Overlapping multimodal representations
make it difficult to extract unique features for each modality. To overcome
these challenges, the authors propose the Essence-Point and Disentangle
Representation Learning (EDRL) strategy, which integrates a self-distillation
mechanism into an end-to-end framework to enhance feature selection and
disentanglement for more robust multimodal learning. Specifically, the
Essence-Point Representation Learning module selects discriminative features
that improve disease grading performance. The Disentangled Representation
Learning module separates multimodal data into modality-common and
modality-unique representations, reducing feature entanglement and enhancing
both robustness and interpretability in ophthalmic disease diagnosis.
Experiments on multimodal ophthalmology datasets show that the proposed EDRL
strategy significantly outperforms current state-of-the-art methods.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 10:58:38 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Wang",
"Xinkun",
""
],
[
"Wang",
"Yifang",
""
],
[
"Liang",
"Senwei",
""
],
[
"Tang",
"Feilong",
""
],
[
"Liu",
"Chengzhi",
""
],
[
"Hu",
"Ming",
""
],
[
"Hu",
"Chao",
""
],
[
"He",
"Junjun",
""
],
[
"Ge",
"Zongyuan",
""
],
[
"Razzak",
"Imran",
""
]
]
| TITLE: Robust Multimodal Learning for Ophthalmic Disease Grading via
Disentangled Representation
ABSTRACT: This paper discusses how ophthalmologists often rely on multimodal data to
improve diagnostic accuracy. However, complete multimodal data is rare in
real-world applications due to a lack of medical equipment and concerns about
data privacy. Traditional deep learning methods typically address these issues
by learning representations in latent space. However, the paper highlights two
key limitations of these approaches: (i) Task-irrelevant redundant information
(e.g., numerous slices) in complex modalities leads to significant redundancy
in latent space representations. (ii) Overlapping multimodal representations
make it difficult to extract unique features for each modality. To overcome
these challenges, the authors propose the Essence-Point and Disentangle
Representation Learning (EDRL) strategy, which integrates a self-distillation
mechanism into an end-to-end framework to enhance feature selection and
disentanglement for more robust multimodal learning. Specifically, the
Essence-Point Representation Learning module selects discriminative features
that improve disease grading performance. The Disentangled Representation
Learning module separates multimodal data into modality-common and
modality-unique representations, reducing feature entanglement and enhancing
both robustness and interpretability in ophthalmic disease diagnosis.
Experiments on multimodal ophthalmology datasets show that the proposed EDRL
strategy significantly outperforms current state-of-the-art methods.
| no_new_dataset | 0.95253 |
2503.05320 | Zitao Fang | Zitao Fang, Guodong DU, Shuyang Yu, Yifei Guo, Yiwei Zhang, Jing Li,
Ho-Kin Tang, Sim Kuan Goh | Disentangling Task Interference within Neurons: Model Merging in
Alignment with Neuronal Mechanisms | null | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Fine-tuning pre-trained models on targeted datasets enhances task-specific
performance but often comes at the expense of generalization. Model merging
techniques, which integrate multiple fine-tuned models into a single multi-task
model through task arithmetic at various levels: model, layer, or parameter,
offer a promising solution. However, task interference remains a fundamental
challenge, leading to performance degradation and suboptimal merged models.
Existing approaches largely overlook the fundamental role of individual neurons
and their connectivity, resulting in a lack of interpretability in both the
merging process and the merged models. In this work, we present the first study
on the impact of neuronal alignment in model merging. We decompose
task-specific representations into two complementary neuronal subspaces that
regulate neuron sensitivity and input adaptability. Leveraging this
decomposition, we introduce NeuroMerging, a novel merging framework developed
to mitigate task interference within neuronal subspaces, enabling training-free
model fusion across diverse tasks. Through extensive experiments, we
demonstrate that NeuroMerging achieves superior performance compared to
existing methods on multi-task benchmarks across both vision and natural
language domains. Our findings highlight the importance of aligning neuronal
mechanisms in model merging, offering new insights into mitigating task
interference and improving knowledge fusion.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 11:00:24 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Fang",
"Zitao",
""
],
[
"DU",
"Guodong",
""
],
[
"Yu",
"Shuyang",
""
],
[
"Guo",
"Yifei",
""
],
[
"Zhang",
"Yiwei",
""
],
[
"Li",
"Jing",
""
],
[
"Tang",
"Ho-Kin",
""
],
[
"Goh",
"Sim Kuan",
""
]
]
| TITLE: Disentangling Task Interference within Neurons: Model Merging in
Alignment with Neuronal Mechanisms
ABSTRACT: Fine-tuning pre-trained models on targeted datasets enhances task-specific
performance but often comes at the expense of generalization. Model merging
techniques, which integrate multiple fine-tuned models into a single multi-task
model through task arithmetic at various levels: model, layer, or parameter,
offer a promising solution. However, task interference remains a fundamental
challenge, leading to performance degradation and suboptimal merged models.
Existing approaches largely overlook the fundamental role of individual neurons
and their connectivity, resulting in a lack of interpretability in both the
merging process and the merged models. In this work, we present the first study
on the impact of neuronal alignment in model merging. We decompose
task-specific representations into two complementary neuronal subspaces that
regulate neuron sensitivity and input adaptability. Leveraging this
decomposition, we introduce NeuroMerging, a novel merging framework developed
to mitigate task interference within neuronal subspaces, enabling training-free
model fusion across diverse tasks. Through extensive experiments, we
demonstrate that NeuroMerging achieves superior performance compared to
existing methods on multi-task benchmarks across both vision and natural
language domains. Our findings highlight the importance of aligning neuronal
mechanisms in model merging, offering new insights into mitigating task
interference and improving knowledge fusion.
| no_new_dataset | 0.942348 |
2503.05328 | Anar Yeginbergen | Anar Yeginbergen and Maite Oronoz and Rodrigo Agerri | Dynamic Knowledge Integration for Evidence-Driven Counter-Argument
Generation with Large Language Models | null | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | This paper investigates the role of dynamic external knowledge integration in
improving counter-argument generation using Large Language Models (LLMs). While
LLMs have shown promise in argumentative tasks, their tendency to generate
lengthy, potentially unfactual responses highlights the need for more
controlled and evidence-based approaches. We introduce a new manually curated
dataset of argument and counter-argument pairs specifically designed to balance
argumentative complexity with evaluative feasibility. We also propose a new
LLM-as-a-Judge evaluation methodology that shows a stronger correlation with
human judgments compared to traditional reference-based metrics. Our
experimental results demonstrate that integrating dynamic external knowledge
from the web significantly improves the quality of generated counter-arguments,
particularly in terms of relatedness, persuasiveness, and factuality. The
findings suggest that combining LLMs with real-time external knowledge
retrieval offers a promising direction for developing more effective and
reliable counter-argumentation systems.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 11:13:33 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Yeginbergen",
"Anar",
""
],
[
"Oronoz",
"Maite",
""
],
[
"Agerri",
"Rodrigo",
""
]
]
| TITLE: Dynamic Knowledge Integration for Evidence-Driven Counter-Argument
Generation with Large Language Models
ABSTRACT: This paper investigates the role of dynamic external knowledge integration in
improving counter-argument generation using Large Language Models (LLMs). While
LLMs have shown promise in argumentative tasks, their tendency to generate
lengthy, potentially unfactual responses highlights the need for more
controlled and evidence-based approaches. We introduce a new manually curated
dataset of argument and counter-argument pairs specifically designed to balance
argumentative complexity with evaluative feasibility. We also propose a new
LLM-as-a-Judge evaluation methodology that shows a stronger correlation with
human judgments compared to traditional reference-based metrics. Our
experimental results demonstrate that integrating dynamic external knowledge
from the web significantly improves the quality of generated counter-arguments,
particularly in terms of relatedness, persuasiveness, and factuality. The
findings suggest that combining LLMs with real-time external knowledge
retrieval offers a promising direction for developing more effective and
reliable counter-argumentation systems.
| new_dataset | 0.952662 |
2503.05331 | Matic Pikovnik | Matic Pikovnik (1) and \v{Z}iga Zaplotnik (2 and 1) ((1) University of
Ljubljana, Faculty of Mathematics and Physics, Jadranska 19, 1000 Ljubljana,
Slovenia, (2) European Centre for Medium-range Weather Forecasts,
Robert-Schuman-Platz 3, 53175 Bonn, Germany) | The Changes of the Northern Hadley Cell Strength in Reanalyses and
Radiosonde Observations | 17 pages (12 main, 5 supplementary), 11 figures (4 main, 7
supplementary) | null | null | null | physics.ao-ph | http://creativecommons.org/licenses/by/4.0/ | This study examines mean meridional winds and their trends in the Northern
Hadley Cell (NHC) from 1980 to 2022 using reanalysis datasets and radiosonde
observations. Compared to radiosonde data, reanalyses underestimate the mean
upper-tropospheric poleward flow of the NHC but accurately capture the mean
equatorward flow in the lower troposphere. While climate models generally
project a weakening of the NHC, our study finds no significant trend in
radiosonde observations, adding to the uncertainty in future climate
projections. In contrast, reanalyses indicate a strengthening, primarily due to
an intensification of the upper-tropospheric poleward flow. Our examination of
ERA5 analysis increments confirms that the NHC strengthening trend in ERA5 is
not an artefact of data assimilation. Instead, the increments correct the first
guess, which underestimates the strength of the NHC, nudging it toward a
stronger circulation.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 11:15:53 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Pikovnik",
"Matic",
"",
"2 and 1"
],
[
"Zaplotnik",
"Žiga",
"",
"2 and 1"
]
]
| TITLE: The Changes of the Northern Hadley Cell Strength in Reanalyses and
Radiosonde Observations
ABSTRACT: This study examines mean meridional winds and their trends in the Northern
Hadley Cell (NHC) from 1980 to 2022 using reanalysis datasets and radiosonde
observations. Compared to radiosonde data, reanalyses underestimate the mean
upper-tropospheric poleward flow of the NHC but accurately capture the mean
equatorward flow in the lower troposphere. While climate models generally
project a weakening of the NHC, our study finds no significant trend in
radiosonde observations, adding to the uncertainty in future climate
projections. In contrast, reanalyses indicate a strengthening, primarily due to
an intensification of the upper-tropospheric poleward flow. Our examination of
ERA5 analysis increments confirms that the NHC strengthening trend in ERA5 is
not an artefact of data assimilation. Instead, the increments correct the first
guess, which underestimates the strength of the NHC, nudging it toward a
stronger circulation.
| no_new_dataset | 0.948394 |
2503.05332 | Jungho Lee | Jungho Lee, Donghyeong Kim, Dogyoon Lee, Suhwan Cho, Minhyeok Lee,
Wonjoon Lee, Taeoh Kim, Dongyoon Wee, Sangyoun Lee | CoMoGaussian: Continuous Motion-Aware Gaussian Splatting from
Motion-Blurred Images | Revised Version of CRiM-GS, Github:
https://github.com/Jho-Yonsei/CoMoGaussian | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | 3D Gaussian Splatting (3DGS) has gained significant attention for their
high-quality novel view rendering, motivating research to address real-world
challenges. A critical issue is the camera motion blur caused by movement
during exposure, which hinders accurate 3D scene reconstruction. In this study,
we propose CoMoGaussian, a Continuous Motion-Aware Gaussian Splatting that
reconstructs precise 3D scenes from motion-blurred images while maintaining
real-time rendering speed. Considering the complex motion patterns inherent in
real-world camera movements, we predict continuous camera trajectories using
neural ordinary differential equations (ODEs). To ensure accurate modeling, we
employ rigid body transformations, preserving the shape and size of the object
but rely on the discrete integration of sampled frames. To better approximate
the continuous nature of motion blur, we introduce a continuous motion
refinement (CMR) transformation that refines rigid transformations by
incorporating additional learnable parameters. By revisiting fundamental camera
theory and leveraging advanced neural ODE techniques, we achieve precise
modeling of continuous camera trajectories, leading to improved reconstruction
accuracy. Extensive experiments demonstrate state-of-the-art performance both
quantitatively and qualitatively on benchmark datasets, which include a wide
range of motion blur scenarios, from moderate to extreme blur.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 11:18:43 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Lee",
"Jungho",
""
],
[
"Kim",
"Donghyeong",
""
],
[
"Lee",
"Dogyoon",
""
],
[
"Cho",
"Suhwan",
""
],
[
"Lee",
"Minhyeok",
""
],
[
"Lee",
"Wonjoon",
""
],
[
"Kim",
"Taeoh",
""
],
[
"Wee",
"Dongyoon",
""
],
[
"Lee",
"Sangyoun",
""
]
]
| TITLE: CoMoGaussian: Continuous Motion-Aware Gaussian Splatting from
Motion-Blurred Images
ABSTRACT: 3D Gaussian Splatting (3DGS) has gained significant attention for their
high-quality novel view rendering, motivating research to address real-world
challenges. A critical issue is the camera motion blur caused by movement
during exposure, which hinders accurate 3D scene reconstruction. In this study,
we propose CoMoGaussian, a Continuous Motion-Aware Gaussian Splatting that
reconstructs precise 3D scenes from motion-blurred images while maintaining
real-time rendering speed. Considering the complex motion patterns inherent in
real-world camera movements, we predict continuous camera trajectories using
neural ordinary differential equations (ODEs). To ensure accurate modeling, we
employ rigid body transformations, preserving the shape and size of the object
but rely on the discrete integration of sampled frames. To better approximate
the continuous nature of motion blur, we introduce a continuous motion
refinement (CMR) transformation that refines rigid transformations by
incorporating additional learnable parameters. By revisiting fundamental camera
theory and leveraging advanced neural ODE techniques, we achieve precise
modeling of continuous camera trajectories, leading to improved reconstruction
accuracy. Extensive experiments demonstrate state-of-the-art performance both
quantitatively and qualitatively on benchmark datasets, which include a wide
range of motion blur scenarios, from moderate to extreme blur.
| no_new_dataset | 0.953622 |
2503.05333 | Martin Spitznagel | Martin Spitznagel, Jan Vaillant, Janis Keuper | PhysicsGen: Can Generative Models Learn from Images to Predict Complex
Physical Relations? | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The image-to-image translation abilities of generative learning models have
recently made significant progress in the estimation of complex (steered)
mappings between image distributions. While appearance based tasks like image
in-painting or style transfer have been studied at length, we propose to
investigate the potential of generative models in the context of physical
simulations. Providing a dataset of 300k image-pairs and baseline evaluations
for three different physical simulation tasks, we propose a benchmark to
investigate the following research questions: i) are generative models able to
learn complex physical relations from input-output image pairs? ii) what
speedups can be achieved by replacing differential equation based simulations?
While baseline evaluations of different current models show the potential for
high speedups (ii), these results also show strong limitations toward the
physical correctness (i). This underlines the need for new methods to enforce
physical correctness. Data, baseline models and evaluation code
http://www.physics-gen.org.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 11:19:13 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Spitznagel",
"Martin",
""
],
[
"Vaillant",
"Jan",
""
],
[
"Keuper",
"Janis",
""
]
]
| TITLE: PhysicsGen: Can Generative Models Learn from Images to Predict Complex
Physical Relations?
ABSTRACT: The image-to-image translation abilities of generative learning models have
recently made significant progress in the estimation of complex (steered)
mappings between image distributions. While appearance based tasks like image
in-painting or style transfer have been studied at length, we propose to
investigate the potential of generative models in the context of physical
simulations. Providing a dataset of 300k image-pairs and baseline evaluations
for three different physical simulation tasks, we propose a benchmark to
investigate the following research questions: i) are generative models able to
learn complex physical relations from input-output image pairs? ii) what
speedups can be achieved by replacing differential equation based simulations?
While baseline evaluations of different current models show the potential for
high speedups (ii), these results also show strong limitations toward the
physical correctness (i). This underlines the need for new methods to enforce
physical correctness. Data, baseline models and evaluation code
http://www.physics-gen.org.
| new_dataset | 0.957715 |
2503.05335 | Joel Honkamaa | Joel Honkamaa and Pekka Marttinen | New multimodal similarity measure for image registration via modeling
local functional dependence with linear combination of learned basis
functions | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | The deformable registration of images of different modalities, essential in
many medical imaging applications, remains challenging. The main challenge is
developing a robust measure for image overlap despite the compared images
capturing different aspects of the underlying tissue. Here, we explore
similarity metrics based on functional dependence between intensity values of
registered images. Although functional dependence is too restrictive on the
global scale, earlier work has shown competitive performance in deformable
registration when such measures are applied over small enough contexts. We
confirm this finding and further develop the idea by modeling local functional
dependence via the linear basis function model with the basis functions learned
jointly with the deformation. The measure can be implemented via convolutions,
making it efficient to compute on GPUs. We release the method as an easy-to-use
tool and show good performance on three datasets compared to well-established
baseline and earlier functional dependence-based methods.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 11:22:33 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Honkamaa",
"Joel",
""
],
[
"Marttinen",
"Pekka",
""
]
]
| TITLE: New multimodal similarity measure for image registration via modeling
local functional dependence with linear combination of learned basis
functions
ABSTRACT: The deformable registration of images of different modalities, essential in
many medical imaging applications, remains challenging. The main challenge is
developing a robust measure for image overlap despite the compared images
capturing different aspects of the underlying tissue. Here, we explore
similarity metrics based on functional dependence between intensity values of
registered images. Although functional dependence is too restrictive on the
global scale, earlier work has shown competitive performance in deformable
registration when such measures are applied over small enough contexts. We
confirm this finding and further develop the idea by modeling local functional
dependence via the linear basis function model with the basis functions learned
jointly with the deformation. The measure can be implemented via convolutions,
making it efficient to compute on GPUs. We release the method as an easy-to-use
tool and show good performance on three datasets compared to well-established
baseline and earlier functional dependence-based methods.
| no_new_dataset | 0.946349 |
2503.05339 | Zhenxuan Zhang | Zhenxuan Zhang, Peiyuan Jing, Coraline Beitone, Jiahao Huang, Zhifan
Gao, Guang Yang, and Pete Lally | Pretext Task Adversarial Learning for Unpaired Low-field to Ultra
High-field MRI Synthesis | null | null | null | null | eess.IV cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Given the scarcity and cost of high-field MRI, the synthesis of high-field
MRI from low-field MRI holds significant potential when there is limited data
for training downstream tasks (e.g. segmentation). Low-field MRI often suffers
from a reduced signal-to-noise ratio (SNR) and spatial resolution compared to
high-field MRI. However, synthesizing high-field MRI data presents challenges.
These involve aligning image features across domains while preserving
anatomical accuracy and enhancing fine details. To address these challenges, we
propose a Pretext Task Adversarial (PTA) learning framework for high-field MRI
synthesis from low-field MRI data. The framework comprises three processes: (1)
The slice-wise gap perception (SGP) network aligns the slice inconsistencies of
low-field and high-field datasets based on contrastive learning. (2) The local
structure correction (LSC) network extracts local structures by restoring the
locally rotated and masked images. (3) The pretext task-guided adversarial
training process introduces additional supervision and incorporates a
discriminator to improve image realism. Extensive experiments on low-field to
ultra high-field task demonstrate the effectiveness of our method, achieving
state-of-the-art performance (16.892 in FID, 1.933 in IS, and 0.324 in
MS-SSIM). This enables the generation of high-quality high-field-like MRI data
from low-field MRI data to augment training datasets for downstream tasks. The
code is available at:
https://github.com/Zhenxuan-Zhang/PTA4Unpaired_HF_MRI_SYN.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 11:28:55 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Zhang",
"Zhenxuan",
""
],
[
"Jing",
"Peiyuan",
""
],
[
"Beitone",
"Coraline",
""
],
[
"Huang",
"Jiahao",
""
],
[
"Gao",
"Zhifan",
""
],
[
"Yang",
"Guang",
""
],
[
"Lally",
"Pete",
""
]
]
| TITLE: Pretext Task Adversarial Learning for Unpaired Low-field to Ultra
High-field MRI Synthesis
ABSTRACT: Given the scarcity and cost of high-field MRI, the synthesis of high-field
MRI from low-field MRI holds significant potential when there is limited data
for training downstream tasks (e.g. segmentation). Low-field MRI often suffers
from a reduced signal-to-noise ratio (SNR) and spatial resolution compared to
high-field MRI. However, synthesizing high-field MRI data presents challenges.
These involve aligning image features across domains while preserving
anatomical accuracy and enhancing fine details. To address these challenges, we
propose a Pretext Task Adversarial (PTA) learning framework for high-field MRI
synthesis from low-field MRI data. The framework comprises three processes: (1)
The slice-wise gap perception (SGP) network aligns the slice inconsistencies of
low-field and high-field datasets based on contrastive learning. (2) The local
structure correction (LSC) network extracts local structures by restoring the
locally rotated and masked images. (3) The pretext task-guided adversarial
training process introduces additional supervision and incorporates a
discriminator to improve image realism. Extensive experiments on low-field to
ultra high-field task demonstrate the effectiveness of our method, achieving
state-of-the-art performance (16.892 in FID, 1.933 in IS, and 0.324 in
MS-SSIM). This enables the generation of high-quality high-field-like MRI data
from low-field MRI data to augment training datasets for downstream tasks. The
code is available at:
https://github.com/Zhenxuan-Zhang/PTA4Unpaired_HF_MRI_SYN.
| no_new_dataset | 0.952882 |
2503.05347 | Zhenxuan Zhang | Zhenxuan Zhang, Kinhei Lee, Weihang Deng, Huichi Zhou, Zihao Jin,
Jiahao Huang, Zhifan Gao, Dominic C Marshall, Yingying Fang, Guang Yang | GEMA-Score: Granular Explainable Multi-Agent Score for Radiology Report
Evaluation | null | null | null | null | cs.CL cs.MA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Automatic medical report generation supports clinical diagnosis, reduces the
workload of radiologists, and holds the promise of improving diagnosis
consistency. However, existing evaluation metrics primarily assess the accuracy
of key medical information coverage in generated reports compared to
human-written reports, while overlooking crucial details such as the location
and certainty of reported abnormalities. These limitations hinder the
comprehensive assessment of the reliability of generated reports and pose risks
in their selection for clinical use. Therefore, we propose a Granular
Explainable Multi-Agent Score (GEMA-Score) in this paper, which conducts both
objective quantification and subjective evaluation through a large language
model-based multi-agent workflow. Our GEMA-Score parses structured reports and
employs NER-F1 calculations through interactive exchanges of information among
agents to assess disease diagnosis, location, severity, and uncertainty.
Additionally, an LLM-based scoring agent evaluates completeness, readability,
and clinical terminology while providing explanatory feedback. Extensive
experiments validate that GEMA-Score achieves the highest correlation with
human expert evaluations on a public dataset, demonstrating its effectiveness
in clinical scoring (Kendall coefficient = 0.70 for Rexval dataset and Kendall
coefficient = 0.54 for RadEvalX dataset). The anonymous project demo is
available at: https://github.com/Zhenxuan-Zhang/GEMA_score.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 11:42:22 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Zhang",
"Zhenxuan",
""
],
[
"Lee",
"Kinhei",
""
],
[
"Deng",
"Weihang",
""
],
[
"Zhou",
"Huichi",
""
],
[
"Jin",
"Zihao",
""
],
[
"Huang",
"Jiahao",
""
],
[
"Gao",
"Zhifan",
""
],
[
"Marshall",
"Dominic C",
""
],
[
"Fang",
"Yingying",
""
],
[
"Yang",
"Guang",
""
]
]
| TITLE: GEMA-Score: Granular Explainable Multi-Agent Score for Radiology Report
Evaluation
ABSTRACT: Automatic medical report generation supports clinical diagnosis, reduces the
workload of radiologists, and holds the promise of improving diagnosis
consistency. However, existing evaluation metrics primarily assess the accuracy
of key medical information coverage in generated reports compared to
human-written reports, while overlooking crucial details such as the location
and certainty of reported abnormalities. These limitations hinder the
comprehensive assessment of the reliability of generated reports and pose risks
in their selection for clinical use. Therefore, we propose a Granular
Explainable Multi-Agent Score (GEMA-Score) in this paper, which conducts both
objective quantification and subjective evaluation through a large language
model-based multi-agent workflow. Our GEMA-Score parses structured reports and
employs NER-F1 calculations through interactive exchanges of information among
agents to assess disease diagnosis, location, severity, and uncertainty.
Additionally, an LLM-based scoring agent evaluates completeness, readability,
and clinical terminology while providing explanatory feedback. Extensive
experiments validate that GEMA-Score achieves the highest correlation with
human expert evaluations on a public dataset, demonstrating its effectiveness
in clinical scoring (Kendall coefficient = 0.70 for Rexval dataset and Kendall
coefficient = 0.54 for RadEvalX dataset). The anonymous project demo is
available at: https://github.com/Zhenxuan-Zhang/GEMA_score.
| no_new_dataset | 0.95594 |
2503.05349 | Dingkun Liu | Dingkun Liu, Siyang Li, Ziwei Wang, Wei Li and Dongrui Wu | Spatial Distillation based Distribution Alignment (SDDA) for
Cross-Headset EEG Classification | 10 pages, 5 figures | null | null | null | cs.LG cs.AI cs.HC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A non-invasive brain-computer interface (BCI) enables direct interaction
between the user and external devices, typically via electroencephalogram (EEG)
signals. However, decoding EEG signals across different headsets remains a
significant challenge due to differences in the number and locations of the
electrodes. To address this challenge, we propose a spatial distillation based
distribution alignment (SDDA) approach for heterogeneous cross-headset transfer
in non-invasive BCIs. SDDA uses first spatial distillation to make use of the
full set of electrodes, and then input/feature/output space distribution
alignments to cope with the significant differences between the source and
target domains. To our knowledge, this is the first work to use knowledge
distillation in cross-headset transfers. Extensive experiments on six EEG
datasets from two BCI paradigms demonstrated that SDDA achieved superior
performance in both offline unsupervised domain adaptation and online
supervised domain adaptation scenarios, consistently outperforming 10 classical
and state-of-the-art transfer learning algorithms.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 11:44:49 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Liu",
"Dingkun",
""
],
[
"Li",
"Siyang",
""
],
[
"Wang",
"Ziwei",
""
],
[
"Li",
"Wei",
""
],
[
"Wu",
"Dongrui",
""
]
]
| TITLE: Spatial Distillation based Distribution Alignment (SDDA) for
Cross-Headset EEG Classification
ABSTRACT: A non-invasive brain-computer interface (BCI) enables direct interaction
between the user and external devices, typically via electroencephalogram (EEG)
signals. However, decoding EEG signals across different headsets remains a
significant challenge due to differences in the number and locations of the
electrodes. To address this challenge, we propose a spatial distillation based
distribution alignment (SDDA) approach for heterogeneous cross-headset transfer
in non-invasive BCIs. SDDA uses first spatial distillation to make use of the
full set of electrodes, and then input/feature/output space distribution
alignments to cope with the significant differences between the source and
target domains. To our knowledge, this is the first work to use knowledge
distillation in cross-headset transfers. Extensive experiments on six EEG
datasets from two BCI paradigms demonstrated that SDDA achieved superior
performance in both offline unsupervised domain adaptation and online
supervised domain adaptation scenarios, consistently outperforming 10 classical
and state-of-the-art transfer learning algorithms.
| no_new_dataset | 0.947186 |
2503.05357 | Jan Fillies | Jan Fillies and Adrian Paschke | Improving Hate Speech Classification with Cross-Taxonomy Dataset
Integration | Accepted for publication at LaTeCH-CLfL 2025. The 9th Joint ACL
Special Interest Group on Language Technologies for the Socio-Economic
Sciences and Humanities (SIGHUM) Workshop on Computational Linguistics for
Cultural Heritage, Social Sciences, Humanities and Literature | null | null | null | cs.CL cs.AI cs.LG cs.SI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Algorithmic hate speech detection faces significant challenges due to the
diverse definitions and datasets used in research and practice. Social media
platforms, legal frameworks, and institutions each apply distinct yet
overlapping definitions, complicating classification efforts. This study
addresses these challenges by demonstrating that existing datasets and
taxonomies can be integrated into a unified model, enhancing prediction
performance and reducing reliance on multiple specialized classifiers. The work
introduces a universal taxonomy and a hate speech classifier capable of
detecting a wide range of definitions within a single framework. Our approach
is validated by combining two widely used but differently annotated datasets,
showing improved classification performance on an independent test set. This
work highlights the potential of dataset and taxonomy integration in advancing
hate speech detection, increasing efficiency, and ensuring broader
applicability across contexts.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 12:01:02 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Fillies",
"Jan",
""
],
[
"Paschke",
"Adrian",
""
]
]
| TITLE: Improving Hate Speech Classification with Cross-Taxonomy Dataset
Integration
ABSTRACT: Algorithmic hate speech detection faces significant challenges due to the
diverse definitions and datasets used in research and practice. Social media
platforms, legal frameworks, and institutions each apply distinct yet
overlapping definitions, complicating classification efforts. This study
addresses these challenges by demonstrating that existing datasets and
taxonomies can be integrated into a unified model, enhancing prediction
performance and reducing reliance on multiple specialized classifiers. The work
introduces a universal taxonomy and a hate speech classifier capable of
detecting a wide range of definitions within a single framework. Our approach
is validated by combining two widely used but differently annotated datasets,
showing improved classification performance on an independent test set. This
work highlights the potential of dataset and taxonomy integration in advancing
hate speech detection, increasing efficiency, and ensuring broader
applicability across contexts.
| no_new_dataset | 0.953708 |
2503.05362 | Weixiang Zhao | Weixiang Zhao, Xingyu Sui, Xinyang Han, Yang Deng, Yulin Hu, Jiahe
Guo, Libo Qin, Qianyun Du, Shijin Wang, Yanyan Zhao, Bing Qin, Ting Liu | Chain of Strategy Optimization Makes Large Language Models Better
Emotional Supporter | 19 pages, 9 figures, 15 tables | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The growing emotional stress in modern society has increased the demand for
Emotional Support Conversations (ESC). While Large Language Models (LLMs) show
promise for ESC, they face two key challenges: (1) low strategy selection
accuracy, and (2) preference bias, limiting their adaptability to emotional
needs of users. Existing supervised fine-tuning (SFT) struggles to address
these issues, as it rigidly trains models on single gold-standard responses
without modeling nuanced strategy trade-offs. To overcome these limitations, we
propose Chain-of-Strategy Optimization (CSO), a novel approach that optimizes
strategy selection preferences at each dialogue turn. We first leverage Monte
Carlo Tree Search to construct ESC-Pro, a high-quality preference dataset with
turn-level strategy-response pairs. Training on ESC-Pro with CSO improves both
strategy accuracy and bias mitigation, enabling LLMs to generate more
empathetic and contextually appropriate responses. Experiments on LLaMA-3.1-8B,
Gemma-2-9B, and Qwen2.5-7B demonstrate that CSO outperforms standard SFT,
highlighting the efficacy of fine-grained, turn-level preference modeling in
ESC.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 12:07:59 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Zhao",
"Weixiang",
""
],
[
"Sui",
"Xingyu",
""
],
[
"Han",
"Xinyang",
""
],
[
"Deng",
"Yang",
""
],
[
"Hu",
"Yulin",
""
],
[
"Guo",
"Jiahe",
""
],
[
"Qin",
"Libo",
""
],
[
"Du",
"Qianyun",
""
],
[
"Wang",
"Shijin",
""
],
[
"Zhao",
"Yanyan",
""
],
[
"Qin",
"Bing",
""
],
[
"Liu",
"Ting",
""
]
]
| TITLE: Chain of Strategy Optimization Makes Large Language Models Better
Emotional Supporter
ABSTRACT: The growing emotional stress in modern society has increased the demand for
Emotional Support Conversations (ESC). While Large Language Models (LLMs) show
promise for ESC, they face two key challenges: (1) low strategy selection
accuracy, and (2) preference bias, limiting their adaptability to emotional
needs of users. Existing supervised fine-tuning (SFT) struggles to address
these issues, as it rigidly trains models on single gold-standard responses
without modeling nuanced strategy trade-offs. To overcome these limitations, we
propose Chain-of-Strategy Optimization (CSO), a novel approach that optimizes
strategy selection preferences at each dialogue turn. We first leverage Monte
Carlo Tree Search to construct ESC-Pro, a high-quality preference dataset with
turn-level strategy-response pairs. Training on ESC-Pro with CSO improves both
strategy accuracy and bias mitigation, enabling LLMs to generate more
empathetic and contextually appropriate responses. Experiments on LLaMA-3.1-8B,
Gemma-2-9B, and Qwen2.5-7B demonstrate that CSO outperforms standard SFT,
highlighting the efficacy of fine-grained, turn-level preference modeling in
ESC.
| new_dataset | 0.937612 |
2503.05365 | Zhigang Wang | Zhigang Wang, Shaojing Fan, Zhenguang Liu, Zheqi Wu, Sifan Wu,
Yingying Jiao | Multi-Grained Feature Pruning for Video-Based Human Pose Estimation | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Human pose estimation, with its broad applications in action recognition and
motion capture, has experienced significant advancements. However, current
Transformer-based methods for video pose estimation often face challenges in
managing redundant temporal information and achieving fine-grained perception
because they only focus on processing low-resolution features. To address these
challenges, we propose a novel multi-scale resolution framework that encodes
spatio-temporal representations at varying granularities and executes
fine-grained perception compensation. Furthermore, we employ a density peaks
clustering method to dynamically identify and prioritize tokens that offer
important semantic information. This strategy effectively prunes redundant
feature tokens, especially those arising from multi-frame features, thereby
optimizing computational efficiency without sacrificing semantic richness.
Empirically, it sets new benchmarks for both performance and efficiency on
three large-scale datasets. Our method achieves a 93.8% improvement in
inference speed compared to the baseline, while also enhancing pose estimation
accuracy, reaching 87.4 mAP on the PoseTrack2017 dataset.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 12:14:51 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Wang",
"Zhigang",
""
],
[
"Fan",
"Shaojing",
""
],
[
"Liu",
"Zhenguang",
""
],
[
"Wu",
"Zheqi",
""
],
[
"Wu",
"Sifan",
""
],
[
"Jiao",
"Yingying",
""
]
]
| TITLE: Multi-Grained Feature Pruning for Video-Based Human Pose Estimation
ABSTRACT: Human pose estimation, with its broad applications in action recognition and
motion capture, has experienced significant advancements. However, current
Transformer-based methods for video pose estimation often face challenges in
managing redundant temporal information and achieving fine-grained perception
because they only focus on processing low-resolution features. To address these
challenges, we propose a novel multi-scale resolution framework that encodes
spatio-temporal representations at varying granularities and executes
fine-grained perception compensation. Furthermore, we employ a density peaks
clustering method to dynamically identify and prioritize tokens that offer
important semantic information. This strategy effectively prunes redundant
feature tokens, especially those arising from multi-frame features, thereby
optimizing computational efficiency without sacrificing semantic richness.
Empirically, it sets new benchmarks for both performance and efficiency on
three large-scale datasets. Our method achieves a 93.8% improvement in
inference speed compared to the baseline, while also enhancing pose estimation
accuracy, reaching 87.4 mAP on the PoseTrack2017 dataset.
| no_new_dataset | 0.951414 |
2503.05370 | Alea Miako Tokita | Alea Miako Tokita, Timoth\'ee Devergne, A. Marco Saitta, J\"org Behler | Free energy profiles for chemical reactions in solution from
high-dimensional neural network potentials: The case of the Strecker
synthesis | null | null | null | null | physics.chem-ph | http://creativecommons.org/licenses/by/4.0/ | Machine learning potentials (MLPs) have become a popular tool in chemistry
and materials science as they combine the accuracy of electronic structure
calculations with the high computational efficiency of analytic potentials.
MLPs are particularly useful for computationally demanding simulations such as
the determination of free energy profiles governing chemical reactions in
solution, but to date such applications are still rare. In this work we show
how umbrella sampling simulations can be combined with active learning of
high-dimensional neural network potentials (HDNNPs) to construct free energy
profiles in a systematic way. For the example of the first step of Strecker
synthesis of glycine in aqueous solution we provide a detailed analysis of the
improving quality of HDNNPs for datasets of increasing size. We find that next
to the typical quantification of energy and force errors with respect to the
underlying density functional theory data also the long-term stability of the
simulations and the convergence of physical properties should be rigorously
monitored to obtain reliable and converged free energy profiles of chemical
reactions in solution.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 12:22:32 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Tokita",
"Alea Miako",
""
],
[
"Devergne",
"Timothée",
""
],
[
"Saitta",
"A. Marco",
""
],
[
"Behler",
"Jörg",
""
]
]
| TITLE: Free energy profiles for chemical reactions in solution from
high-dimensional neural network potentials: The case of the Strecker
synthesis
ABSTRACT: Machine learning potentials (MLPs) have become a popular tool in chemistry
and materials science as they combine the accuracy of electronic structure
calculations with the high computational efficiency of analytic potentials.
MLPs are particularly useful for computationally demanding simulations such as
the determination of free energy profiles governing chemical reactions in
solution, but to date such applications are still rare. In this work we show
how umbrella sampling simulations can be combined with active learning of
high-dimensional neural network potentials (HDNNPs) to construct free energy
profiles in a systematic way. For the example of the first step of Strecker
synthesis of glycine in aqueous solution we provide a detailed analysis of the
improving quality of HDNNPs for datasets of increasing size. We find that next
to the typical quantification of energy and force errors with respect to the
underlying density functional theory data also the long-term stability of the
simulations and the convergence of physical properties should be rigorously
monitored to obtain reliable and converged free energy profiles of chemical
reactions in solution.
| no_new_dataset | 0.947332 |
2503.05371 | Zara Siddique | Zara Siddique, Irtaza Khalid, Liam D. Turner, Luis Espinosa-Anke | Shifting Perspectives: Steering Vector Ensembles for Robust Bias
Mitigation in LLMs | Submitted to ACL 2025 | null | null | null | cs.LG cs.AI cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | We present a novel approach to bias mitigation in large language models
(LLMs) by applying steering vectors to modify model activations in forward
passes. We employ Bayesian optimization to systematically identify effective
contrastive pair datasets across nine bias axes. When optimized on the BBQ
dataset, our individually tuned steering vectors achieve average improvements
of 12.2%, 4.7%, and 3.2% over the baseline for Mistral, Llama, and Qwen,
respectively. Building on these promising results, we introduce Steering Vector
Ensembles (SVE), a method that averages multiple individually optimized
steering vectors, each targeting a specific bias axis such as age, race, or
gender. By leveraging their collective strength, SVE outperforms individual
steering vectors in both bias reduction and maintaining model performance. The
work presents the first systematic investigation of steering vectors for bias
mitigation, and we demonstrate that SVE is a powerful and computationally
efficient strategy for reducing bias in LLMs, with broader implications for
enhancing AI safety.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 12:25:29 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Siddique",
"Zara",
""
],
[
"Khalid",
"Irtaza",
""
],
[
"Turner",
"Liam D.",
""
],
[
"Espinosa-Anke",
"Luis",
""
]
]
| TITLE: Shifting Perspectives: Steering Vector Ensembles for Robust Bias
Mitigation in LLMs
ABSTRACT: We present a novel approach to bias mitigation in large language models
(LLMs) by applying steering vectors to modify model activations in forward
passes. We employ Bayesian optimization to systematically identify effective
contrastive pair datasets across nine bias axes. When optimized on the BBQ
dataset, our individually tuned steering vectors achieve average improvements
of 12.2%, 4.7%, and 3.2% over the baseline for Mistral, Llama, and Qwen,
respectively. Building on these promising results, we introduce Steering Vector
Ensembles (SVE), a method that averages multiple individually optimized
steering vectors, each targeting a specific bias axis such as age, race, or
gender. By leveraging their collective strength, SVE outperforms individual
steering vectors in both bias reduction and maintaining model performance. The
work presents the first systematic investigation of steering vectors for bias
mitigation, and we demonstrate that SVE is a powerful and computationally
efficient strategy for reducing bias in LLMs, with broader implications for
enhancing AI safety.
| no_new_dataset | 0.945651 |
2503.05373 | Linh Le | Linh Le, Guido Zuccon, Gianluca Demartini, Genghong Zhao, Xia Zhang | Leveraging Semantic Type Dependencies for Clinical Named Entity
Recognition | null | AMIA - American Medical Informatics Association 2022 | null | null | cs.CL | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Previous work on clinical relation extraction from free-text sentences
leveraged information about semantic types from clinical knowledge bases as a
part of entity representations. In this paper, we exploit additional evidence
by also making use of domain-specific semantic type dependencies. We encode the
relation between a span of tokens matching a Unified Medical Language System
(UMLS) concept and other tokens in the sentence. We implement our method and
compare against different named entity recognition (NER) architectures (i.e.,
BiLSTM-CRF and BiLSTM-GCN-CRF) using different pre-trained clinical embeddings
(i.e., BERT, BioBERT, UMLSBert). Our experimental results on clinical datasets
show that in some cases NER effectiveness can be significantly improved by
making use of domain-specific semantic type dependencies. Our work is also the
first study generating a matrix encoding to make use of more than three
dependencies in one pass for the NER task.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 12:29:21 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Le",
"Linh",
""
],
[
"Zuccon",
"Guido",
""
],
[
"Demartini",
"Gianluca",
""
],
[
"Zhao",
"Genghong",
""
],
[
"Zhang",
"Xia",
""
]
]
| TITLE: Leveraging Semantic Type Dependencies for Clinical Named Entity
Recognition
ABSTRACT: Previous work on clinical relation extraction from free-text sentences
leveraged information about semantic types from clinical knowledge bases as a
part of entity representations. In this paper, we exploit additional evidence
by also making use of domain-specific semantic type dependencies. We encode the
relation between a span of tokens matching a Unified Medical Language System
(UMLS) concept and other tokens in the sentence. We implement our method and
compare against different named entity recognition (NER) architectures (i.e.,
BiLSTM-CRF and BiLSTM-GCN-CRF) using different pre-trained clinical embeddings
(i.e., BERT, BioBERT, UMLSBert). Our experimental results on clinical datasets
show that in some cases NER effectiveness can be significantly improved by
making use of domain-specific semantic type dependencies. Our work is also the
first study generating a matrix encoding to make use of more than three
dependencies in one pass for the NER task.
| no_new_dataset | 0.946745 |
2503.05388 | Mohammad Javad Saeedizade | Anna Sofia Lippolis, Mohammad Javad Saeedizade, Robin Keskis\"arkk\"a,
Sara Zuppiroli, Miguel Ceriani, Aldo Gangemi, Eva Blomqvist, Andrea Giovanni
Nuzzolese | Ontology Generation using Large Language Models | 2 figures and 3 tables. 20 pages | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The ontology engineering process is complex, time-consuming, and error-prone,
even for experienced ontology engineers. In this work, we investigate the
potential of Large Language Models (LLMs) to provide effective OWL ontology
drafts directly from ontological requirements described using user stories and
competency questions. Our main contribution is the presentation and evaluation
of two new prompting techniques for automated ontology development: Memoryless
CQbyCQ and Ontogenia. We also emphasize the importance of three structural
criteria for ontology assessment, alongside expert qualitative evaluation,
highlighting the need for a multi-dimensional evaluation in order to capture
the quality and usability of the generated ontologies. Our experiments,
conducted on a benchmark dataset of ten ontologies with 100 distinct CQs and 29
different user stories, compare the performance of three LLMs using the two
prompting techniques. The results demonstrate improvements over the current
state-of-the-art in LLM-supported ontology engineering. More specifically, the
model OpenAI o1-preview with Ontogenia produces ontologies of sufficient
quality to meet the requirements of ontology engineers, significantly
outperforming novice ontology engineers in modelling ability. However, we still
note some common mistakes and variability of result quality, which is important
to take into account when using LLMs for ontology authoring support. We discuss
these limitations and propose directions for future research.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 13:03:28 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Lippolis",
"Anna Sofia",
""
],
[
"Saeedizade",
"Mohammad Javad",
""
],
[
"Keskisärkkä",
"Robin",
""
],
[
"Zuppiroli",
"Sara",
""
],
[
"Ceriani",
"Miguel",
""
],
[
"Gangemi",
"Aldo",
""
],
[
"Blomqvist",
"Eva",
""
],
[
"Nuzzolese",
"Andrea Giovanni",
""
]
]
| TITLE: Ontology Generation using Large Language Models
ABSTRACT: The ontology engineering process is complex, time-consuming, and error-prone,
even for experienced ontology engineers. In this work, we investigate the
potential of Large Language Models (LLMs) to provide effective OWL ontology
drafts directly from ontological requirements described using user stories and
competency questions. Our main contribution is the presentation and evaluation
of two new prompting techniques for automated ontology development: Memoryless
CQbyCQ and Ontogenia. We also emphasize the importance of three structural
criteria for ontology assessment, alongside expert qualitative evaluation,
highlighting the need for a multi-dimensional evaluation in order to capture
the quality and usability of the generated ontologies. Our experiments,
conducted on a benchmark dataset of ten ontologies with 100 distinct CQs and 29
different user stories, compare the performance of three LLMs using the two
prompting techniques. The results demonstrate improvements over the current
state-of-the-art in LLM-supported ontology engineering. More specifically, the
model OpenAI o1-preview with Ontogenia produces ontologies of sufficient
quality to meet the requirements of ontology engineers, significantly
outperforming novice ontology engineers in modelling ability. However, we still
note some common mistakes and variability of result quality, which is important
to take into account when using LLMs for ontology authoring support. We discuss
these limitations and propose directions for future research.
| no_new_dataset | 0.60092 |
2503.05423 | Run He | Run He, Di Fang, Yicheng Xu, Yawen Cui, Ming Li, Cen Chen, Ziqian
Zeng, Huiping Zhuang | Semantic Shift Estimation via Dual-Projection and Classifier
Reconstruction for Exemplar-Free Class-Incremental Learning | 14 pages, 7 figures | null | null | null | cs.CV cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Exemplar-Free Class-Incremental Learning (EFCIL) aims to sequentially learn
from distinct categories without retaining exemplars but easily suffers from
catastrophic forgetting of learned knowledge. While existing EFCIL methods
leverage knowledge distillation to alleviate forgetting, they still face two
critical challenges: semantic shift and decision bias. Specifically, the
embeddings of old tasks shift in the embedding space after learning new tasks,
and the classifier becomes biased towards new tasks due to training solely with
new data, thereby hindering the balance between old and new knowledge. To
address these issues, we propose the Dual-Projection Shift Estimation and
Classifier Reconstruction (DPCR) approach for EFCIL. DPCR effectively estimates
semantic shift through a dual-projection, which combines a learnable
transformation with a row-space projection to capture both task-wise and
category-wise shifts. Furthermore, to mitigate decision bias, DPCR employs
ridge regression to reformulate classifier training as a reconstruction
process. This reconstruction exploits previous information encoded in
covariance and prototype of each class after calibration with estimated shift,
thereby reducing decision bias. Extensive experiments demonstrate that, across
various datasets, DPCR effectively balances old and new tasks, outperforming
state-of-the-art EFCIL methods.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 13:50:29 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"He",
"Run",
""
],
[
"Fang",
"Di",
""
],
[
"Xu",
"Yicheng",
""
],
[
"Cui",
"Yawen",
""
],
[
"Li",
"Ming",
""
],
[
"Chen",
"Cen",
""
],
[
"Zeng",
"Ziqian",
""
],
[
"Zhuang",
"Huiping",
""
]
]
| TITLE: Semantic Shift Estimation via Dual-Projection and Classifier
Reconstruction for Exemplar-Free Class-Incremental Learning
ABSTRACT: Exemplar-Free Class-Incremental Learning (EFCIL) aims to sequentially learn
from distinct categories without retaining exemplars but easily suffers from
catastrophic forgetting of learned knowledge. While existing EFCIL methods
leverage knowledge distillation to alleviate forgetting, they still face two
critical challenges: semantic shift and decision bias. Specifically, the
embeddings of old tasks shift in the embedding space after learning new tasks,
and the classifier becomes biased towards new tasks due to training solely with
new data, thereby hindering the balance between old and new knowledge. To
address these issues, we propose the Dual-Projection Shift Estimation and
Classifier Reconstruction (DPCR) approach for EFCIL. DPCR effectively estimates
semantic shift through a dual-projection, which combines a learnable
transformation with a row-space projection to capture both task-wise and
category-wise shifts. Furthermore, to mitigate decision bias, DPCR employs
ridge regression to reformulate classifier training as a reconstruction
process. This reconstruction exploits previous information encoded in
covariance and prototype of each class after calibration with estimated shift,
thereby reducing decision bias. Extensive experiments demonstrate that, across
various datasets, DPCR effectively balances old and new tasks, outperforming
state-of-the-art EFCIL methods.
| no_new_dataset | 0.94625 |
2503.05425 | Huai Yu | Jian Shen, Huai Yu, Ji Wu, Wen Yang, Gui-Song Xia | LiDAR-enhanced 3D Gaussian Splatting Mapping | Accepted by ICRA 2025 | null | null | null | cs.RO | http://creativecommons.org/licenses/by/4.0/ | This paper introduces LiGSM, a novel LiDAR-enhanced 3D Gaussian Splatting
(3DGS) mapping framework that improves the accuracy and robustness of 3D scene
mapping by integrating LiDAR data. LiGSM constructs joint loss from images and
LiDAR point clouds to estimate the poses and optimize their extrinsic
parameters, enabling dynamic adaptation to variations in sensor alignment.
Furthermore, it leverages LiDAR point clouds to initialize 3DGS, providing a
denser and more reliable starting points compared to sparse SfM points. In
scene rendering, the framework augments standard image-based supervision with
depth maps generated from LiDAR projections, ensuring an accurate scene
representation in both geometry and photometry. Experiments on public and
self-collected datasets demonstrate that LiGSM outperforms comparative methods
in pose tracking and scene rendering.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 13:51:34 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Shen",
"Jian",
""
],
[
"Yu",
"Huai",
""
],
[
"Wu",
"Ji",
""
],
[
"Yang",
"Wen",
""
],
[
"Xia",
"Gui-Song",
""
]
]
| TITLE: LiDAR-enhanced 3D Gaussian Splatting Mapping
ABSTRACT: This paper introduces LiGSM, a novel LiDAR-enhanced 3D Gaussian Splatting
(3DGS) mapping framework that improves the accuracy and robustness of 3D scene
mapping by integrating LiDAR data. LiGSM constructs joint loss from images and
LiDAR point clouds to estimate the poses and optimize their extrinsic
parameters, enabling dynamic adaptation to variations in sensor alignment.
Furthermore, it leverages LiDAR point clouds to initialize 3DGS, providing a
denser and more reliable starting points compared to sparse SfM points. In
scene rendering, the framework augments standard image-based supervision with
depth maps generated from LiDAR projections, ensuring an accurate scene
representation in both geometry and photometry. Experiments on public and
self-collected datasets demonstrate that LiGSM outperforms comparative methods
in pose tracking and scene rendering.
| no_new_dataset | 0.662469 |
2503.05427 | Simon Malacek | Simon Malacek, Jos\'e Portela, Yannick Marcus Werner, Sonja Wogrin | Generating Building-Level Heat Demand Time Series by Combining Occupancy
Simulations and Thermal Modeling | null | null | null | null | eess.SY cs.SY | http://creativecommons.org/licenses/by/4.0/ | Despite various efforts, decarbonizing the heating sector remains a
significant challenge. To tackle it by smart planning, the availability of
highly resolved heating demand data is key. Several existing models provide
heating demand only for specific applications. Typically, they either offer
time series for a larger area or annual demand data on a building level, but
not both simultaneously. Additionally, the diversity in heating demand across
different buildings is often not considered. To address these limitations, this
paper presents a novel method for generating temporally resolved heat demand
time series at the building level using publicly available data. The approach
integrates a thermal building model with stochastic occupancy simulations that
account for variability in user behavior. As a result, the tool serves as a
cost-effective resource for cross-sectoral energy system planning and policy
development, particularly with a focus on the heating sector. The obtained data
can be used to assess the impact of renovation and retrofitting strategies, or
to analyze district heating expansion. To illustrate the potential applications
of this approach, we conducted a case study in Puertollano (Spain), where we
prepared a dataset of heating demand with hourly resolution for each of 9,298
residential buildings. This data was then used to compare two different
pathways for the thermal renovation of these buildings. By relying on publicly
available data, this method can be adapted and applied to various European
regions, offering broad usability in energy system optimization and analysis of
decarbonization strategies.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 13:56:14 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Malacek",
"Simon",
""
],
[
"Portela",
"José",
""
],
[
"Werner",
"Yannick Marcus",
""
],
[
"Wogrin",
"Sonja",
""
]
]
| TITLE: Generating Building-Level Heat Demand Time Series by Combining Occupancy
Simulations and Thermal Modeling
ABSTRACT: Despite various efforts, decarbonizing the heating sector remains a
significant challenge. To tackle it by smart planning, the availability of
highly resolved heating demand data is key. Several existing models provide
heating demand only for specific applications. Typically, they either offer
time series for a larger area or annual demand data on a building level, but
not both simultaneously. Additionally, the diversity in heating demand across
different buildings is often not considered. To address these limitations, this
paper presents a novel method for generating temporally resolved heat demand
time series at the building level using publicly available data. The approach
integrates a thermal building model with stochastic occupancy simulations that
account for variability in user behavior. As a result, the tool serves as a
cost-effective resource for cross-sectoral energy system planning and policy
development, particularly with a focus on the heating sector. The obtained data
can be used to assess the impact of renovation and retrofitting strategies, or
to analyze district heating expansion. To illustrate the potential applications
of this approach, we conducted a case study in Puertollano (Spain), where we
prepared a dataset of heating demand with hourly resolution for each of 9,298
residential buildings. This data was then used to compare two different
pathways for the thermal renovation of these buildings. By relying on publicly
available data, this method can be adapted and applied to various European
regions, offering broad usability in energy system optimization and analysis of
decarbonization strategies.
| new_dataset | 0.971699 |
2503.05439 | Lachlan McPheat | Navdeep Kaur, Lachlan McPheat, Alessandra Russo, Anthony G Cohn,
Pranava Madhyastha | An Empirical Study of Conformal Prediction in LLM with ASP Scaffolds for
Robust Reasoning | null | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | In this paper, we examine the use of Conformal Language Modelling (CLM)
alongside Answer Set Programming (ASP) to enhance the performance of standard
open-weight LLMs on complex multi-step reasoning tasks. Using the StepGame
dataset, which requires spatial reasoning, we apply CLM to generate sets of ASP
programs from an LLM, providing statistical guarantees on the correctness of
the outputs. Experimental results show that CLM significantly outperforms
baseline models that use standard sampling methods, achieving substantial
accuracy improvements across different levels of reasoning complexity.
Additionally, the LLM-as-Judge metric enhances CLM's performance, especially in
assessing structurally and logically correct ASP outputs. However, calibrating
CLM with diverse calibration sets did not improve generalizability for tasks
requiring much longer reasoning steps, indicating limitations in handling more
complex tasks.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 14:10:10 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Kaur",
"Navdeep",
""
],
[
"McPheat",
"Lachlan",
""
],
[
"Russo",
"Alessandra",
""
],
[
"Cohn",
"Anthony G",
""
],
[
"Madhyastha",
"Pranava",
""
]
]
| TITLE: An Empirical Study of Conformal Prediction in LLM with ASP Scaffolds for
Robust Reasoning
ABSTRACT: In this paper, we examine the use of Conformal Language Modelling (CLM)
alongside Answer Set Programming (ASP) to enhance the performance of standard
open-weight LLMs on complex multi-step reasoning tasks. Using the StepGame
dataset, which requires spatial reasoning, we apply CLM to generate sets of ASP
programs from an LLM, providing statistical guarantees on the correctness of
the outputs. Experimental results show that CLM significantly outperforms
baseline models that use standard sampling methods, achieving substantial
accuracy improvements across different levels of reasoning complexity.
Additionally, the LLM-as-Judge metric enhances CLM's performance, especially in
assessing structurally and logically correct ASP outputs. However, calibrating
CLM with diverse calibration sets did not improve generalizability for tasks
requiring much longer reasoning steps, indicating limitations in handling more
complex tasks.
| no_new_dataset | 0.945045 |
2503.05474 | Ziran Zhou | Ziran Zhou and Guanyu Gao and Xiaohu Wu and Yan Lyu | Personalized Federated Learning via Learning Dynamic Graphs | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Personalized Federated Learning (PFL) aims to train a personalized model for
each client that is tailored to its local data distribution, learning fails to
perform well on individual clients due to variations in their local data
distributions. Most existing PFL methods focus on personalizing the aggregated
global model for each client, neglecting the fundamental aspect of federated
learning: the regulation of how client models are aggregated. Additionally,
almost all of them overlook the graph structure formed by clients in federated
learning. In this paper, we propose a novel method, Personalized Federated
Learning with Graph Attention Network (pFedGAT), which captures the latent
graph structure between clients and dynamically determines the importance of
other clients for each client, enabling fine-grained control over the
aggregation process. We evaluate pFedGAT across multiple data distribution
scenarios, comparing it with twelve state of the art methods on three datasets:
Fashion MNIST, CIFAR-10, and CIFAR-100, and find that it consistently performs
well.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 14:47:03 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Zhou",
"Ziran",
""
],
[
"Gao",
"Guanyu",
""
],
[
"Wu",
"Xiaohu",
""
],
[
"Lyu",
"Yan",
""
]
]
| TITLE: Personalized Federated Learning via Learning Dynamic Graphs
ABSTRACT: Personalized Federated Learning (PFL) aims to train a personalized model for
each client that is tailored to its local data distribution, learning fails to
perform well on individual clients due to variations in their local data
distributions. Most existing PFL methods focus on personalizing the aggregated
global model for each client, neglecting the fundamental aspect of federated
learning: the regulation of how client models are aggregated. Additionally,
almost all of them overlook the graph structure formed by clients in federated
learning. In this paper, we propose a novel method, Personalized Federated
Learning with Graph Attention Network (pFedGAT), which captures the latent
graph structure between clients and dynamically determines the importance of
other clients for each client, enabling fine-grained control over the
aggregation process. We evaluate pFedGAT across multiple data distribution
scenarios, comparing it with twelve state of the art methods on three datasets:
Fashion MNIST, CIFAR-10, and CIFAR-100, and find that it consistently performs
well.
| no_new_dataset | 0.948298 |
2503.05477 | Gazi Tanbhir | Nizo Jaman Shohan, Gazi Tanbhir, Faria Elahi, Ahsan Ullah, Md. Nazmus
Sakib | Enhancing Network Security: A Hybrid Approach for Detection and
Mitigation of Distributed Denial-of-Service Attacks Using Machine Learning | Part of the book series: Communications in Computer and Information
Science ((CCIS,volume 2091)) | ANTIC 2023. Communications in Computer and Information Science,
vol 2091. Springer, Cham | 10.1007/978-3-031-64064-3_7 | null | cs.CR cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The distributed denial-of-service (DDoS) attack stands out as a highly
formidable cyber threat, representing an advanced form of the denial-of-service
(DoS) attack. A DDoS attack involves multiple computers working together to
overwhelm a system, making it unavailable. On the other hand, a DoS attack is a
one-on-one attempt to make a system or website inaccessible. Thus, it is
crucial to construct an effective model for identifying various DDoS incidents.
Although extensive research has focused on binary detection models for DDoS
identification, they face challenges to adapt evolving threats, necessitating
frequent updates. Whereas multiclass detection models offer a comprehensive
defense against diverse DDoS attacks, ensuring adaptability in the
ever-changing cyber threat landscape. In this paper, we propose a Hybrid Model
to strengthen network security by combining the featureextraction abilities of
1D Convolutional Neural Networks (CNNs) with the classification skills of
Random Forest (RF) and Multi-layer Perceptron (MLP) classifiers. Using the
CIC-DDoS2019 dataset, we perform multiclass classification of various DDoS
attacks and conduct a comparative analysis of evaluation metrics for RF, MLP,
and our proposed Hybrid Model. After analyzing the results, we draw meaningful
conclusions and confirm the superiority of our Hybrid Model by performing
thorough cross-validation. Additionally, we integrate our machine learning
model with Snort, which provides a robust and adaptive solution for detecting
and mitigating various DDoS attacks.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 14:47:56 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Shohan",
"Nizo Jaman",
""
],
[
"Tanbhir",
"Gazi",
""
],
[
"Elahi",
"Faria",
""
],
[
"Ullah",
"Ahsan",
""
],
[
"Sakib",
"Md. Nazmus",
""
]
]
| TITLE: Enhancing Network Security: A Hybrid Approach for Detection and
Mitigation of Distributed Denial-of-Service Attacks Using Machine Learning
ABSTRACT: The distributed denial-of-service (DDoS) attack stands out as a highly
formidable cyber threat, representing an advanced form of the denial-of-service
(DoS) attack. A DDoS attack involves multiple computers working together to
overwhelm a system, making it unavailable. On the other hand, a DoS attack is a
one-on-one attempt to make a system or website inaccessible. Thus, it is
crucial to construct an effective model for identifying various DDoS incidents.
Although extensive research has focused on binary detection models for DDoS
identification, they face challenges to adapt evolving threats, necessitating
frequent updates. Whereas multiclass detection models offer a comprehensive
defense against diverse DDoS attacks, ensuring adaptability in the
ever-changing cyber threat landscape. In this paper, we propose a Hybrid Model
to strengthen network security by combining the featureextraction abilities of
1D Convolutional Neural Networks (CNNs) with the classification skills of
Random Forest (RF) and Multi-layer Perceptron (MLP) classifiers. Using the
CIC-DDoS2019 dataset, we perform multiclass classification of various DDoS
attacks and conduct a comparative analysis of evaluation metrics for RF, MLP,
and our proposed Hybrid Model. After analyzing the results, we draw meaningful
conclusions and confirm the superiority of our Hybrid Model by performing
thorough cross-validation. Additionally, we integrate our machine learning
model with Snort, which provides a robust and adaptive solution for detecting
and mitigating various DDoS attacks.
| no_new_dataset | 0.945701 |
2503.05482 | Christofer Fellicious | Christofer Fellicious, Hans P. Reiser, Michael Granitzer | Bridging the Semantic Gap in Virtual Machine Introspection and Forensic
Memory Analysis | null | null | null | null | cs.CR cs.LG | http://creativecommons.org/licenses/by/4.0/ | Forensic Memory Analysis (FMA) and Virtual Machine Introspection (VMI) are
critical tools for security in a virtualization-based approach. VMI and FMA
involves using digital forensic methods to extract information from the system
to identify and explain security incidents. A key challenge in both FMA and VMI
is the "Semantic Gap", which is the difficulty of interpreting raw memory data
without specialized tools and expertise. In this work, we investigate how a
priori knowledge, metadata and engineered features can aid VMI and FMA,
leveraging machine learning to automate information extraction and reduce the
workload of forensic investigators. We choose OpenSSH as our use case to test
different methods to extract high level structures. We also test our method on
complete physical memory dumps to showcase the effectiveness of the engineered
features. Our features range from basic statistical features to advanced
graph-based representations using malloc headers and pointer translations. The
training and testing are carried out on public datasets that we compare against
already recognized baseline methods. We show that using metadata, we can
improve the performance of the algorithm when there is very little training
data and also quantify how having more data results in better generalization
performance. The final contribution is an open dataset of physical memory
dumps, totalling more than 1 TB of different memory state, software
environments, main memory capacities and operating system versions. Our methods
show that having more metadata boosts performance with all methods obtaining an
F1-Score of over 80%. Our research underscores the possibility of using feature
engineering and machine learning techniques to bridge the semantic gap.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 14:51:32 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Fellicious",
"Christofer",
""
],
[
"Reiser",
"Hans P.",
""
],
[
"Granitzer",
"Michael",
""
]
]
| TITLE: Bridging the Semantic Gap in Virtual Machine Introspection and Forensic
Memory Analysis
ABSTRACT: Forensic Memory Analysis (FMA) and Virtual Machine Introspection (VMI) are
critical tools for security in a virtualization-based approach. VMI and FMA
involves using digital forensic methods to extract information from the system
to identify and explain security incidents. A key challenge in both FMA and VMI
is the "Semantic Gap", which is the difficulty of interpreting raw memory data
without specialized tools and expertise. In this work, we investigate how a
priori knowledge, metadata and engineered features can aid VMI and FMA,
leveraging machine learning to automate information extraction and reduce the
workload of forensic investigators. We choose OpenSSH as our use case to test
different methods to extract high level structures. We also test our method on
complete physical memory dumps to showcase the effectiveness of the engineered
features. Our features range from basic statistical features to advanced
graph-based representations using malloc headers and pointer translations. The
training and testing are carried out on public datasets that we compare against
already recognized baseline methods. We show that using metadata, we can
improve the performance of the algorithm when there is very little training
data and also quantify how having more data results in better generalization
performance. The final contribution is an open dataset of physical memory
dumps, totalling more than 1 TB of different memory state, software
environments, main memory capacities and operating system versions. Our methods
show that having more metadata boosts performance with all methods obtaining an
F1-Score of over 80%. Our research underscores the possibility of using feature
engineering and machine learning techniques to bridge the semantic gap.
| no_new_dataset | 0.951097 |
2503.05490 | Amit Levy | Amit Levy, Itzik Klein | Adaptive Neural Unscented Kalman Filter | eight pages, ten figures | null | null | null | cs.RO eess.SP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The unscented Kalman filter is an algorithm capable of handling nonlinear
scenarios. Uncertainty in process noise covariance may decrease the filter
estimation performance or even lead to its divergence. Therefore, it is
important to adjust the process noise covariance matrix in real time. In this
paper, we developed an adaptive neural unscented Kalman filter to cope with
time-varying uncertainties during platform operation. To this end, we devised
ProcessNet, a simple yet efficient end-to-end regression network to adaptively
estimate the process noise covariance matrix. We focused on the nonlinear
inertial sensor and Doppler velocity log fusion problem in the case of
autonomous underwater vehicle navigation. Using a real-world recorded dataset
from an autonomous underwater vehicle, we demonstrated our filter performance
and showed its advantages over other adaptive and non-adaptive nonlinear
filters.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 14:59:30 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Levy",
"Amit",
""
],
[
"Klein",
"Itzik",
""
]
]
| TITLE: Adaptive Neural Unscented Kalman Filter
ABSTRACT: The unscented Kalman filter is an algorithm capable of handling nonlinear
scenarios. Uncertainty in process noise covariance may decrease the filter
estimation performance or even lead to its divergence. Therefore, it is
important to adjust the process noise covariance matrix in real time. In this
paper, we developed an adaptive neural unscented Kalman filter to cope with
time-varying uncertainties during platform operation. To this end, we devised
ProcessNet, a simple yet efficient end-to-end regression network to adaptively
estimate the process noise covariance matrix. We focused on the nonlinear
inertial sensor and Doppler velocity log fusion problem in the case of
autonomous underwater vehicle navigation. Using a real-world recorded dataset
from an autonomous underwater vehicle, we demonstrated our filter performance
and showed its advantages over other adaptive and non-adaptive nonlinear
filters.
| no_new_dataset | 0.950824 |
2503.05492 | Haotian Hu | Haotian Hu and Jingwei Xu and Fanyi Wang and Toyota Li and Yaonong
Wang and Laifeng Hu and Zhiwang Zhang | FastMap: Fast Queries Initialization Based Vectorized HD Map
Reconstruction Framework | null | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Reconstruction of high-definition maps is a crucial task in perceiving the
autonomous driving environment, as its accuracy directly impacts the
reliability of prediction and planning capabilities in downstream modules.
Current vectorized map reconstruction methods based on the DETR framework
encounter limitations due to the redundancy in the decoder structure,
necessitating the stacking of six decoder layers to maintain performance, which
significantly hampers computational efficiency. To tackle this issue, we
introduce FastMap, an innovative framework designed to reduce decoder
redundancy in existing approaches. FastMap optimizes the decoder architecture
by employing a single-layer, two-stage transformer that achieves multilevel
representation capabilities. Our framework eliminates the conventional practice
of randomly initializing queries and instead incorporates a heatmap-guided
query generation module during the decoding phase, which effectively maps image
features into structured query vectors using learnable positional encoding.
Additionally, we propose a geometry-constrained point-to-line loss mechanism
for FastMap, which adeptly addresses the challenge of distinguishing highly
homogeneous features that often arise in traditional point-to-point loss
computations. Extensive experiments demonstrate that FastMap achieves
state-of-the-art performance in both nuScenes and Argoverse2 datasets, with its
decoder operating 3.2 faster than the baseline. Code and more demos are
available at https://github.com/hht1996ok/FastMap.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 15:01:55 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Hu",
"Haotian",
""
],
[
"Xu",
"Jingwei",
""
],
[
"Wang",
"Fanyi",
""
],
[
"Li",
"Toyota",
""
],
[
"Wang",
"Yaonong",
""
],
[
"Hu",
"Laifeng",
""
],
[
"Zhang",
"Zhiwang",
""
]
]
| TITLE: FastMap: Fast Queries Initialization Based Vectorized HD Map
Reconstruction Framework
ABSTRACT: Reconstruction of high-definition maps is a crucial task in perceiving the
autonomous driving environment, as its accuracy directly impacts the
reliability of prediction and planning capabilities in downstream modules.
Current vectorized map reconstruction methods based on the DETR framework
encounter limitations due to the redundancy in the decoder structure,
necessitating the stacking of six decoder layers to maintain performance, which
significantly hampers computational efficiency. To tackle this issue, we
introduce FastMap, an innovative framework designed to reduce decoder
redundancy in existing approaches. FastMap optimizes the decoder architecture
by employing a single-layer, two-stage transformer that achieves multilevel
representation capabilities. Our framework eliminates the conventional practice
of randomly initializing queries and instead incorporates a heatmap-guided
query generation module during the decoding phase, which effectively maps image
features into structured query vectors using learnable positional encoding.
Additionally, we propose a geometry-constrained point-to-line loss mechanism
for FastMap, which adeptly addresses the challenge of distinguishing highly
homogeneous features that often arise in traditional point-to-point loss
computations. Extensive experiments demonstrate that FastMap achieves
state-of-the-art performance in both nuScenes and Argoverse2 datasets, with its
decoder operating 3.2 faster than the baseline. Code and more demos are
available at https://github.com/hht1996ok/FastMap.
| no_new_dataset | 0.943452 |
2503.05493 | Qijiong Liu | Qijiong Liu, Jieming Zhu, Lu Fan, Kun Wang, Hengchang Hu, Wei Guo,
Yong Liu, Xiao-Ming Wu | Benchmarking LLMs in Recommendation Tasks: A Comparative Evaluation with
Conventional Recommenders | null | null | null | null | cs.IR cs.CL | http://creativecommons.org/licenses/by/4.0/ | In recent years, integrating large language models (LLMs) into recommender
systems has created new opportunities for improving recommendation quality.
However, a comprehensive benchmark is needed to thoroughly evaluate and compare
the recommendation capabilities of LLMs with traditional recommender systems.
In this paper, we introduce RecBench, which systematically investigates various
item representation forms (including unique identifier, text, semantic
embedding, and semantic identifier) and evaluates two primary recommendation
tasks, i.e., click-through rate prediction (CTR) and sequential recommendation
(SeqRec). Our extensive experiments cover up to 17 large models and are
conducted across five diverse datasets from fashion, news, video, books, and
music domains. Our findings indicate that LLM-based recommenders outperform
conventional recommenders, achieving up to a 5% AUC improvement in the CTR
scenario and up to a 170% NDCG@10 improvement in the SeqRec scenario. However,
these substantial performance gains come at the expense of significantly
reduced inference efficiency, rendering the LLM-as-RS paradigm impractical for
real-time recommendation environments. We aim for our findings to inspire
future research, including recommendation-specific model acceleration methods.
We will release our code, data, configurations, and platform to enable other
researchers to reproduce and build upon our experimental results.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 15:05:23 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Liu",
"Qijiong",
""
],
[
"Zhu",
"Jieming",
""
],
[
"Fan",
"Lu",
""
],
[
"Wang",
"Kun",
""
],
[
"Hu",
"Hengchang",
""
],
[
"Guo",
"Wei",
""
],
[
"Liu",
"Yong",
""
],
[
"Wu",
"Xiao-Ming",
""
]
]
| TITLE: Benchmarking LLMs in Recommendation Tasks: A Comparative Evaluation with
Conventional Recommenders
ABSTRACT: In recent years, integrating large language models (LLMs) into recommender
systems has created new opportunities for improving recommendation quality.
However, a comprehensive benchmark is needed to thoroughly evaluate and compare
the recommendation capabilities of LLMs with traditional recommender systems.
In this paper, we introduce RecBench, which systematically investigates various
item representation forms (including unique identifier, text, semantic
embedding, and semantic identifier) and evaluates two primary recommendation
tasks, i.e., click-through rate prediction (CTR) and sequential recommendation
(SeqRec). Our extensive experiments cover up to 17 large models and are
conducted across five diverse datasets from fashion, news, video, books, and
music domains. Our findings indicate that LLM-based recommenders outperform
conventional recommenders, achieving up to a 5% AUC improvement in the CTR
scenario and up to a 170% NDCG@10 improvement in the SeqRec scenario. However,
these substantial performance gains come at the expense of significantly
reduced inference efficiency, rendering the LLM-as-RS paradigm impractical for
real-time recommendation environments. We aim for our findings to inspire
future research, including recommendation-specific model acceleration methods.
We will release our code, data, configurations, and platform to enable other
researchers to reproduce and build upon our experimental results.
| no_new_dataset | 0.772144 |
2503.05505 | Yusong Ke | Yusong Ke | Statistical Guarantees of Correctness Coverage for Medical
Multiple-Choice Question Answering | Under Review | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Large language models (LLMs) are increasingly deployed in real-world
question-answering (QA) applications. However, LLMs have been proven to
generate hallucinations and nonfactual information, undermining their
trustworthiness in high-stakes medical tasks. Conformal prediction (CP) is
well-known to be model-agnostic and distribution-free, which creates
statistically rigorous prediction sets in classification tasks. In this work,
we for the first time adapt the CP framework to medical multiple-choice
question-answering (MCQA) tasks, by correlating the nonconformity score with
the frequency score of correct options grounded in self-consistency theory,
assuming no access to internal model information. Considering that the adapted
CP framework can only control the (mis)coverage rate, we employ a risk control
framework, which can manage task-specific metrics by devising a monotonically
decreasing loss function. We evaluate our framework on 3 popular medical MCQA
datasets utilizing 4 ``off-the-shelf'' LLMs. Empirical results demonstrate that
we achieve user-specified average (or marginal) error rates on the test set.
Furthermore, we observe that the average prediction set size (APSS) on the test
set decreases as the risk level increases, which concludes a promising
evaluation metric for the uncertainty of LLMs.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 15:22:10 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Ke",
"Yusong",
""
]
]
| TITLE: Statistical Guarantees of Correctness Coverage for Medical
Multiple-Choice Question Answering
ABSTRACT: Large language models (LLMs) are increasingly deployed in real-world
question-answering (QA) applications. However, LLMs have been proven to
generate hallucinations and nonfactual information, undermining their
trustworthiness in high-stakes medical tasks. Conformal prediction (CP) is
well-known to be model-agnostic and distribution-free, which creates
statistically rigorous prediction sets in classification tasks. In this work,
we for the first time adapt the CP framework to medical multiple-choice
question-answering (MCQA) tasks, by correlating the nonconformity score with
the frequency score of correct options grounded in self-consistency theory,
assuming no access to internal model information. Considering that the adapted
CP framework can only control the (mis)coverage rate, we employ a risk control
framework, which can manage task-specific metrics by devising a monotonically
decreasing loss function. We evaluate our framework on 3 popular medical MCQA
datasets utilizing 4 ``off-the-shelf'' LLMs. Empirical results demonstrate that
we achieve user-specified average (or marginal) error rates on the test set.
Furthermore, we observe that the average prediction set size (APSS) on the test
set decreases as the risk level increases, which concludes a promising
evaluation metric for the uncertainty of LLMs.
| no_new_dataset | 0.947527 |
2503.05520 | Romain Hermary | Romain Hermary, Vincent Gaudilli\`ere, Abd El Rahman Shabayek, Djamila
Aouada | Removing Geometric Bias in One-Class Anomaly Detection with Adaptive
Feature Perturbation | Published in WACV 2025 | null | null | null | cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | One-class anomaly detection aims to detect objects that do not belong to a
predefined normal class. In practice training data lack those anomalous
samples; hence state-of-the-art methods are trained to discriminate between
normal and synthetically-generated pseudo-anomalous data. Most methods use data
augmentation techniques on normal images to simulate anomalies. However the
best-performing ones implicitly leverage a geometric bias present in the
benchmarking datasets. This limits their usability in more general conditions.
Others are relying on basic noising schemes that may be suboptimal in capturing
the underlying structure of normal data. In addition most still favour the
image domain to generate pseudo-anomalies training models end-to-end from only
the normal class and overlooking richer representations of the information. To
overcome these limitations we consider frozen yet rich feature spaces given by
pretrained models and create pseudo-anomalous features with a novel adaptive
linear feature perturbation technique. It adapts the noise distribution to each
sample applies decaying linear perturbations to feature vectors and further
guides the classification process using a contrastive learning objective.
Experimental evaluation conducted on both standard and geometric bias-free
datasets demonstrates the superiority of our approach with respect to
comparable baselines. The codebase is accessible via our public repository.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 15:42:51 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Hermary",
"Romain",
""
],
[
"Gaudillière",
"Vincent",
""
],
[
"Shabayek",
"Abd El Rahman",
""
],
[
"Aouada",
"Djamila",
""
]
]
| TITLE: Removing Geometric Bias in One-Class Anomaly Detection with Adaptive
Feature Perturbation
ABSTRACT: One-class anomaly detection aims to detect objects that do not belong to a
predefined normal class. In practice training data lack those anomalous
samples; hence state-of-the-art methods are trained to discriminate between
normal and synthetically-generated pseudo-anomalous data. Most methods use data
augmentation techniques on normal images to simulate anomalies. However the
best-performing ones implicitly leverage a geometric bias present in the
benchmarking datasets. This limits their usability in more general conditions.
Others are relying on basic noising schemes that may be suboptimal in capturing
the underlying structure of normal data. In addition most still favour the
image domain to generate pseudo-anomalies training models end-to-end from only
the normal class and overlooking richer representations of the information. To
overcome these limitations we consider frozen yet rich feature spaces given by
pretrained models and create pseudo-anomalous features with a novel adaptive
linear feature perturbation technique. It adapts the noise distribution to each
sample applies decaying linear perturbations to feature vectors and further
guides the classification process using a contrastive learning objective.
Experimental evaluation conducted on both standard and geometric bias-free
datasets demonstrates the superiority of our approach with respect to
comparable baselines. The codebase is accessible via our public repository.
| no_new_dataset | 0.947817 |
2503.05522 | Eren Erogullari | Eren Erogullari, Sebastian Lapuschkin, Wojciech Samek, Frederik Pahde | Post-Hoc Concept Disentanglement: From Correlated to Isolated Concept
Representations | null | null | null | null | cs.CV cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Concept Activation Vectors (CAVs) are widely used to model
human-understandable concepts as directions within the latent space of neural
networks. They are trained by identifying directions from the activations of
concept samples to those of non-concept samples. However, this method often
produces similar, non-orthogonal directions for correlated concepts, such as
"beard" and "necktie" within the CelebA dataset, which frequently co-occur in
images of men. This entanglement complicates the interpretation of concepts in
isolation and can lead to undesired effects in CAV applications, such as
activation steering. To address this issue, we introduce a post-hoc concept
disentanglement method that employs a non-orthogonality loss, facilitating the
identification of orthogonal concept directions while preserving directional
correctness. We evaluate our approach with real-world and controlled correlated
concepts in CelebA and a synthetic FunnyBirds dataset with VGG16 and ResNet18
architectures. We further demonstrate the superiority of orthogonalized concept
representations in activation steering tasks, allowing (1) the insertion of
isolated concepts into input images through generative models and (2) the
removal of concepts for effective shortcut suppression with reduced impact on
correlated concepts in comparison to baseline CAVs.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 15:45:43 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Erogullari",
"Eren",
""
],
[
"Lapuschkin",
"Sebastian",
""
],
[
"Samek",
"Wojciech",
""
],
[
"Pahde",
"Frederik",
""
]
]
| TITLE: Post-Hoc Concept Disentanglement: From Correlated to Isolated Concept
Representations
ABSTRACT: Concept Activation Vectors (CAVs) are widely used to model
human-understandable concepts as directions within the latent space of neural
networks. They are trained by identifying directions from the activations of
concept samples to those of non-concept samples. However, this method often
produces similar, non-orthogonal directions for correlated concepts, such as
"beard" and "necktie" within the CelebA dataset, which frequently co-occur in
images of men. This entanglement complicates the interpretation of concepts in
isolation and can lead to undesired effects in CAV applications, such as
activation steering. To address this issue, we introduce a post-hoc concept
disentanglement method that employs a non-orthogonality loss, facilitating the
identification of orthogonal concept directions while preserving directional
correctness. We evaluate our approach with real-world and controlled correlated
concepts in CelebA and a synthetic FunnyBirds dataset with VGG16 and ResNet18
architectures. We further demonstrate the superiority of orthogonalized concept
representations in activation steering tasks, allowing (1) the insertion of
isolated concepts into input images through generative models and (2) the
removal of concepts for effective shortcut suppression with reduced impact on
correlated concepts in comparison to baseline CAVs.
| no_new_dataset | 0.947332 |
2503.05531 | Alex Fedorov | Alex Fedorov, Yutong Bu, Xiao Hu, Chris Rorden, Sergey Plis | State-of-the-Art Stroke Lesion Segmentation at 1/1000th of Parameters | International Symposium on Biomedical Imaging, April 14-17, 2025 | null | null | null | eess.IV cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Efficient and accurate whole-brain lesion segmentation remains a challenge in
medical image analysis. In this work, we revisit MeshNet, a parameter-efficient
segmentation model, and introduce a novel multi-scale dilation pattern with an
encoder-decoder structure. This innovation enables capturing broad contextual
information and fine-grained details without traditional downsampling,
upsampling, or skip-connections. Unlike previous approaches processing
subvolumes or slices, we operate directly on whole-brain $256^3$ MRI volumes.
Evaluations on the Aphasia Recovery Cohort (ARC) dataset demonstrate that
MeshNet achieves superior or comparable DICE scores to state-of-the-art
architectures such as MedNeXt and U-MAMBA at 1/1000th of parameters. Our
results validate MeshNet's strong balance of efficiency and performance, making
it particularly suitable for resource-limited environments such as web-based
applications and opening new possibilities for the widespread deployment of
advanced medical image analysis tools.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 15:58:36 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Fedorov",
"Alex",
""
],
[
"Bu",
"Yutong",
""
],
[
"Hu",
"Xiao",
""
],
[
"Rorden",
"Chris",
""
],
[
"Plis",
"Sergey",
""
]
]
| TITLE: State-of-the-Art Stroke Lesion Segmentation at 1/1000th of Parameters
ABSTRACT: Efficient and accurate whole-brain lesion segmentation remains a challenge in
medical image analysis. In this work, we revisit MeshNet, a parameter-efficient
segmentation model, and introduce a novel multi-scale dilation pattern with an
encoder-decoder structure. This innovation enables capturing broad contextual
information and fine-grained details without traditional downsampling,
upsampling, or skip-connections. Unlike previous approaches processing
subvolumes or slices, we operate directly on whole-brain $256^3$ MRI volumes.
Evaluations on the Aphasia Recovery Cohort (ARC) dataset demonstrate that
MeshNet achieves superior or comparable DICE scores to state-of-the-art
architectures such as MedNeXt and U-MAMBA at 1/1000th of parameters. Our
results validate MeshNet's strong balance of efficiency and performance, making
it particularly suitable for resource-limited environments such as web-based
applications and opening new possibilities for the widespread deployment of
advanced medical image analysis tools.
| no_new_dataset | 0.942981 |
2503.05534 | Adrien Meyer | Adrien Meyer, Lorenzo Arboit, Giuseppe Massimiani, Francesco Brucchi,
Luca Emanuele Amodio, Didier Mutter, Nicolas Padoy | S4M: Segment Anything with 4 Extreme Points | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | The Segment Anything Model (SAM) has revolutionized open-set interactive
image segmentation, inspiring numerous adapters for the medical domain.
However, SAM primarily relies on sparse prompts such as point or bounding box,
which may be suboptimal for fine-grained instance segmentation, particularly in
endoscopic imagery, where precise localization is critical and existing prompts
struggle to capture object boundaries effectively. To address this, we
introduce S4M (Segment Anything with 4 Extreme Points), which augments SAM by
leveraging extreme points -- the top-, bottom-, left-, and right-most points of
an instance -- prompts. These points are intuitive to identify and provide a
faster, structured alternative to box prompts. However, a na\"ive use of
extreme points degrades performance, due to SAM's inability to interpret their
semantic roles. To resolve this, we introduce dedicated learnable embeddings,
enabling the model to distinguish extreme points from generic free-form points
and better reason about their spatial relationships. We further propose an
auxiliary training task through the Canvas module, which operates solely on
prompts -- without vision input -- to predict a coarse instance mask. This
encourages the model to internalize the relationship between extreme points and
mask distributions, leading to more robust segmentation. S4M outperforms other
SAM-based approaches on three endoscopic surgical datasets, demonstrating its
effectiveness in complex scenarios. Finally, we validate our approach through a
human annotation study on surgical endoscopic videos, confirming that extreme
points are faster to acquire than bounding boxes.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 16:02:11 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Meyer",
"Adrien",
""
],
[
"Arboit",
"Lorenzo",
""
],
[
"Massimiani",
"Giuseppe",
""
],
[
"Brucchi",
"Francesco",
""
],
[
"Amodio",
"Luca Emanuele",
""
],
[
"Mutter",
"Didier",
""
],
[
"Padoy",
"Nicolas",
""
]
]
| TITLE: S4M: Segment Anything with 4 Extreme Points
ABSTRACT: The Segment Anything Model (SAM) has revolutionized open-set interactive
image segmentation, inspiring numerous adapters for the medical domain.
However, SAM primarily relies on sparse prompts such as point or bounding box,
which may be suboptimal for fine-grained instance segmentation, particularly in
endoscopic imagery, where precise localization is critical and existing prompts
struggle to capture object boundaries effectively. To address this, we
introduce S4M (Segment Anything with 4 Extreme Points), which augments SAM by
leveraging extreme points -- the top-, bottom-, left-, and right-most points of
an instance -- prompts. These points are intuitive to identify and provide a
faster, structured alternative to box prompts. However, a na\"ive use of
extreme points degrades performance, due to SAM's inability to interpret their
semantic roles. To resolve this, we introduce dedicated learnable embeddings,
enabling the model to distinguish extreme points from generic free-form points
and better reason about their spatial relationships. We further propose an
auxiliary training task through the Canvas module, which operates solely on
prompts -- without vision input -- to predict a coarse instance mask. This
encourages the model to internalize the relationship between extreme points and
mask distributions, leading to more robust segmentation. S4M outperforms other
SAM-based approaches on three endoscopic surgical datasets, demonstrating its
effectiveness in complex scenarios. Finally, we validate our approach through a
human annotation study on surgical endoscopic videos, confirming that extreme
points are faster to acquire than bounding boxes.
| no_new_dataset | 0.94625 |
2503.05541 | Juan Miguel Valverde | Juan Miguel Valverde, Maja {\O}stergaard, Adrian Rodriguez-Palomo,
Peter Alling Strange Vibe, Nina K{\o}lln Wittig, Henrik Birkedal, Anders
Bjorholm Dahl | Disconnect to Connect: A Data Augmentation Method for Improving Topology
Accuracy in Image Segmentation | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Accurate segmentation of thin, tubular structures (e.g., blood vessels) is
challenging for deep neural networks. These networks classify individual
pixels, and even minor misclassifications can break the thin connections within
these structures. Existing methods for improving topology accuracy, such as
topology loss functions, rely on very precise, topologically-accurate training
labels, which are difficult to obtain. This is because annotating images,
especially 3D images, is extremely laborious and time-consuming. Low image
resolution and contrast further complicates the annotation by causing tubular
structures to appear disconnected. We present CoLeTra, a data augmentation
strategy that integrates to the models the prior knowledge that structures that
appear broken are actually connected. This is achieved by creating images with
the appearance of disconnected structures while maintaining the original
labels. Our extensive experiments, involving different architectures, loss
functions, and datasets, demonstrate that CoLeTra leads to segmentations
topologically more accurate while often improving the Dice coefficient and
Hausdorff distance. CoLeTra's hyper-parameters are intuitive to tune, and our
sensitivity analysis shows that CoLeTra is robust to changes in these
hyper-parameters. We also release a dataset specifically suited for image
segmentation methods with a focus on topology accuracy. CoLetra's code can be
found at https://github.com/jmlipman/CoLeTra.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 16:11:55 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Valverde",
"Juan Miguel",
""
],
[
"Østergaard",
"Maja",
""
],
[
"Rodriguez-Palomo",
"Adrian",
""
],
[
"Vibe",
"Peter Alling Strange",
""
],
[
"Wittig",
"Nina Kølln",
""
],
[
"Birkedal",
"Henrik",
""
],
[
"Dahl",
"Anders Bjorholm",
""
]
]
| TITLE: Disconnect to Connect: A Data Augmentation Method for Improving Topology
Accuracy in Image Segmentation
ABSTRACT: Accurate segmentation of thin, tubular structures (e.g., blood vessels) is
challenging for deep neural networks. These networks classify individual
pixels, and even minor misclassifications can break the thin connections within
these structures. Existing methods for improving topology accuracy, such as
topology loss functions, rely on very precise, topologically-accurate training
labels, which are difficult to obtain. This is because annotating images,
especially 3D images, is extremely laborious and time-consuming. Low image
resolution and contrast further complicates the annotation by causing tubular
structures to appear disconnected. We present CoLeTra, a data augmentation
strategy that integrates to the models the prior knowledge that structures that
appear broken are actually connected. This is achieved by creating images with
the appearance of disconnected structures while maintaining the original
labels. Our extensive experiments, involving different architectures, loss
functions, and datasets, demonstrate that CoLeTra leads to segmentations
topologically more accurate while often improving the Dice coefficient and
Hausdorff distance. CoLeTra's hyper-parameters are intuitive to tune, and our
sensitivity analysis shows that CoLeTra is robust to changes in these
hyper-parameters. We also release a dataset specifically suited for image
segmentation methods with a focus on topology accuracy. CoLetra's code can be
found at https://github.com/jmlipman/CoLeTra.
| new_dataset | 0.960768 |
2503.05549 | Junpeng Jing | Junpeng Jing, Weixun Luo, Ye Mao, Krystian Mikolajczyk | Stereo Any Video: Temporally Consistent Stereo Matching | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper introduces Stereo Any Video, a powerful framework for video stereo
matching. It can estimate spatially accurate and temporally consistent
disparities without relying on auxiliary information such as camera poses or
optical flow. The strong capability is driven by rich priors from monocular
video depth models, which are integrated with convolutional features to produce
stable representations. To further enhance performance, key architectural
innovations are introduced: all-to-all-pairs correlation, which constructs
smooth and robust matching cost volumes, and temporal convex upsampling, which
improves temporal coherence. These components collectively ensure robustness,
accuracy, and temporal consistency, setting a new standard in video stereo
matching. Extensive experiments demonstrate that our method achieves
state-of-the-art performance across multiple datasets both qualitatively and
quantitatively in zero-shot settings, as well as strong generalization to
real-world indoor and outdoor scenarios.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 16:20:36 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Jing",
"Junpeng",
""
],
[
"Luo",
"Weixun",
""
],
[
"Mao",
"Ye",
""
],
[
"Mikolajczyk",
"Krystian",
""
]
]
| TITLE: Stereo Any Video: Temporally Consistent Stereo Matching
ABSTRACT: This paper introduces Stereo Any Video, a powerful framework for video stereo
matching. It can estimate spatially accurate and temporally consistent
disparities without relying on auxiliary information such as camera poses or
optical flow. The strong capability is driven by rich priors from monocular
video depth models, which are integrated with convolutional features to produce
stable representations. To further enhance performance, key architectural
innovations are introduced: all-to-all-pairs correlation, which constructs
smooth and robust matching cost volumes, and temporal convex upsampling, which
improves temporal coherence. These components collectively ensure robustness,
accuracy, and temporal consistency, setting a new standard in video stereo
matching. Extensive experiments demonstrate that our method achieves
state-of-the-art performance across multiple datasets both qualitatively and
quantitatively in zero-shot settings, as well as strong generalization to
real-world indoor and outdoor scenarios.
| no_new_dataset | 0.949902 |
2503.05568 | Xiaobei Zhao | Xiaobei Zhao (1), Xiangrong Zeng (1), Yihang Ma (1), Pengjin Tang (1),
Xiang Li (1) ((1) China Agricultural University) | TomatoScanner: phenotyping tomato fruit based on only RGB image | 12 pages, 37 figures. Codes and datasets are open-sourced in
https://github.com/AlexTraveling/TomatoScanner | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In tomato greenhouse, phenotypic measurement is meaningful for researchers
and farmers to monitor crop growth, thereby precisely control environmental
conditions in time, leading to better quality and higher yield. Traditional
phenotyping mainly relies on manual measurement, which is accurate but
inefficient, more importantly, endangering the health and safety of people.
Several studies have explored computer vision-based methods to replace manual
phenotyping. However, the 2D-based need extra calibration, or cause destruction
to fruit, or can only measure limited and meaningless traits. The 3D-based need
extra depth camera, which is expensive and unacceptable for most farmers. In
this paper, we propose a non-contact tomato fruit phenotyping method, titled
TomatoScanner, where RGB image is all you need for input. First, pixel feature
is extracted by instance segmentation of our proposed EdgeYOLO with
preprocessing of individual separation and pose correction. Second, depth
feature is extracted by depth estimation of Depth Pro. Third, pixel and depth
feature are fused to output phenotype results in reality. We establish
self-built Tomato Phenotype Dataset to test TomatoScanner, which achieves
excellent phenotyping on width, height, vertical area and volume, with median
relative error of 5.63%, 7.03%, -0.64% and 37.06%, respectively. We propose and
add three innovative modules - EdgeAttention, EdgeLoss and EdgeBoost - into
EdgeYOLO, to enhance the segmentation accuracy on edge portion. Precision and
mean Edge Error greatly improve from 0.943 and 5.641% to 0.986 and 2.963%,
respectively. Meanwhile, EdgeYOLO keeps lightweight and efficient, with 48.7 M
weights size and 76.34 FPS. Codes and datasets:
https://github.com/AlexTraveling/TomatoScanner.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 16:47:48 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Zhao",
"Xiaobei",
"",
"China Agricultural University"
],
[
"Zeng",
"Xiangrong",
"",
"China Agricultural University"
],
[
"Ma",
"Yihang",
"",
"China Agricultural University"
],
[
"Tang",
"Pengjin",
"",
"China Agricultural University"
],
[
"Li",
"Xiang",
"",
"China Agricultural University"
]
]
| TITLE: TomatoScanner: phenotyping tomato fruit based on only RGB image
ABSTRACT: In tomato greenhouse, phenotypic measurement is meaningful for researchers
and farmers to monitor crop growth, thereby precisely control environmental
conditions in time, leading to better quality and higher yield. Traditional
phenotyping mainly relies on manual measurement, which is accurate but
inefficient, more importantly, endangering the health and safety of people.
Several studies have explored computer vision-based methods to replace manual
phenotyping. However, the 2D-based need extra calibration, or cause destruction
to fruit, or can only measure limited and meaningless traits. The 3D-based need
extra depth camera, which is expensive and unacceptable for most farmers. In
this paper, we propose a non-contact tomato fruit phenotyping method, titled
TomatoScanner, where RGB image is all you need for input. First, pixel feature
is extracted by instance segmentation of our proposed EdgeYOLO with
preprocessing of individual separation and pose correction. Second, depth
feature is extracted by depth estimation of Depth Pro. Third, pixel and depth
feature are fused to output phenotype results in reality. We establish
self-built Tomato Phenotype Dataset to test TomatoScanner, which achieves
excellent phenotyping on width, height, vertical area and volume, with median
relative error of 5.63%, 7.03%, -0.64% and 37.06%, respectively. We propose and
add three innovative modules - EdgeAttention, EdgeLoss and EdgeBoost - into
EdgeYOLO, to enhance the segmentation accuracy on edge portion. Precision and
mean Edge Error greatly improve from 0.943 and 5.641% to 0.986 and 2.963%,
respectively. Meanwhile, EdgeYOLO keeps lightweight and efficient, with 48.7 M
weights size and 76.34 FPS. Codes and datasets:
https://github.com/AlexTraveling/TomatoScanner.
| no_new_dataset | 0.948058 |
2503.05578 | Jian Liu | Jian Liu, Wei Sun, Kai Zeng, Jin Zheng, Hui Yang, Lin Wang, Hossein
Rahmani, Ajmal Mian | Novel Object 6D Pose Estimation with a Single Reference View | 17 pages, 12 figures (including supplementary material) | null | null | null | cs.CV cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Existing novel object 6D pose estimation methods typically rely on CAD models
or dense reference views, which are both difficult to acquire. Using only a
single reference view is more scalable, but challenging due to large pose
discrepancies and limited geometric and spatial information. To address these
issues, we propose a Single-Reference-based novel object 6D (SinRef-6D) pose
estimation method. Our key idea is to iteratively establish point-wise
alignment in the camera coordinate system based on state space models (SSMs).
Specifically, iterative camera-space point-wise alignment can effectively
handle large pose discrepancies, while our proposed RGB and Points SSMs can
capture long-range dependencies and spatial information from a single view,
offering linear complexity and superior spatial modeling capability. Once
pre-trained on synthetic data, SinRef-6D can estimate the 6D pose of a novel
object using only a single reference view, without requiring retraining or a
CAD model. Extensive experiments on six popular datasets and real-world robotic
scenes demonstrate that we achieve on-par performance with CAD-based and dense
reference view-based methods, despite operating in the more challenging single
reference setting. Code will be released at
https://github.com/CNJianLiu/SinRef-6D.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 17:00:41 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Liu",
"Jian",
""
],
[
"Sun",
"Wei",
""
],
[
"Zeng",
"Kai",
""
],
[
"Zheng",
"Jin",
""
],
[
"Yang",
"Hui",
""
],
[
"Wang",
"Lin",
""
],
[
"Rahmani",
"Hossein",
""
],
[
"Mian",
"Ajmal",
""
]
]
| TITLE: Novel Object 6D Pose Estimation with a Single Reference View
ABSTRACT: Existing novel object 6D pose estimation methods typically rely on CAD models
or dense reference views, which are both difficult to acquire. Using only a
single reference view is more scalable, but challenging due to large pose
discrepancies and limited geometric and spatial information. To address these
issues, we propose a Single-Reference-based novel object 6D (SinRef-6D) pose
estimation method. Our key idea is to iteratively establish point-wise
alignment in the camera coordinate system based on state space models (SSMs).
Specifically, iterative camera-space point-wise alignment can effectively
handle large pose discrepancies, while our proposed RGB and Points SSMs can
capture long-range dependencies and spatial information from a single view,
offering linear complexity and superior spatial modeling capability. Once
pre-trained on synthetic data, SinRef-6D can estimate the 6D pose of a novel
object using only a single reference view, without requiring retraining or a
CAD model. Extensive experiments on six popular datasets and real-world robotic
scenes demonstrate that we achieve on-par performance with CAD-based and dense
reference view-based methods, despite operating in the more challenging single
reference setting. Code will be released at
https://github.com/CNJianLiu/SinRef-6D.
| no_new_dataset | 0.948917 |
2503.05582 | Yang Mu | Yang Mu, Muhammad Shahzad, Xiao Xiang Zhu | MPTSNet: Integrating Multiscale Periodic Local Patterns and Global
Dependencies for Multivariate Time Series Classification | Accepted by AAAI2025 | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multivariate Time Series Classification (MTSC) is crucial in extensive
practical applications, such as environmental monitoring, medical EEG analysis,
and action recognition. Real-world time series datasets typically exhibit
complex dynamics. To capture this complexity, RNN-based, CNN-based,
Transformer-based, and hybrid models have been proposed. Unfortunately, current
deep learning-based methods often neglect the simultaneous construction of
local features and global dependencies at different time scales, lacking
sufficient feature extraction capabilities to achieve satisfactory
classification accuracy. To address these challenges, we propose a novel
Multiscale Periodic Time Series Network (MPTSNet), which integrates multiscale
local patterns and global correlations to fully exploit the inherent
information in time series. Recognizing the multi-periodicity and complex
variable correlations in time series, we use the Fourier transform to extract
primary periods, enabling us to decompose data into multiscale periodic
segments. Leveraging the inherent strengths of CNN and attention mechanism, we
introduce the PeriodicBlock, which adaptively captures local patterns and
global dependencies while offering enhanced interpretability through attention
integration across different periodic scales. The experiments on UEA benchmark
datasets demonstrate that the proposed MPTSNet outperforms 21 existing advanced
baselines in the MTSC tasks.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 17:07:51 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Mu",
"Yang",
""
],
[
"Shahzad",
"Muhammad",
""
],
[
"Zhu",
"Xiao Xiang",
""
]
]
| TITLE: MPTSNet: Integrating Multiscale Periodic Local Patterns and Global
Dependencies for Multivariate Time Series Classification
ABSTRACT: Multivariate Time Series Classification (MTSC) is crucial in extensive
practical applications, such as environmental monitoring, medical EEG analysis,
and action recognition. Real-world time series datasets typically exhibit
complex dynamics. To capture this complexity, RNN-based, CNN-based,
Transformer-based, and hybrid models have been proposed. Unfortunately, current
deep learning-based methods often neglect the simultaneous construction of
local features and global dependencies at different time scales, lacking
sufficient feature extraction capabilities to achieve satisfactory
classification accuracy. To address these challenges, we propose a novel
Multiscale Periodic Time Series Network (MPTSNet), which integrates multiscale
local patterns and global correlations to fully exploit the inherent
information in time series. Recognizing the multi-periodicity and complex
variable correlations in time series, we use the Fourier transform to extract
primary periods, enabling us to decompose data into multiscale periodic
segments. Leveraging the inherent strengths of CNN and attention mechanism, we
introduce the PeriodicBlock, which adaptively captures local patterns and
global dependencies while offering enhanced interpretability through attention
integration across different periodic scales. The experiments on UEA benchmark
datasets demonstrate that the proposed MPTSNet outperforms 21 existing advanced
baselines in the MTSC tasks.
| no_new_dataset | 0.949201 |
2503.05587 | Shiping Yang | Shiping Yang, Jie Wu, Wenbiao Ding, Ning Wu, Shining Liang, Ming Gong,
Hengyuan Zhang, Dongmei Zhang | Quantifying the Robustness of Retrieval-Augmented Language Models
Against Spurious Features in Grounding Data | null | null | null | null | cs.CL cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Robustness has become a critical attribute for the deployment of RAG systems
in real-world applications. Existing research focuses on robustness to explicit
noise (e.g., document semantics) but overlooks spurious features (a.k.a.
implicit noise). While previous works have explored spurious features in LLMs,
they are limited to specific features (e.g., formats) and narrow scenarios
(e.g., ICL). In this work, we statistically confirm the presence of spurious
features in the RAG paradigm, a robustness problem caused by the sensitivity of
LLMs to semantic-agnostic features. Moreover, we provide a comprehensive
taxonomy of spurious features and empirically quantify their impact through
controlled experiments. Further analysis reveals that not all spurious features
are harmful and they can even be beneficial sometimes. Extensive evaluation
results across multiple LLMs suggest that spurious features are a widespread
and challenging problem in the field of RAG. The code and dataset will be
released to facilitate future research. We release all codes and data at:
$\\\href{https://github.com/maybenotime/RAG-SpuriousFeatures}{https://github.com/maybenotime/RAG-SpuriousFeatures}$.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 17:11:34 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Yang",
"Shiping",
""
],
[
"Wu",
"Jie",
""
],
[
"Ding",
"Wenbiao",
""
],
[
"Wu",
"Ning",
""
],
[
"Liang",
"Shining",
""
],
[
"Gong",
"Ming",
""
],
[
"Zhang",
"Hengyuan",
""
],
[
"Zhang",
"Dongmei",
""
]
]
| TITLE: Quantifying the Robustness of Retrieval-Augmented Language Models
Against Spurious Features in Grounding Data
ABSTRACT: Robustness has become a critical attribute for the deployment of RAG systems
in real-world applications. Existing research focuses on robustness to explicit
noise (e.g., document semantics) but overlooks spurious features (a.k.a.
implicit noise). While previous works have explored spurious features in LLMs,
they are limited to specific features (e.g., formats) and narrow scenarios
(e.g., ICL). In this work, we statistically confirm the presence of spurious
features in the RAG paradigm, a robustness problem caused by the sensitivity of
LLMs to semantic-agnostic features. Moreover, we provide a comprehensive
taxonomy of spurious features and empirically quantify their impact through
controlled experiments. Further analysis reveals that not all spurious features
are harmful and they can even be beneficial sometimes. Extensive evaluation
results across multiple LLMs suggest that spurious features are a widespread
and challenging problem in the field of RAG. The code and dataset will be
released to facilitate future research. We release all codes and data at:
$\\\href{https://github.com/maybenotime/RAG-SpuriousFeatures}{https://github.com/maybenotime/RAG-SpuriousFeatures}$.
| no_new_dataset | 0.917525 |
2503.05595 | Zheng Li | Zheng Li, Liangbin Xie, Jiantao Zhou, Xintao Wang, Haiwei Wu, Jinyu
Tian | Anti-Diffusion: Preventing Abuse of Modifications of Diffusion-Based
Models | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Although diffusion-based techniques have shown remarkable success in image
generation and editing tasks, their abuse can lead to severe negative social
impacts. Recently, some works have been proposed to provide defense against the
abuse of diffusion-based methods. However, their protection may be limited in
specific scenarios by manually defined prompts or the stable diffusion (SD)
version. Furthermore, these methods solely focus on tuning methods, overlooking
editing methods that could also pose a significant threat. In this work, we
propose Anti-Diffusion, a privacy protection system designed for general
diffusion-based methods, applicable to both tuning and editing techniques. To
mitigate the limitations of manually defined prompts on defense performance, we
introduce the prompt tuning (PT) strategy that enables precise expression of
original images. To provide defense against both tuning and editing methods, we
propose the semantic disturbance loss (SDL) to disrupt the semantic information
of protected images. Given the limited research on the defense against editing
methods, we develop a dataset named Defense-Edit to assess the defense
performance of various methods. Experiments demonstrate that our Anti-Diffusion
achieves superior defense performance across a wide range of diffusion-based
techniques in different scenarios.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 17:23:52 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Li",
"Zheng",
""
],
[
"Xie",
"Liangbin",
""
],
[
"Zhou",
"Jiantao",
""
],
[
"Wang",
"Xintao",
""
],
[
"Wu",
"Haiwei",
""
],
[
"Tian",
"Jinyu",
""
]
]
| TITLE: Anti-Diffusion: Preventing Abuse of Modifications of Diffusion-Based
Models
ABSTRACT: Although diffusion-based techniques have shown remarkable success in image
generation and editing tasks, their abuse can lead to severe negative social
impacts. Recently, some works have been proposed to provide defense against the
abuse of diffusion-based methods. However, their protection may be limited in
specific scenarios by manually defined prompts or the stable diffusion (SD)
version. Furthermore, these methods solely focus on tuning methods, overlooking
editing methods that could also pose a significant threat. In this work, we
propose Anti-Diffusion, a privacy protection system designed for general
diffusion-based methods, applicable to both tuning and editing techniques. To
mitigate the limitations of manually defined prompts on defense performance, we
introduce the prompt tuning (PT) strategy that enables precise expression of
original images. To provide defense against both tuning and editing methods, we
propose the semantic disturbance loss (SDL) to disrupt the semantic information
of protected images. Given the limited research on the defense against editing
methods, we develop a dataset named Defense-Edit to assess the defense
performance of various methods. Experiments demonstrate that our Anti-Diffusion
achieves superior defense performance across a wide range of diffusion-based
techniques in different scenarios.
| new_dataset | 0.959687 |
2503.05602 | Jan Schnabel | Roberto Fl\'orez Ablan, Marco Roth, and Jan Schnabel | On the similarity of bandwidth-tuned quantum kernels and classical
kernels | 9 main pages with 5 figures, and 9 appendix pages with 12 figures | null | null | null | quant-ph cs.LG | http://creativecommons.org/licenses/by/4.0/ | Quantum kernels (QK) are widely used in quantum machine learning
applications; yet, their potential to surpass classical machine learning
methods on classical datasets remains uncertain. This limitation can be
attributed to the exponential concentration phenomenon, which can impair both
trainability and generalization. A common strategy to alleviate this is
bandwidth tuning, which involves rescaling data points in the quantum model to
improve generalization. In this work, we numerically demonstrate that optimal
bandwidth tuning results in QKs that closely resemble radial basis function
(RBF) kernels, leading to a lack of quantum advantage over classical methods.
Moreover, we reveal that the size of optimal bandwidth tuning parameters
further simplifies QKs, causing them to behave like polynomial kernels,
corresponding to a low-order Taylor approximation of a RBF kernel. We
thoroughly investigate this for fidelity quantum kernels and projected quantum
kernels using various data encoding circuits across several classification
datasets. We provide numerical evidence and derive a simple analytical model
that elucidates how bandwidth tuning influences key quantities in
classification tasks. Overall, our findings shed light on the mechanisms that
render QK methods classically simulatable.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 17:28:02 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Ablan",
"Roberto Flórez",
""
],
[
"Roth",
"Marco",
""
],
[
"Schnabel",
"Jan",
""
]
]
| TITLE: On the similarity of bandwidth-tuned quantum kernels and classical
kernels
ABSTRACT: Quantum kernels (QK) are widely used in quantum machine learning
applications; yet, their potential to surpass classical machine learning
methods on classical datasets remains uncertain. This limitation can be
attributed to the exponential concentration phenomenon, which can impair both
trainability and generalization. A common strategy to alleviate this is
bandwidth tuning, which involves rescaling data points in the quantum model to
improve generalization. In this work, we numerically demonstrate that optimal
bandwidth tuning results in QKs that closely resemble radial basis function
(RBF) kernels, leading to a lack of quantum advantage over classical methods.
Moreover, we reveal that the size of optimal bandwidth tuning parameters
further simplifies QKs, causing them to behave like polynomial kernels,
corresponding to a low-order Taylor approximation of a RBF kernel. We
thoroughly investigate this for fidelity quantum kernels and projected quantum
kernels using various data encoding circuits across several classification
datasets. We provide numerical evidence and derive a simple analytical model
that elucidates how bandwidth tuning influences key quantities in
classification tasks. Overall, our findings shed light on the mechanisms that
render QK methods classically simulatable.
| no_new_dataset | 0.947769 |
2503.05604 | Ahmed Alagha | Hanae Elmekki, Ahmed Alagha, Hani Sami, Amanda Spilkin, Antonela
Mariel Zanuttini, Ehsan Zakeri, Jamal Bentahar, Lyes Kadem, Wen-Fang Xie,
Philippe Pibarot, Rabeb Mizouni, Hadi Otrok, Shakti Singh, Azzam Mourad | CACTUS: An Open Dataset and Framework for Automated Cardiac Assessment
and Classification of Ultrasound Images Using Deep Transfer Learning | null | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Cardiac ultrasound (US) scanning is a commonly used techniques in cardiology
to diagnose the health of the heart and its proper functioning. Therefore, it
is necessary to consider ways to automate these tasks and assist medical
professionals in classifying and assessing cardiac US images. Machine learning
(ML) techniques are regarded as a prominent solution due to their success in
numerous applications aimed at enhancing the medical field, including
addressing the shortage of echography technicians. However, the limited
availability of medical data presents a significant barrier to applying ML in
cardiology, particularly regarding US images of the heart. This paper addresses
this challenge by introducing the first open graded dataset for Cardiac
Assessment and ClassificaTion of UltraSound (CACTUS), which is available
online. This dataset contains images obtained from scanning a CAE Blue Phantom
and representing various heart views and different quality levels, exceeding
the conventional cardiac views typically found in the literature. Additionally,
the paper introduces a Deep Learning (DL) framework consisting of two main
components. The first component classifies cardiac US images based on the heart
view using a Convolutional Neural Network (CNN). The second component uses
Transfer Learning (TL) to fine-tune the knowledge from the first component and
create a model for grading and assessing cardiac images. The framework
demonstrates high performance in both classification and grading, achieving up
to 99.43% accuracy and as low as 0.3067 error, respectively. To showcase its
robustness, the framework is further fine-tuned using new images representing
additional cardiac views and compared to several other state-of-the-art
architectures. The framework's outcomes and performance in handling real-time
scans were also assessed using a questionnaire answered by cardiac experts.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 17:29:04 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Elmekki",
"Hanae",
""
],
[
"Alagha",
"Ahmed",
""
],
[
"Sami",
"Hani",
""
],
[
"Spilkin",
"Amanda",
""
],
[
"Zanuttini",
"Antonela Mariel",
""
],
[
"Zakeri",
"Ehsan",
""
],
[
"Bentahar",
"Jamal",
""
],
[
"Kadem",
"Lyes",
""
],
[
"Xie",
"Wen-Fang",
""
],
[
"Pibarot",
"Philippe",
""
],
[
"Mizouni",
"Rabeb",
""
],
[
"Otrok",
"Hadi",
""
],
[
"Singh",
"Shakti",
""
],
[
"Mourad",
"Azzam",
""
]
]
| TITLE: CACTUS: An Open Dataset and Framework for Automated Cardiac Assessment
and Classification of Ultrasound Images Using Deep Transfer Learning
ABSTRACT: Cardiac ultrasound (US) scanning is a commonly used techniques in cardiology
to diagnose the health of the heart and its proper functioning. Therefore, it
is necessary to consider ways to automate these tasks and assist medical
professionals in classifying and assessing cardiac US images. Machine learning
(ML) techniques are regarded as a prominent solution due to their success in
numerous applications aimed at enhancing the medical field, including
addressing the shortage of echography technicians. However, the limited
availability of medical data presents a significant barrier to applying ML in
cardiology, particularly regarding US images of the heart. This paper addresses
this challenge by introducing the first open graded dataset for Cardiac
Assessment and ClassificaTion of UltraSound (CACTUS), which is available
online. This dataset contains images obtained from scanning a CAE Blue Phantom
and representing various heart views and different quality levels, exceeding
the conventional cardiac views typically found in the literature. Additionally,
the paper introduces a Deep Learning (DL) framework consisting of two main
components. The first component classifies cardiac US images based on the heart
view using a Convolutional Neural Network (CNN). The second component uses
Transfer Learning (TL) to fine-tune the knowledge from the first component and
create a model for grading and assessing cardiac images. The framework
demonstrates high performance in both classification and grading, achieving up
to 99.43% accuracy and as low as 0.3067 error, respectively. To showcase its
robustness, the framework is further fine-tuned using new images representing
additional cardiac views and compared to several other state-of-the-art
architectures. The framework's outcomes and performance in handling real-time
scans were also assessed using a questionnaire answered by cardiac experts.
| new_dataset | 0.910107 |
2503.05605 | Silvia Garc\'ia-M\'endez | Francisco de Arriba-P\'erez, Silvia Garc\'ia-M\'endez, F\'atima Leal,
Benedita Malheiro, Juan C Burguillo | Identification and explanation of disinformation in wiki data streams | (2025) Integrated Computer-Aided Engineering | null | 10.1177/10692509241306580 | null | cs.IR cs.CY cs.SI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Social media platforms, increasingly used as news sources for varied data
analytics, have transformed how information is generated and disseminated.
However, the unverified nature of this content raises concerns about
trustworthiness and accuracy, potentially negatively impacting readers'
critical judgment due to disinformation. This work aims to contribute to the
automatic data quality validation field, addressing the rapid growth of online
content on wiki pages. Our scalable solution includes stream-based data
processing with feature engineering, feature analysis and selection,
stream-based classification, and real-time explanation of prediction outcomes.
The explainability dashboard is designed for the general public, who may need
more specialized knowledge to interpret the model's prediction. Experimental
results on two datasets attain approximately 90 % values across all evaluation
metrics, demonstrating robust and competitive performance compared to works in
the literature. In summary, the system assists editors by reducing their effort
and time in detecting disinformation.
| [
{
"version": "v1",
"created": "Mon, 3 Feb 2025 08:34:39 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"de Arriba-Pérez",
"Francisco",
""
],
[
"García-Méndez",
"Silvia",
""
],
[
"Leal",
"Fátima",
""
],
[
"Malheiro",
"Benedita",
""
],
[
"Burguillo",
"Juan C",
""
]
]
| TITLE: Identification and explanation of disinformation in wiki data streams
ABSTRACT: Social media platforms, increasingly used as news sources for varied data
analytics, have transformed how information is generated and disseminated.
However, the unverified nature of this content raises concerns about
trustworthiness and accuracy, potentially negatively impacting readers'
critical judgment due to disinformation. This work aims to contribute to the
automatic data quality validation field, addressing the rapid growth of online
content on wiki pages. Our scalable solution includes stream-based data
processing with feature engineering, feature analysis and selection,
stream-based classification, and real-time explanation of prediction outcomes.
The explainability dashboard is designed for the general public, who may need
more specialized knowledge to interpret the model's prediction. Experimental
results on two datasets attain approximately 90 % values across all evaluation
metrics, demonstrating robust and competitive performance compared to works in
the literature. In summary, the system assists editors by reducing their effort
and time in detecting disinformation.
| no_new_dataset | 0.952353 |
2503.05609 | Charvi Rastogi | Pushkar Mishra, Charvi Rastogi, Stephen R. Pfohl, Alicia Parrish, Roma
Patel, Mark Diaz, Ding Wang, Michela Paganini, Vinodkumar Prabhakaran, Lora
Aroyo, Verena Rieser | Nuanced Safety for Generative AI: How Demographics Shape Responsiveness
to Severity | null | null | null | null | cs.CY cs.HC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Ensuring safety of Generative AI requires a nuanced understanding of
pluralistic viewpoints. In this paper, we introduce a novel data-driven
approach for calibrating granular ratings in pluralistic datasets.
Specifically, we address the challenge of interpreting responses of a diverse
population to safety expressed via ordinal scales (e.g., Likert scale). We
distill non-parametric responsiveness metrics that quantify the consistency of
raters in scoring the varying levels of the severity of safety violations.
Using safety evaluation of AI-generated content as a case study, we investigate
how raters from different demographic groups (age, gender, ethnicity) use an
ordinal scale to express their perception of the severity of violations in a
pluralistic safety dataset. We apply our metrics across violation types,
demonstrating their utility in extracting nuanced insights that are crucial for
developing reliable AI systems in a multi-cultural contexts. We show that our
approach offers improved capabilities for prioritizing safety concerns by
capturing nuanced viewpoints across different demographic groups, hence
improving the reliability of pluralistic data collection and in turn
contributing to more robust AI evaluations.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 17:32:31 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Mishra",
"Pushkar",
""
],
[
"Rastogi",
"Charvi",
""
],
[
"Pfohl",
"Stephen R.",
""
],
[
"Parrish",
"Alicia",
""
],
[
"Patel",
"Roma",
""
],
[
"Diaz",
"Mark",
""
],
[
"Wang",
"Ding",
""
],
[
"Paganini",
"Michela",
""
],
[
"Prabhakaran",
"Vinodkumar",
""
],
[
"Aroyo",
"Lora",
""
],
[
"Rieser",
"Verena",
""
]
]
| TITLE: Nuanced Safety for Generative AI: How Demographics Shape Responsiveness
to Severity
ABSTRACT: Ensuring safety of Generative AI requires a nuanced understanding of
pluralistic viewpoints. In this paper, we introduce a novel data-driven
approach for calibrating granular ratings in pluralistic datasets.
Specifically, we address the challenge of interpreting responses of a diverse
population to safety expressed via ordinal scales (e.g., Likert scale). We
distill non-parametric responsiveness metrics that quantify the consistency of
raters in scoring the varying levels of the severity of safety violations.
Using safety evaluation of AI-generated content as a case study, we investigate
how raters from different demographic groups (age, gender, ethnicity) use an
ordinal scale to express their perception of the severity of violations in a
pluralistic safety dataset. We apply our metrics across violation types,
demonstrating their utility in extracting nuanced insights that are crucial for
developing reliable AI systems in a multi-cultural contexts. We show that our
approach offers improved capabilities for prioritizing safety concerns by
capturing nuanced viewpoints across different demographic groups, hence
improving the reliability of pluralistic data collection and in turn
contributing to more robust AI evaluations.
| no_new_dataset | 0.946101 |
2503.05618 | Luca Mossina | Luca Mossina and Corentin Friedrich | Conformal Prediction for Image Segmentation Using Morphological
Prediction Sets | null | null | null | null | cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Image segmentation is a challenging task influenced by multiple sources of
uncertainty, such as the data labeling process or the sampling of training
data. In this paper we focus on binary segmentation and address these
challenges using conformal prediction, a family of model- and data-agnostic
methods for uncertainty quantification that provide finite-sample theoretical
guarantees and applicable to any pretrained predictor. Our approach involves
computing nonconformity scores, a type of prediction residual, on held-out
calibration data not used during training. We use dilation, one of the
fundamental operations in mathematical morphology, to construct a margin added
to the borders of predicted segmentation masks. At inference, the predicted set
formed by the mask and its margin contains the ground-truth mask with high
probability, at a confidence level specified by the user. The size of the
margin serves as an indicator of predictive uncertainty for a given model and
dataset. We work in a regime of minimal information as we do not require any
feedback from the predictor: only the predicted masks are needed for computing
the prediction sets. Hence, our method is applicable to any segmentation model,
including those based on deep learning; we evaluate our approach on several
medical imaging applications.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 17:42:30 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Mossina",
"Luca",
""
],
[
"Friedrich",
"Corentin",
""
]
]
| TITLE: Conformal Prediction for Image Segmentation Using Morphological
Prediction Sets
ABSTRACT: Image segmentation is a challenging task influenced by multiple sources of
uncertainty, such as the data labeling process or the sampling of training
data. In this paper we focus on binary segmentation and address these
challenges using conformal prediction, a family of model- and data-agnostic
methods for uncertainty quantification that provide finite-sample theoretical
guarantees and applicable to any pretrained predictor. Our approach involves
computing nonconformity scores, a type of prediction residual, on held-out
calibration data not used during training. We use dilation, one of the
fundamental operations in mathematical morphology, to construct a margin added
to the borders of predicted segmentation masks. At inference, the predicted set
formed by the mask and its margin contains the ground-truth mask with high
probability, at a confidence level specified by the user. The size of the
margin serves as an indicator of predictive uncertainty for a given model and
dataset. We work in a regime of minimal information as we do not require any
feedback from the predictor: only the predicted masks are needed for computing
the prediction sets. Hence, our method is applicable to any segmentation model,
including those based on deep learning; we evaluate our approach on several
medical imaging applications.
| no_new_dataset | 0.950227 |
2503.05620 | Xuanqing Liu | Xuanqing Liu, Luyang Kong, Wei Niu, Afshin Khashei, Belinda Zeng,
Steve Johnson, Jon Jay, Davor Golac, Matt Pope | Learning LLM Preference over Intra-Dialogue Pairs: A Framework for
Utterance-level Understandings | 7 pages, 4 figures | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Large language models (LLMs) have demonstrated remarkable capabilities in
handling complex dialogue tasks without requiring use case-specific
fine-tuning. However, analyzing live dialogues in real-time necessitates
low-latency processing systems, making it impractical to deploy models with
billions of parameters due to latency constraints. As a result, practitioners
often prefer smaller models with millions of parameters, trained on
high-quality, human-annotated datasets. Yet, curating such datasets is both
time-consuming and costly. Consequently, there is a growing need to combine the
scalability of LLM-generated labels with the precision of human annotations,
enabling fine-tuned smaller models to achieve both higher speed and accuracy
comparable to larger models. In this paper, we introduce a simple yet effective
framework to address this challenge. Our approach is specifically designed for
per-utterance classification problems, which encompass tasks such as intent
detection, dialogue state tracking, and more. To mitigate the impact of
labeling errors from LLMs -- the primary source of inaccuracies in student
models -- we propose a noise-reduced preference learning loss. Experimental
results demonstrate that our method significantly improves accuracy across
utterance-level dialogue tasks, including sentiment detection (over $2\%$),
dialogue act classification (over $1.5\%$), etc.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 17:46:13 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Liu",
"Xuanqing",
""
],
[
"Kong",
"Luyang",
""
],
[
"Niu",
"Wei",
""
],
[
"Khashei",
"Afshin",
""
],
[
"Zeng",
"Belinda",
""
],
[
"Johnson",
"Steve",
""
],
[
"Jay",
"Jon",
""
],
[
"Golac",
"Davor",
""
],
[
"Pope",
"Matt",
""
]
]
| TITLE: Learning LLM Preference over Intra-Dialogue Pairs: A Framework for
Utterance-level Understandings
ABSTRACT: Large language models (LLMs) have demonstrated remarkable capabilities in
handling complex dialogue tasks without requiring use case-specific
fine-tuning. However, analyzing live dialogues in real-time necessitates
low-latency processing systems, making it impractical to deploy models with
billions of parameters due to latency constraints. As a result, practitioners
often prefer smaller models with millions of parameters, trained on
high-quality, human-annotated datasets. Yet, curating such datasets is both
time-consuming and costly. Consequently, there is a growing need to combine the
scalability of LLM-generated labels with the precision of human annotations,
enabling fine-tuned smaller models to achieve both higher speed and accuracy
comparable to larger models. In this paper, we introduce a simple yet effective
framework to address this challenge. Our approach is specifically designed for
per-utterance classification problems, which encompass tasks such as intent
detection, dialogue state tracking, and more. To mitigate the impact of
labeling errors from LLMs -- the primary source of inaccuracies in student
models -- we propose a noise-reduced preference learning loss. Experimental
results demonstrate that our method significantly improves accuracy across
utterance-level dialogue tasks, including sentiment detection (over $2\%$),
dialogue act classification (over $1.5\%$), etc.
| no_new_dataset | 0.950595 |
2503.05626 | Jingyu Xu | Jingyu Xu, Yang Wang | FMT:A Multimodal Pneumonia Detection Model Based on Stacking MOE
Framework | null | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Artificial intelligence has shown the potential to improve diagnostic
accuracy through medical image analysis for pneumonia diagnosis. However,
traditional multimodal approaches often fail to address real-world challenges
such as incomplete data and modality loss. In this study, a Flexible Multimodal
Transformer (FMT) was proposed, which uses ResNet-50 and BERT for joint
representation learning, followed by a dynamic masked attention strategy that
simulates clinical modality loss to improve robustness; finally, a sequential
mixture of experts (MOE) architecture was used to achieve multi-level decision
refinement. After evaluation on a small multimodal pneumonia dataset, FMT
achieved state-of-the-art performance with 94% accuracy, 95% recall, and 93% F1
score, outperforming single-modal baselines (ResNet: 89%; BERT: 79%) and the
medical benchmark CheXMed (90%), providing a scalable solution for multimodal
diagnosis of pneumonia in resource-constrained medical settings.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 17:52:12 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Xu",
"Jingyu",
""
],
[
"Wang",
"Yang",
""
]
]
| TITLE: FMT:A Multimodal Pneumonia Detection Model Based on Stacking MOE
Framework
ABSTRACT: Artificial intelligence has shown the potential to improve diagnostic
accuracy through medical image analysis for pneumonia diagnosis. However,
traditional multimodal approaches often fail to address real-world challenges
such as incomplete data and modality loss. In this study, a Flexible Multimodal
Transformer (FMT) was proposed, which uses ResNet-50 and BERT for joint
representation learning, followed by a dynamic masked attention strategy that
simulates clinical modality loss to improve robustness; finally, a sequential
mixture of experts (MOE) architecture was used to achieve multi-level decision
refinement. After evaluation on a small multimodal pneumonia dataset, FMT
achieved state-of-the-art performance with 94% accuracy, 95% recall, and 93% F1
score, outperforming single-modal baselines (ResNet: 89%; BERT: 79%) and the
medical benchmark CheXMed (90%), providing a scalable solution for multimodal
diagnosis of pneumonia in resource-constrained medical settings.
| no_new_dataset | 0.946151 |
2503.05629 | Samaneh Zolfaghari Dr. | Ali Samimi Fard, Mohammadreza Mashhadigholamali, Samaneh Zolfaghari,
Hajar Abedi, Mainak Chakraborty, Luigi Borz\`i, Masoud Daneshtalab, George
Shaker | Exploring FMCW Radars and Feature Maps for Activity Recognition: A
Benchmark Study | null | null | null | null | cs.ET cs.AI | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Human Activity Recognition has gained significant attention due to its
diverse applications, including ambient assisted living and remote sensing.
Wearable sensor-based solutions often suffer from user discomfort and
reliability issues, while video-based methods raise privacy concerns and
perform poorly in low-light conditions or long ranges. This study introduces a
Frequency-Modulated Continuous Wave radar-based framework for human activity
recognition, leveraging a 60 GHz radar and multi-dimensional feature maps.
Unlike conventional approaches that process feature maps as images, this study
feeds multi-dimensional feature maps -- Range-Doppler, Range-Azimuth, and
Range-Elevation -- as data vectors directly into the machine learning (SVM,
MLP) and deep learning (CNN, LSTM, ConvLSTM) models, preserving the spatial and
temporal structures of the data. These features were extracted from a novel
dataset with seven activity classes and validated using two different
validation approaches. The ConvLSTM model outperformed conventional machine
learning and deep learning models, achieving an accuracy of 90.51% and an
F1-score of 87.31% on cross-scene validation and an accuracy of 89.56% and an
F1-score of 87.15% on leave-one-person-out cross-validation. The results
highlight the approach's potential for scalable, non-intrusive, and
privacy-preserving activity monitoring in real-world scenarios.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 17:53:29 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Fard",
"Ali Samimi",
""
],
[
"Mashhadigholamali",
"Mohammadreza",
""
],
[
"Zolfaghari",
"Samaneh",
""
],
[
"Abedi",
"Hajar",
""
],
[
"Chakraborty",
"Mainak",
""
],
[
"Borzì",
"Luigi",
""
],
[
"Daneshtalab",
"Masoud",
""
],
[
"Shaker",
"George",
""
]
]
| TITLE: Exploring FMCW Radars and Feature Maps for Activity Recognition: A
Benchmark Study
ABSTRACT: Human Activity Recognition has gained significant attention due to its
diverse applications, including ambient assisted living and remote sensing.
Wearable sensor-based solutions often suffer from user discomfort and
reliability issues, while video-based methods raise privacy concerns and
perform poorly in low-light conditions or long ranges. This study introduces a
Frequency-Modulated Continuous Wave radar-based framework for human activity
recognition, leveraging a 60 GHz radar and multi-dimensional feature maps.
Unlike conventional approaches that process feature maps as images, this study
feeds multi-dimensional feature maps -- Range-Doppler, Range-Azimuth, and
Range-Elevation -- as data vectors directly into the machine learning (SVM,
MLP) and deep learning (CNN, LSTM, ConvLSTM) models, preserving the spatial and
temporal structures of the data. These features were extracted from a novel
dataset with seven activity classes and validated using two different
validation approaches. The ConvLSTM model outperformed conventional machine
learning and deep learning models, achieving an accuracy of 90.51% and an
F1-score of 87.31% on cross-scene validation and an accuracy of 89.56% and an
F1-score of 87.15% on leave-one-person-out cross-validation. The results
highlight the approach's potential for scalable, non-intrusive, and
privacy-preserving activity monitoring in real-world scenarios.
| no_new_dataset | 0.580471 |
2503.05630 | Yu Jiang | Tian Qiu, Ruiming Du, Nikolai Spine, Lailiang Cheng, Yu Jiang | Joint 3D Point Cloud Segmentation using Real-Sim Loop: From Panels to
Trees and Branches | Accepted by ICRA 2025 | null | null | null | cs.RO cs.CV q-bio.QM | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Modern orchards are planted in structured rows with distinct panel divisions
to improve management. Accurate and efficient joint segmentation of point cloud
from Panel to Tree and Branch (P2TB) is essential for robotic operations.
However, most current segmentation methods focus on single instance
segmentation and depend on a sequence of deep networks to perform joint tasks.
This strategy hinders the use of hierarchical information embedded in the data,
leading to both error accumulation and increased costs for annotation and
computation, which limits its scalability for real-world applications. In this
study, we proposed a novel approach that incorporated a Real2Sim L-TreeGen for
training data generation and a joint model (J-P2TB) designed for the P2TB task.
The J-P2TB model, trained on the generated simulation dataset, was used for
joint segmentation of real-world panel point clouds via zero-shot learning.
Compared to representative methods, our model outperformed them in most
segmentation metrics while using 40% fewer learnable parameters. This Sim2Real
result highlighted the efficacy of L-TreeGen in model training and the
performance of J-P2TB for joint segmentation, demonstrating its strong
accuracy, efficiency, and generalizability for real-world applications. These
improvements would not only greatly benefit the development of robots for
automated orchard operations but also advance digital twin technology.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 17:54:02 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Qiu",
"Tian",
""
],
[
"Du",
"Ruiming",
""
],
[
"Spine",
"Nikolai",
""
],
[
"Cheng",
"Lailiang",
""
],
[
"Jiang",
"Yu",
""
]
]
| TITLE: Joint 3D Point Cloud Segmentation using Real-Sim Loop: From Panels to
Trees and Branches
ABSTRACT: Modern orchards are planted in structured rows with distinct panel divisions
to improve management. Accurate and efficient joint segmentation of point cloud
from Panel to Tree and Branch (P2TB) is essential for robotic operations.
However, most current segmentation methods focus on single instance
segmentation and depend on a sequence of deep networks to perform joint tasks.
This strategy hinders the use of hierarchical information embedded in the data,
leading to both error accumulation and increased costs for annotation and
computation, which limits its scalability for real-world applications. In this
study, we proposed a novel approach that incorporated a Real2Sim L-TreeGen for
training data generation and a joint model (J-P2TB) designed for the P2TB task.
The J-P2TB model, trained on the generated simulation dataset, was used for
joint segmentation of real-world panel point clouds via zero-shot learning.
Compared to representative methods, our model outperformed them in most
segmentation metrics while using 40% fewer learnable parameters. This Sim2Real
result highlighted the efficacy of L-TreeGen in model training and the
performance of J-P2TB for joint segmentation, demonstrating its strong
accuracy, efficiency, and generalizability for real-world applications. These
improvements would not only greatly benefit the development of robots for
automated orchard operations but also advance digital twin technology.
| no_new_dataset | 0.946101 |
2503.05638 | Wenbo Hu | Mark YU, Wenbo Hu, Jinbo Xing, Ying Shan | TrajectoryCrafter: Redirecting Camera Trajectory for Monocular Videos
via Diffusion Models | Project webpage: https://trajectorycrafter.github.io/ | null | null | null | cs.CV cs.AI cs.GR | http://creativecommons.org/licenses/by/4.0/ | We present TrajectoryCrafter, a novel approach to redirect camera
trajectories for monocular videos. By disentangling deterministic view
transformations from stochastic content generation, our method achieves precise
control over user-specified camera trajectories. We propose a novel dual-stream
conditional video diffusion model that concurrently integrates point cloud
renders and source videos as conditions, ensuring accurate view transformations
and coherent 4D content generation. Instead of leveraging scarce multi-view
videos, we curate a hybrid training dataset combining web-scale monocular
videos with static multi-view datasets, by our innovative double-reprojection
strategy, significantly fostering robust generalization across diverse scenes.
Extensive evaluations on multi-view and large-scale monocular videos
demonstrate the superior performance of our method.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 17:57:53 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"YU",
"Mark",
""
],
[
"Hu",
"Wenbo",
""
],
[
"Xing",
"Jinbo",
""
],
[
"Shan",
"Ying",
""
]
]
| TITLE: TrajectoryCrafter: Redirecting Camera Trajectory for Monocular Videos
via Diffusion Models
ABSTRACT: We present TrajectoryCrafter, a novel approach to redirect camera
trajectories for monocular videos. By disentangling deterministic view
transformations from stochastic content generation, our method achieves precise
control over user-specified camera trajectories. We propose a novel dual-stream
conditional video diffusion model that concurrently integrates point cloud
renders and source videos as conditions, ensuring accurate view transformations
and coherent 4D content generation. Instead of leveraging scarce multi-view
videos, we curate a hybrid training dataset combining web-scale monocular
videos with static multi-view datasets, by our innovative double-reprojection
strategy, significantly fostering robust generalization across diverse scenes.
Extensive evaluations on multi-view and large-scale monocular videos
demonstrate the superior performance of our method.
| new_dataset | 0.856152 |
2503.05648 | Bharat Jayaprakash | Harish Panneer Selvam, Bharat Jayaprakash, Yan Li, Shashi Shekhar,
William F. Northrop | Physics-based machine learning framework for predicting NOx emissions
from compression ignition engines using on-board diagnostics data | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work presents a physics-based machine learning framework to predict and
analyze oxides of nitrogen (NOx) emissions from compression-ignition
engine-powered vehicles using on-board diagnostics (OBD) data as input.
Accurate NOx prediction from OBD datasets is difficult because NOx formation
inside an engine combustion chamber is governed by complex processes occurring
on timescales much shorter than the data collection rate. Thus, emissions
generally cannot be predicted accurately using simple empirically derived
physics models. Black box models like genetic algorithms or neural networks can
be more accurate, but have poor interpretability. The transparent model
presented in this paper has both high accuracy and can explain potential
sources of high emissions. The proposed framework consists of two major steps:
a physics-based NOx prediction model combined with a novel Divergent Window
Co-occurrence (DWC) Pattern detection algorithm to analyze operating conditions
that are not adequately addressed by the physics-based model. The proposed
framework is validated for generalizability with a second vehicle OBD dataset,
a sensitivity analysis is performed, and model predictions are compared with
that from a deep neural network. The results show that NOx emissions
predictions using the proposed model has around 55% better root mean square
error, and around 60% higher mean absolute error compared to the baseline NOx
prediction model from previously published work. The DWC Pattern Detection
Algorithm identified low engine power conditions to have high statistical
significance, indicating an operating regime where the model can be improved.
This work shows that the physics-based machine learning framework is a viable
method for predicting NOx emissions from engines that do not incorporate NOx
sensing.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 18:11:23 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Selvam",
"Harish Panneer",
""
],
[
"Jayaprakash",
"Bharat",
""
],
[
"Li",
"Yan",
""
],
[
"Shekhar",
"Shashi",
""
],
[
"Northrop",
"William F.",
""
]
]
| TITLE: Physics-based machine learning framework for predicting NOx emissions
from compression ignition engines using on-board diagnostics data
ABSTRACT: This work presents a physics-based machine learning framework to predict and
analyze oxides of nitrogen (NOx) emissions from compression-ignition
engine-powered vehicles using on-board diagnostics (OBD) data as input.
Accurate NOx prediction from OBD datasets is difficult because NOx formation
inside an engine combustion chamber is governed by complex processes occurring
on timescales much shorter than the data collection rate. Thus, emissions
generally cannot be predicted accurately using simple empirically derived
physics models. Black box models like genetic algorithms or neural networks can
be more accurate, but have poor interpretability. The transparent model
presented in this paper has both high accuracy and can explain potential
sources of high emissions. The proposed framework consists of two major steps:
a physics-based NOx prediction model combined with a novel Divergent Window
Co-occurrence (DWC) Pattern detection algorithm to analyze operating conditions
that are not adequately addressed by the physics-based model. The proposed
framework is validated for generalizability with a second vehicle OBD dataset,
a sensitivity analysis is performed, and model predictions are compared with
that from a deep neural network. The results show that NOx emissions
predictions using the proposed model has around 55% better root mean square
error, and around 60% higher mean absolute error compared to the baseline NOx
prediction model from previously published work. The DWC Pattern Detection
Algorithm identified low engine power conditions to have high statistical
significance, indicating an operating regime where the model can be improved.
This work shows that the physics-based machine learning framework is a viable
method for predicting NOx emissions from engines that do not incorporate NOx
sensing.
| no_new_dataset | 0.949153 |
2503.05657 | Yasser Khalil | Yasser H. Khalil, Leo Brunswic, Soufiane Lamghari, Xu Li, Mahdi
Beitollahi, Xi Chen | NoT: Federated Unlearning via Weight Negation | The 42nd IEEE/CVF Conference on Computer Vision and Pattern
Recognition, Nashville TN, US. 2025 | null | null | null | cs.LG cs.CV | http://creativecommons.org/licenses/by/4.0/ | Federated unlearning (FU) aims to remove a participant's data contributions
from a trained federated learning (FL) model, ensuring privacy and regulatory
compliance. Traditional FU methods often depend on auxiliary storage on either
the client or server side or require direct access to the data targeted for
removal-a dependency that may not be feasible if the data is no longer
available. To overcome these limitations, we propose NoT, a novel and efficient
FU algorithm based on weight negation (multiplying by -1), which circumvents
the need for additional storage and access to the target data. We argue that
effective and efficient unlearning can be achieved by perturbing model
parameters away from the set of optimal parameters, yet being well-positioned
for quick re-optimization. This technique, though seemingly contradictory, is
theoretically grounded: we prove that the weight negation perturbation
effectively disrupts inter-layer co-adaptation, inducing unlearning while
preserving an approximate optimality property, thereby enabling rapid recovery.
Experimental results across three datasets and three model architectures
demonstrate that NoT significantly outperforms existing baselines in unlearning
efficacy as well as in communication and computational efficiency.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 18:19:19 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Khalil",
"Yasser H.",
""
],
[
"Brunswic",
"Leo",
""
],
[
"Lamghari",
"Soufiane",
""
],
[
"Li",
"Xu",
""
],
[
"Beitollahi",
"Mahdi",
""
],
[
"Chen",
"Xi",
""
]
]
| TITLE: NoT: Federated Unlearning via Weight Negation
ABSTRACT: Federated unlearning (FU) aims to remove a participant's data contributions
from a trained federated learning (FL) model, ensuring privacy and regulatory
compliance. Traditional FU methods often depend on auxiliary storage on either
the client or server side or require direct access to the data targeted for
removal-a dependency that may not be feasible if the data is no longer
available. To overcome these limitations, we propose NoT, a novel and efficient
FU algorithm based on weight negation (multiplying by -1), which circumvents
the need for additional storage and access to the target data. We argue that
effective and efficient unlearning can be achieved by perturbing model
parameters away from the set of optimal parameters, yet being well-positioned
for quick re-optimization. This technique, though seemingly contradictory, is
theoretically grounded: we prove that the weight negation perturbation
effectively disrupts inter-layer co-adaptation, inducing unlearning while
preserving an approximate optimality property, thereby enabling rapid recovery.
Experimental results across three datasets and three model architectures
demonstrate that NoT significantly outperforms existing baselines in unlearning
efficacy as well as in communication and computational efficiency.
| no_new_dataset | 0.943504 |
2503.05665 | Zengqun Zhao | Zengqun Zhao, Ziquan Liu, Yu Cao, Shaogang Gong, Ioannis Patras | AIM-Fair: Advancing Algorithmic Fairness via Selectively Fine-Tuning
Biased Models with Contextual Synthetic Data | Accepted at CVPR 2025. Github:
https://github.com/zengqunzhao/AIM-Fair. Project page:
https://zengqunzhao.github.io/AIMFair | null | null | null | cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | Recent advances in generative models have sparked research on improving model
fairness with AI-generated data. However, existing methods often face
limitations in the diversity and quality of synthetic data, leading to
compromised fairness and overall model accuracy. Moreover, many approaches rely
on the availability of demographic group labels, which are often costly to
annotate. This paper proposes AIM-Fair, aiming to overcome these limitations
and harness the potential of cutting-edge generative models in promoting
algorithmic fairness. We investigate a fine-tuning paradigm starting from a
biased model initially trained on real-world data without demographic
annotations. This model is then fine-tuned using unbiased synthetic data
generated by a state-of-the-art diffusion model to improve its fairness. Two
key challenges are identified in this fine-tuning paradigm, 1) the low quality
of synthetic data, which can still happen even with advanced generative models,
and 2) the domain and bias gap between real and synthetic data. To address the
limitation of synthetic data quality, we propose Contextual Synthetic Data
Generation (CSDG) to generate data using a text-to-image diffusion model (T2I)
with prompts generated by a context-aware LLM, ensuring both data diversity and
control of bias in synthetic data. To resolve domain and bias shifts, we
introduce a novel selective fine-tuning scheme in which only model parameters
more sensitive to bias and less sensitive to domain shift are updated.
Experiments on CelebA and UTKFace datasets show that our AIM-Fair improves
model fairness while maintaining utility, outperforming both fully and
partially fine-tuned approaches to model fairness.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 18:26:48 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Zhao",
"Zengqun",
""
],
[
"Liu",
"Ziquan",
""
],
[
"Cao",
"Yu",
""
],
[
"Gong",
"Shaogang",
""
],
[
"Patras",
"Ioannis",
""
]
]
| TITLE: AIM-Fair: Advancing Algorithmic Fairness via Selectively Fine-Tuning
Biased Models with Contextual Synthetic Data
ABSTRACT: Recent advances in generative models have sparked research on improving model
fairness with AI-generated data. However, existing methods often face
limitations in the diversity and quality of synthetic data, leading to
compromised fairness and overall model accuracy. Moreover, many approaches rely
on the availability of demographic group labels, which are often costly to
annotate. This paper proposes AIM-Fair, aiming to overcome these limitations
and harness the potential of cutting-edge generative models in promoting
algorithmic fairness. We investigate a fine-tuning paradigm starting from a
biased model initially trained on real-world data without demographic
annotations. This model is then fine-tuned using unbiased synthetic data
generated by a state-of-the-art diffusion model to improve its fairness. Two
key challenges are identified in this fine-tuning paradigm, 1) the low quality
of synthetic data, which can still happen even with advanced generative models,
and 2) the domain and bias gap between real and synthetic data. To address the
limitation of synthetic data quality, we propose Contextual Synthetic Data
Generation (CSDG) to generate data using a text-to-image diffusion model (T2I)
with prompts generated by a context-aware LLM, ensuring both data diversity and
control of bias in synthetic data. To resolve domain and bias shifts, we
introduce a novel selective fine-tuning scheme in which only model parameters
more sensitive to bias and less sensitive to domain shift are updated.
Experiments on CelebA and UTKFace datasets show that our AIM-Fair improves
model fairness while maintaining utility, outperforming both fully and
partially fine-tuned approaches to model fairness.
| no_new_dataset | 0.953405 |
2503.05678 | Zhongyi Shui | Zhongyi Shui, Ruizhe Guo, Honglin Li, Yuxuan Sun, Yunlong Zhang,
Chenglu Zhu, Jiatong Cai, Pingyi Chen, Yanzhou Su, Lin Yang | Towards Effective and Efficient Context-aware Nucleus Detection in
Histopathology Whole Slide Images | under review | null | null | null | eess.IV cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Nucleus detection in histopathology whole slide images (WSIs) is crucial for
a broad spectrum of clinical applications. Current approaches for nucleus
detection in gigapixel WSIs utilize a sliding window methodology, which
overlooks boarder contextual information (eg, tissue structure) and easily
leads to inaccurate predictions. To address this problem, recent studies
additionally crops a large Filed-of-View (FoV) region around each sliding
window to extract contextual features. However, such methods substantially
increases the inference latency. In this paper, we propose an effective and
efficient context-aware nucleus detection algorithm. Specifically, instead of
leveraging large FoV regions, we aggregate contextual clues from off-the-shelf
features of historically visited sliding windows. This design greatly reduces
computational overhead. Moreover, compared to large FoV regions at a low
magnification, the sliding window patches have higher magnification and provide
finer-grained tissue details, thereby enhancing the detection accuracy. To
further improve the efficiency, we propose a grid pooling technique to compress
dense feature maps of each patch into a few contextual tokens. Finally, we
craft OCELOT-seg, the first benchmark dedicated to context-aware nucleus
instance segmentation. Code, dataset, and model checkpoints will be available
at https://github.com/windygoo/PathContext.
| [
{
"version": "v1",
"created": "Tue, 4 Mar 2025 02:01:53 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Shui",
"Zhongyi",
""
],
[
"Guo",
"Ruizhe",
""
],
[
"Li",
"Honglin",
""
],
[
"Sun",
"Yuxuan",
""
],
[
"Zhang",
"Yunlong",
""
],
[
"Zhu",
"Chenglu",
""
],
[
"Cai",
"Jiatong",
""
],
[
"Chen",
"Pingyi",
""
],
[
"Su",
"Yanzhou",
""
],
[
"Yang",
"Lin",
""
]
]
| TITLE: Towards Effective and Efficient Context-aware Nucleus Detection in
Histopathology Whole Slide Images
ABSTRACT: Nucleus detection in histopathology whole slide images (WSIs) is crucial for
a broad spectrum of clinical applications. Current approaches for nucleus
detection in gigapixel WSIs utilize a sliding window methodology, which
overlooks boarder contextual information (eg, tissue structure) and easily
leads to inaccurate predictions. To address this problem, recent studies
additionally crops a large Filed-of-View (FoV) region around each sliding
window to extract contextual features. However, such methods substantially
increases the inference latency. In this paper, we propose an effective and
efficient context-aware nucleus detection algorithm. Specifically, instead of
leveraging large FoV regions, we aggregate contextual clues from off-the-shelf
features of historically visited sliding windows. This design greatly reduces
computational overhead. Moreover, compared to large FoV regions at a low
magnification, the sliding window patches have higher magnification and provide
finer-grained tissue details, thereby enhancing the detection accuracy. To
further improve the efficiency, we propose a grid pooling technique to compress
dense feature maps of each patch into a few contextual tokens. Finally, we
craft OCELOT-seg, the first benchmark dedicated to context-aware nucleus
instance segmentation. Code, dataset, and model checkpoints will be available
at https://github.com/windygoo/PathContext.
| new_dataset | 0.739046 |
2503.05684 | Parameswaran Kamalaruban Dr. | Parameswaran Kamalaruban, Mark Anderson, Stuart Burrell, Maeve
Madigan, Piotr Skalski, David Sutton | Fairness-Aware Low-Rank Adaptation Under Demographic Privacy Constraints | null | null | null | null | cs.LG cs.CV | http://creativecommons.org/licenses/by/4.0/ | Pre-trained foundation models can be adapted for specific tasks using
Low-Rank Adaptation (LoRA). However, the fairness properties of these adapted
classifiers remain underexplored. Existing fairness-aware fine-tuning methods
rely on direct access to sensitive attributes or their predictors, but in
practice, these sensitive attributes are often held under strict consumer
privacy controls, and neither the attributes nor their predictors are available
to model developers, hampering the development of fair models. To address this
issue, we introduce a set of LoRA-based fine-tuning methods that can be trained
in a distributed fashion, where model developers and fairness auditors
collaborate without sharing sensitive attributes or predictors. In this paper,
we evaluate three such methods - sensitive unlearning, adversarial training,
and orthogonality loss - against a fairness-unaware baseline, using experiments
on the CelebA and UTK-Face datasets with an ImageNet pre-trained ViT-Base
model. We find that orthogonality loss consistently reduces bias while
maintaining or improving utility, whereas adversarial training improves False
Positive Rate Parity and Demographic Parity in some cases, and sensitive
unlearning provides no clear benefit. In tasks where significant biases are
present, distributed fairness-aware fine-tuning methods can effectively
eliminate bias without compromising consumer privacy and, in most cases,
improve model utility.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 18:49:57 GMT"
}
]
| 2025-03-10T00:00:00 | [
[
"Kamalaruban",
"Parameswaran",
""
],
[
"Anderson",
"Mark",
""
],
[
"Burrell",
"Stuart",
""
],
[
"Madigan",
"Maeve",
""
],
[
"Skalski",
"Piotr",
""
],
[
"Sutton",
"David",
""
]
]
| TITLE: Fairness-Aware Low-Rank Adaptation Under Demographic Privacy Constraints
ABSTRACT: Pre-trained foundation models can be adapted for specific tasks using
Low-Rank Adaptation (LoRA). However, the fairness properties of these adapted
classifiers remain underexplored. Existing fairness-aware fine-tuning methods
rely on direct access to sensitive attributes or their predictors, but in
practice, these sensitive attributes are often held under strict consumer
privacy controls, and neither the attributes nor their predictors are available
to model developers, hampering the development of fair models. To address this
issue, we introduce a set of LoRA-based fine-tuning methods that can be trained
in a distributed fashion, where model developers and fairness auditors
collaborate without sharing sensitive attributes or predictors. In this paper,
we evaluate three such methods - sensitive unlearning, adversarial training,
and orthogonality loss - against a fairness-unaware baseline, using experiments
on the CelebA and UTK-Face datasets with an ImageNet pre-trained ViT-Base
model. We find that orthogonality loss consistently reduces bias while
maintaining or improving utility, whereas adversarial training improves False
Positive Rate Parity and Demographic Parity in some cases, and sensitive
unlearning provides no clear benefit. In tasks where significant biases are
present, distributed fairness-aware fine-tuning methods can effectively
eliminate bias without compromising consumer privacy and, in most cases,
improve model utility.
| no_new_dataset | 0.948155 |
2206.02796 | Xihong Yang | Xihong Yang, Yiqi Wang, Yue Liu, Yi Wen, Lingyuan Meng, Sihang Zhou,
Xinwang Liu, En Zhu | Mixed Graph Contrastive Network for Semi-Supervised Node Classification | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Graph Neural Networks (GNNs) have achieved promising performance in
semi-supervised node classification in recent years. However, the problem of
insufficient supervision, together with representation collapse, largely limits
the performance of the GNNs in this field. To alleviate the collapse of node
representations in semi-supervised scenario, we propose a novel graph
contrastive learning method, termed Mixed Graph Contrastive Network (MGCN). In
our method, we improve the discriminative capability of the latent embeddings
by an interpolation-based augmentation strategy and a correlation reduction
mechanism. Specifically, we first conduct the interpolation-based augmentation
in the latent space and then force the prediction model to change linearly
between samples. Second, we enable the learned network to tell apart samples
across two interpolation-perturbed views through forcing the correlation matrix
across views to approximate an identity matrix. By combining the two settings,
we extract rich supervision information from both the abundant unlabeled nodes
and the rare yet valuable labeled nodes for discriminative representation
learning. Extensive experimental results on six datasets demonstrate the
effectiveness and the generality of MGCN compared to the existing
state-of-the-art methods. The code of MGCN is available at
https://github.com/xihongyang1999/MGCN on Github.
| [
{
"version": "v1",
"created": "Mon, 6 Jun 2022 14:26:34 GMT"
},
{
"version": "v2",
"created": "Thu, 16 Jun 2022 14:25:36 GMT"
},
{
"version": "v3",
"created": "Thu, 6 Mar 2025 09:10:18 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Yang",
"Xihong",
""
],
[
"Wang",
"Yiqi",
""
],
[
"Liu",
"Yue",
""
],
[
"Wen",
"Yi",
""
],
[
"Meng",
"Lingyuan",
""
],
[
"Zhou",
"Sihang",
""
],
[
"Liu",
"Xinwang",
""
],
[
"Zhu",
"En",
""
]
]
| TITLE: Mixed Graph Contrastive Network for Semi-Supervised Node Classification
ABSTRACT: Graph Neural Networks (GNNs) have achieved promising performance in
semi-supervised node classification in recent years. However, the problem of
insufficient supervision, together with representation collapse, largely limits
the performance of the GNNs in this field. To alleviate the collapse of node
representations in semi-supervised scenario, we propose a novel graph
contrastive learning method, termed Mixed Graph Contrastive Network (MGCN). In
our method, we improve the discriminative capability of the latent embeddings
by an interpolation-based augmentation strategy and a correlation reduction
mechanism. Specifically, we first conduct the interpolation-based augmentation
in the latent space and then force the prediction model to change linearly
between samples. Second, we enable the learned network to tell apart samples
across two interpolation-perturbed views through forcing the correlation matrix
across views to approximate an identity matrix. By combining the two settings,
we extract rich supervision information from both the abundant unlabeled nodes
and the rare yet valuable labeled nodes for discriminative representation
learning. Extensive experimental results on six datasets demonstrate the
effectiveness and the generality of MGCN compared to the existing
state-of-the-art methods. The code of MGCN is available at
https://github.com/xihongyang1999/MGCN on Github.
| no_new_dataset | 0.950411 |
2210.06746 | Hao Cui | Hao Cui, Rahmadi Trimananda, Scott Jordan, Athina Markopoulou | PoliGraph: Automated Privacy Policy Analysis using Knowledge Graphs
(Journal Version) | 46 pages, 15 figures (including subfigures). This is the 2025 updated
journal version. For the conference version published at USENIX Security '23,
see arXiv:2210.06746v2 | null | null | null | cs.CR | http://creativecommons.org/licenses/by/4.0/ | Privacy policies disclose how an organization collects and handles personal
information. Recent work has made progress in leveraging natural language
processing (NLP) to automate privacy policy analysis and extract data
collection statements from different sentences, considered in isolation from
each other. In this paper, we view and analyze, for the first time, the entire
text of a privacy policy in an integrated way. In terms of methodology: (1) we
define PoliGraph, a type of knowledge graph that captures statements in a
policy as relations between different parts of the text; and (2) we revisit the
notion of ontologies, previously defined in heuristic ways, to capture
subsumption relations between terms. We make a clear distinction between local
and global ontologies to capture the context of individual policies,
application domains, and privacy laws. We develop PoliGrapher, an NLP tool to
automatically extract PoliGraph from the text using linguistic analysis. Using
a public dataset for evaluation, we show that PoliGrapher identifies 40% more
collection statements than prior state-of-the-art, with 97% precision. In terms
of applications, PoliGraph enables automated analysis of a corpus of policies
and allows us to: (1) reveal common patterns in the texts across different
policies, and (2) assess the correctness of the terms as defined within a
policy. We also apply PoliGraph to: (3) detect contradictions in a policy,
where we show false alarms by prior work, and (4) analyze the consistency of
policies and network traffic, where we identify significantly more clear
disclosures than prior work. Finally, leveraging the capabilities of the
emerging large language models (LLMs), we also present PoliGrapher-LM, a tool
that uses LLM prompting instead of NLP linguistic analysis, to extract
PoliGraph from the policy text, and we show that it further improves coverage.
| [
{
"version": "v1",
"created": "Thu, 13 Oct 2022 05:16:22 GMT"
},
{
"version": "v2",
"created": "Tue, 20 Jun 2023 19:45:23 GMT"
},
{
"version": "v3",
"created": "Thu, 6 Mar 2025 06:47:40 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Cui",
"Hao",
""
],
[
"Trimananda",
"Rahmadi",
""
],
[
"Jordan",
"Scott",
""
],
[
"Markopoulou",
"Athina",
""
]
]
| TITLE: PoliGraph: Automated Privacy Policy Analysis using Knowledge Graphs
(Journal Version)
ABSTRACT: Privacy policies disclose how an organization collects and handles personal
information. Recent work has made progress in leveraging natural language
processing (NLP) to automate privacy policy analysis and extract data
collection statements from different sentences, considered in isolation from
each other. In this paper, we view and analyze, for the first time, the entire
text of a privacy policy in an integrated way. In terms of methodology: (1) we
define PoliGraph, a type of knowledge graph that captures statements in a
policy as relations between different parts of the text; and (2) we revisit the
notion of ontologies, previously defined in heuristic ways, to capture
subsumption relations between terms. We make a clear distinction between local
and global ontologies to capture the context of individual policies,
application domains, and privacy laws. We develop PoliGrapher, an NLP tool to
automatically extract PoliGraph from the text using linguistic analysis. Using
a public dataset for evaluation, we show that PoliGrapher identifies 40% more
collection statements than prior state-of-the-art, with 97% precision. In terms
of applications, PoliGraph enables automated analysis of a corpus of policies
and allows us to: (1) reveal common patterns in the texts across different
policies, and (2) assess the correctness of the terms as defined within a
policy. We also apply PoliGraph to: (3) detect contradictions in a policy,
where we show false alarms by prior work, and (4) analyze the consistency of
policies and network traffic, where we identify significantly more clear
disclosures than prior work. Finally, leveraging the capabilities of the
emerging large language models (LLMs), we also present PoliGrapher-LM, a tool
that uses LLM prompting instead of NLP linguistic analysis, to extract
PoliGraph from the policy text, and we show that it further improves coverage.
| no_new_dataset | 0.951863 |
2306.00723 | Lakmal Meegahapola | Wageesha Bangamuarachchi and Anju Chamantha and Lakmal Meegahapola and
Haeeun Kim and Salvador Ruiz-Correa and Indika Perera and Daniel Gatica-Perez | Inferring Mood-While-Eating with Smartphone Sensing and Community-Based
Model Personalization | ACM Transactions on Computing for Healthcare (HEALTH) | null | null | null | cs.HC cs.CY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The interplay between mood and eating episodes has been extensively
researched, revealing a connection between the two. Previous studies have
relied on questionnaires and mobile phone self-reports to investigate the
relationship between mood and eating. However, current literature exhibits
several limitations: a lack of investigation into the generalization of mood
inference models trained with data from various everyday life situations to
specific contexts like eating; an absence of studies using sensor data to
explore the intersection of mood and eating; and inadequate examination of
model personalization techniques within limited label settings, a common
challenge in mood inference (i.e., far fewer negative mood reports compared to
positive or neutral reports). In this study, we sought to examine everyday
eating behavior and mood using two datasets of college students in Mexico
(N_mex = 84, 1843 mood-while-eating reports) and eight countries (N_mul = 678,
24K mood-while-eating reports), which contain both passive smartphone sensing
and self-report data. Our results indicate that generic mood inference models
experience a decline in performance in specific contexts, such as during
eating, highlighting the issue of sub-context shifts in mobile sensing.
Moreover, we discovered that population-level (non-personalized) and hybrid
(partially personalized) modeling techniques fall short in the commonly used
three-class mood inference task (positive, neutral, negative). To overcome
these limitations, we implemented a novel community-based personalization
approach. Our findings demonstrate that mood-while-eating can be inferred with
accuracies 63.8% (with F1-score of 62.5) for the MEX dataset and 88.3% (with
F1-score of 85.7) with the MUL dataset using community-based models, surpassing
those achieved with traditional methods.
| [
{
"version": "v1",
"created": "Thu, 1 Jun 2023 14:24:10 GMT"
},
{
"version": "v2",
"created": "Sat, 9 Dec 2023 21:58:40 GMT"
},
{
"version": "v3",
"created": "Wed, 5 Mar 2025 21:05:04 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Bangamuarachchi",
"Wageesha",
""
],
[
"Chamantha",
"Anju",
""
],
[
"Meegahapola",
"Lakmal",
""
],
[
"Kim",
"Haeeun",
""
],
[
"Ruiz-Correa",
"Salvador",
""
],
[
"Perera",
"Indika",
""
],
[
"Gatica-Perez",
"Daniel",
""
]
]
| TITLE: Inferring Mood-While-Eating with Smartphone Sensing and Community-Based
Model Personalization
ABSTRACT: The interplay between mood and eating episodes has been extensively
researched, revealing a connection between the two. Previous studies have
relied on questionnaires and mobile phone self-reports to investigate the
relationship between mood and eating. However, current literature exhibits
several limitations: a lack of investigation into the generalization of mood
inference models trained with data from various everyday life situations to
specific contexts like eating; an absence of studies using sensor data to
explore the intersection of mood and eating; and inadequate examination of
model personalization techniques within limited label settings, a common
challenge in mood inference (i.e., far fewer negative mood reports compared to
positive or neutral reports). In this study, we sought to examine everyday
eating behavior and mood using two datasets of college students in Mexico
(N_mex = 84, 1843 mood-while-eating reports) and eight countries (N_mul = 678,
24K mood-while-eating reports), which contain both passive smartphone sensing
and self-report data. Our results indicate that generic mood inference models
experience a decline in performance in specific contexts, such as during
eating, highlighting the issue of sub-context shifts in mobile sensing.
Moreover, we discovered that population-level (non-personalized) and hybrid
(partially personalized) modeling techniques fall short in the commonly used
three-class mood inference task (positive, neutral, negative). To overcome
these limitations, we implemented a novel community-based personalization
approach. Our findings demonstrate that mood-while-eating can be inferred with
accuracies 63.8% (with F1-score of 62.5) for the MEX dataset and 88.3% (with
F1-score of 85.7) with the MUL dataset using community-based models, surpassing
those achieved with traditional methods.
| no_new_dataset | 0.937268 |
2307.04014 | Amirhossein Askari Farsangi | Amirhossein Askari Farsangi, Ali Sharifi-Zarchi, Mohammad Hossein
Rohban | Novel Pipeline for Diagnosing Acute Lymphoblastic Leukemia Sensitive to
Related Biomarkers | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Acute Lymphoblastic Leukemia (ALL) is one of the most common types of
childhood blood cancer. The quick start of the treatment process is critical to
saving the patient's life, and for this reason, early diagnosis of this disease
is essential. Examining the blood smear images of these patients is one of the
methods used by expert doctors to diagnose this disease. Deep learning-based
methods have numerous applications in medical fields, as they have
significantly advanced in recent years. ALL diagnosis is not an exception in
this field, and several machine learning-based methods for this problem have
been proposed. In previous methods, high diagnostic accuracy was reported, but
our work showed that this alone is not sufficient, as it can lead to models
taking shortcuts and not making meaningful decisions. This issue arises due to
the small size of medical training datasets. To address this, we constrained
our model to follow a pipeline inspired by experts' work. We also demonstrated
that, since a judgement based on only one image is insufficient, redefining the
problem as a multiple-instance learning problem is necessary for achieving a
practical result. Our model is the first to provide a solution to this problem
in a multiple-instance learning setup. We introduced a novel pipeline for
diagnosing ALL that approximates the process used by hematologists, is
sensitive to disease biomarkers, and achieves an accuracy of 96.15%, an
F1-score of 94.24%, a sensitivity of 97.56%, and a specificity of 90.91% on ALL
IDB 1. Our method was further evaluated on an out-of-distribution dataset,
which posed a challenging test and had acceptable performance. Notably, our
model was trained on a relatively small dataset, highlighting the potential for
our approach to be applied to other medical datasets with limited data
availability.
| [
{
"version": "v1",
"created": "Sat, 8 Jul 2023 16:46:16 GMT"
},
{
"version": "v2",
"created": "Tue, 11 Jul 2023 11:16:17 GMT"
},
{
"version": "v3",
"created": "Thu, 6 Mar 2025 06:55:15 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Farsangi",
"Amirhossein Askari",
""
],
[
"Sharifi-Zarchi",
"Ali",
""
],
[
"Rohban",
"Mohammad Hossein",
""
]
]
| TITLE: Novel Pipeline for Diagnosing Acute Lymphoblastic Leukemia Sensitive to
Related Biomarkers
ABSTRACT: Acute Lymphoblastic Leukemia (ALL) is one of the most common types of
childhood blood cancer. The quick start of the treatment process is critical to
saving the patient's life, and for this reason, early diagnosis of this disease
is essential. Examining the blood smear images of these patients is one of the
methods used by expert doctors to diagnose this disease. Deep learning-based
methods have numerous applications in medical fields, as they have
significantly advanced in recent years. ALL diagnosis is not an exception in
this field, and several machine learning-based methods for this problem have
been proposed. In previous methods, high diagnostic accuracy was reported, but
our work showed that this alone is not sufficient, as it can lead to models
taking shortcuts and not making meaningful decisions. This issue arises due to
the small size of medical training datasets. To address this, we constrained
our model to follow a pipeline inspired by experts' work. We also demonstrated
that, since a judgement based on only one image is insufficient, redefining the
problem as a multiple-instance learning problem is necessary for achieving a
practical result. Our model is the first to provide a solution to this problem
in a multiple-instance learning setup. We introduced a novel pipeline for
diagnosing ALL that approximates the process used by hematologists, is
sensitive to disease biomarkers, and achieves an accuracy of 96.15%, an
F1-score of 94.24%, a sensitivity of 97.56%, and a specificity of 90.91% on ALL
IDB 1. Our method was further evaluated on an out-of-distribution dataset,
which posed a challenging test and had acceptable performance. Notably, our
model was trained on a relatively small dataset, highlighting the potential for
our approach to be applied to other medical datasets with limited data
availability.
| no_new_dataset | 0.945248 |
2307.15421 | Wei Jiang | Wei Jiang, Jiayu Yang, Yongqi Zhai, Feng Gao, Ronggang Wang | MLIC++: Linear Complexity Multi-Reference Entropy Modeling for Learned
Image Compression | Accepted to ICML 2023 Neural Compression Workshop and ACM
Transactions on Multimedia Computing, Communications, and Applications 2025 | null | 10.1145/3719011 | null | eess.IV cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The latent representation in learned image compression encompasses
channel-wise, local spatial, and global spatial correlations, which are
essential for the entropy model to capture for conditional entropy
minimization. Efficiently capturing these contexts within a single entropy
model, especially in high-resolution image coding, presents a challenge due to
the computational complexity of existing global context modules. To address
this challenge, we propose the Linear Complexity Multi-Reference Entropy Model
(MEM$^{++}$). Specifically, the latent representation is partitioned into
multiple slices. For channel-wise contexts, previously compressed slices serve
as the context for compressing a particular slice. For local contexts, we
introduce a shifted-window-based checkerboard attention module. This module
ensures linear complexity without sacrificing performance. For global contexts,
we propose a linear complexity attention mechanism. It captures global
correlations by decomposing the softmax operation, enabling the implicit
computation of attention maps from previously decoded slices. Using MEM$^{++}$
as the entropy model, we develop the image compression method MLIC$^{++}$.
Extensive experimental results demonstrate that MLIC$^{++}$ achieves
state-of-the-art performance, reducing BD-rate by $13.39\%$ on the Kodak
dataset compared to VTM-17.0 in Peak Signal-to-Noise Ratio (PSNR). Furthermore,
MLIC$^{++}$ exhibits linear computational complexity and memory consumption
with resolution, making it highly suitable for high-resolution image coding.
Code and pre-trained models are available at
https://github.com/JiangWeibeta/MLIC. Training dataset is available at
https://huggingface.co/datasets/Whiteboat/MLIC-Train-100K.
| [
{
"version": "v1",
"created": "Fri, 28 Jul 2023 09:11:37 GMT"
},
{
"version": "v10",
"created": "Sat, 8 Feb 2025 08:12:31 GMT"
},
{
"version": "v11",
"created": "Mon, 17 Feb 2025 08:41:30 GMT"
},
{
"version": "v2",
"created": "Sun, 3 Sep 2023 09:01:43 GMT"
},
{
"version": "v3",
"created": "Mon, 30 Oct 2023 05:56:08 GMT"
},
{
"version": "v4",
"created": "Mon, 18 Dec 2023 09:02:57 GMT"
},
{
"version": "v5",
"created": "Sun, 7 Jan 2024 03:52:03 GMT"
},
{
"version": "v6",
"created": "Tue, 16 Jan 2024 15:15:49 GMT"
},
{
"version": "v7",
"created": "Sat, 3 Feb 2024 09:12:10 GMT"
},
{
"version": "v8",
"created": "Wed, 14 Feb 2024 11:13:49 GMT"
},
{
"version": "v9",
"created": "Tue, 20 Feb 2024 03:25:43 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Jiang",
"Wei",
""
],
[
"Yang",
"Jiayu",
""
],
[
"Zhai",
"Yongqi",
""
],
[
"Gao",
"Feng",
""
],
[
"Wang",
"Ronggang",
""
]
]
| TITLE: MLIC++: Linear Complexity Multi-Reference Entropy Modeling for Learned
Image Compression
ABSTRACT: The latent representation in learned image compression encompasses
channel-wise, local spatial, and global spatial correlations, which are
essential for the entropy model to capture for conditional entropy
minimization. Efficiently capturing these contexts within a single entropy
model, especially in high-resolution image coding, presents a challenge due to
the computational complexity of existing global context modules. To address
this challenge, we propose the Linear Complexity Multi-Reference Entropy Model
(MEM$^{++}$). Specifically, the latent representation is partitioned into
multiple slices. For channel-wise contexts, previously compressed slices serve
as the context for compressing a particular slice. For local contexts, we
introduce a shifted-window-based checkerboard attention module. This module
ensures linear complexity without sacrificing performance. For global contexts,
we propose a linear complexity attention mechanism. It captures global
correlations by decomposing the softmax operation, enabling the implicit
computation of attention maps from previously decoded slices. Using MEM$^{++}$
as the entropy model, we develop the image compression method MLIC$^{++}$.
Extensive experimental results demonstrate that MLIC$^{++}$ achieves
state-of-the-art performance, reducing BD-rate by $13.39\%$ on the Kodak
dataset compared to VTM-17.0 in Peak Signal-to-Noise Ratio (PSNR). Furthermore,
MLIC$^{++}$ exhibits linear computational complexity and memory consumption
with resolution, making it highly suitable for high-resolution image coding.
Code and pre-trained models are available at
https://github.com/JiangWeibeta/MLIC. Training dataset is available at
https://huggingface.co/datasets/Whiteboat/MLIC-Train-100K.
| no_new_dataset | 0.950824 |
2309.00889 | Alper Ahmetoglu | Alper Ahmetoglu, Batuhan Celik, Erhan Oztop, Emre Ugur | Discovering Predictive Relational Object Symbols with Symbolic Attentive
Layers | arXiv admin note: text overlap with arXiv:2208.01021 | null | 10.1109/LRA.2024.3350994 | null | cs.RO cs.LG | http://creativecommons.org/licenses/by/4.0/ | In this paper, we propose and realize a new deep learning architecture for
discovering symbolic representations for objects and their relations based on
the self-supervised continuous interaction of a manipulator robot with multiple
objects on a tabletop environment. The key feature of the model is that it can
handle a changing number number of objects naturally and map the object-object
relations into symbolic domain explicitly. In the model, we employ a
self-attention layer that computes discrete attention weights from object
features, which are treated as relational symbols between objects. These
relational symbols are then used to aggregate the learned object symbols and
predict the effects of executed actions on each object. The result is a
pipeline that allows the formation of object symbols and relational symbols
from a dataset of object features, actions, and effects in an end-to-end
manner. We compare the performance of our proposed architecture with
state-of-the-art symbol discovery methods in a simulated tabletop environment
where the robot needs to discover symbols related to the relative positions of
objects to predict the observed effect successfully. Our experiments show that
the proposed architecture performs better than other baselines in effect
prediction while forming not only object symbols but also relational symbols.
Furthermore, we analyze the learned symbols and relational patterns between
objects to learn about how the model interprets the environment. Our analysis
shows that the learned symbols relate to the relative positions of objects,
object types, and their horizontal alignment on the table, which reflect the
regularities in the environment.
| [
{
"version": "v1",
"created": "Sat, 2 Sep 2023 10:06:10 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Ahmetoglu",
"Alper",
""
],
[
"Celik",
"Batuhan",
""
],
[
"Oztop",
"Erhan",
""
],
[
"Ugur",
"Emre",
""
]
]
| TITLE: Discovering Predictive Relational Object Symbols with Symbolic Attentive
Layers
ABSTRACT: In this paper, we propose and realize a new deep learning architecture for
discovering symbolic representations for objects and their relations based on
the self-supervised continuous interaction of a manipulator robot with multiple
objects on a tabletop environment. The key feature of the model is that it can
handle a changing number number of objects naturally and map the object-object
relations into symbolic domain explicitly. In the model, we employ a
self-attention layer that computes discrete attention weights from object
features, which are treated as relational symbols between objects. These
relational symbols are then used to aggregate the learned object symbols and
predict the effects of executed actions on each object. The result is a
pipeline that allows the formation of object symbols and relational symbols
from a dataset of object features, actions, and effects in an end-to-end
manner. We compare the performance of our proposed architecture with
state-of-the-art symbol discovery methods in a simulated tabletop environment
where the robot needs to discover symbols related to the relative positions of
objects to predict the observed effect successfully. Our experiments show that
the proposed architecture performs better than other baselines in effect
prediction while forming not only object symbols but also relational symbols.
Furthermore, we analyze the learned symbols and relational patterns between
objects to learn about how the model interprets the environment. Our analysis
shows that the learned symbols relate to the relative positions of objects,
object types, and their horizontal alignment on the table, which reflect the
regularities in the environment.
| no_new_dataset | 0.949529 |
2311.07978 | David Adelani | Jessica Ojo, Odunayo Ogundepo, Akintunde Oladipo, Kelechi Ogueji,
Jimmy Lin, Pontus Stenetorp, David Ifeoluwa Adelani | AfroBench: How Good are Large Language Models on African Languages? | Under review | null | null | null | cs.CL cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Large-scale multilingual evaluations, such as MEGA, often include only a
handful of African languages due to the scarcity of high-quality evaluation
data and the limited discoverability of existing African datasets. This lack of
representation hinders comprehensive LLM evaluation across a diverse range of
languages and tasks. To address these challenges, we introduce AfroBench -- a
multi-task benchmark for evaluating the performance of LLMs across 64 African
languages, 15 tasks and 22 datasets. AfroBench consists of nine natural
language understanding datasets, six text generation datasets, six knowledge
and question answering tasks, and one mathematical reasoning task. We present
results comparing the performance of prompting LLMs to fine-tuned baselines
based on BERT and T5-style models. Our results suggest large gaps in
performance between high-resource languages, such as English, and African
languages across most tasks; but performance also varies based on the
availability of monolingual data resources. Our findings confirm that
performance on African languages continues to remain a hurdle for current LLMs,
underscoring the need for additional efforts to close this gap.
https://mcgill-nlp.github.io/AfroBench/
| [
{
"version": "v1",
"created": "Tue, 14 Nov 2023 08:10:14 GMT"
},
{
"version": "v2",
"created": "Tue, 30 Apr 2024 16:04:16 GMT"
},
{
"version": "v3",
"created": "Wed, 26 Feb 2025 15:16:47 GMT"
},
{
"version": "v4",
"created": "Thu, 6 Mar 2025 13:29:24 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Ojo",
"Jessica",
""
],
[
"Ogundepo",
"Odunayo",
""
],
[
"Oladipo",
"Akintunde",
""
],
[
"Ogueji",
"Kelechi",
""
],
[
"Lin",
"Jimmy",
""
],
[
"Stenetorp",
"Pontus",
""
],
[
"Adelani",
"David Ifeoluwa",
""
]
]
| TITLE: AfroBench: How Good are Large Language Models on African Languages?
ABSTRACT: Large-scale multilingual evaluations, such as MEGA, often include only a
handful of African languages due to the scarcity of high-quality evaluation
data and the limited discoverability of existing African datasets. This lack of
representation hinders comprehensive LLM evaluation across a diverse range of
languages and tasks. To address these challenges, we introduce AfroBench -- a
multi-task benchmark for evaluating the performance of LLMs across 64 African
languages, 15 tasks and 22 datasets. AfroBench consists of nine natural
language understanding datasets, six text generation datasets, six knowledge
and question answering tasks, and one mathematical reasoning task. We present
results comparing the performance of prompting LLMs to fine-tuned baselines
based on BERT and T5-style models. Our results suggest large gaps in
performance between high-resource languages, such as English, and African
languages across most tasks; but performance also varies based on the
availability of monolingual data resources. Our findings confirm that
performance on African languages continues to remain a hurdle for current LLMs,
underscoring the need for additional efforts to close this gap.
https://mcgill-nlp.github.io/AfroBench/
| new_dataset | 0.969928 |
2312.03286 | Hongsin Lee | Hongsin Lee, Seungju Cho, Changick Kim | Indirect Gradient Matching for Adversarial Robust Distillation | ICLR 2025 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Adversarial training significantly improves adversarial robustness, but
superior performance is primarily attained with large models. This substantial
performance gap for smaller models has spurred active research into adversarial
distillation (AD) to mitigate the difference. Existing AD methods leverage the
teacher's logits as a guide. In contrast to these approaches, we aim to
transfer another piece of knowledge from the teacher, the input gradient. In
this paper, we propose a distillation module termed Indirect Gradient
Distillation Module (IGDM) that indirectly matches the student's input gradient
with that of the teacher. Experimental results show that IGDM seamlessly
integrates with existing AD methods, significantly enhancing their performance.
Particularly, utilizing IGDM on the CIFAR-100 dataset improves the AutoAttack
accuracy from 28.06% to 30.32% with the ResNet-18 architecture and from 26.18%
to 29.32% with the MobileNetV2 architecture when integrated into the SOTA
method without additional data augmentation.
| [
{
"version": "v1",
"created": "Wed, 6 Dec 2023 04:32:38 GMT"
},
{
"version": "v2",
"created": "Thu, 6 Mar 2025 10:12:31 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Lee",
"Hongsin",
""
],
[
"Cho",
"Seungju",
""
],
[
"Kim",
"Changick",
""
]
]
| TITLE: Indirect Gradient Matching for Adversarial Robust Distillation
ABSTRACT: Adversarial training significantly improves adversarial robustness, but
superior performance is primarily attained with large models. This substantial
performance gap for smaller models has spurred active research into adversarial
distillation (AD) to mitigate the difference. Existing AD methods leverage the
teacher's logits as a guide. In contrast to these approaches, we aim to
transfer another piece of knowledge from the teacher, the input gradient. In
this paper, we propose a distillation module termed Indirect Gradient
Distillation Module (IGDM) that indirectly matches the student's input gradient
with that of the teacher. Experimental results show that IGDM seamlessly
integrates with existing AD methods, significantly enhancing their performance.
Particularly, utilizing IGDM on the CIFAR-100 dataset improves the AutoAttack
accuracy from 28.06% to 30.32% with the ResNet-18 architecture and from 26.18%
to 29.32% with the MobileNetV2 architecture when integrated into the SOTA
method without additional data augmentation.
| no_new_dataset | 0.948917 |
2401.13097 | Michelle Greene | Michelle R. Greene, Mariam Josyula, Wentao Si and Jennifer A. Hart | Digital Divides in Scene Recognition: Uncovering Socioeconomic Biases in
Deep Learning Systems | 28 pages, 3 figures, 6 tables | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | Computer-based scene understanding has influenced fields ranging from urban
planning to autonomous vehicle performance, yet little is known about how well
these technologies work across social differences. We investigate the biases of
deep convolutional neural networks (dCNNs) in scene classification, using
nearly one million images from global and US sources, including user-submitted
home photographs and Airbnb listings. We applied statistical models to quantify
the impact of socioeconomic indicators such as family income, Human Development
Index (HDI), and demographic factors from public data sources (CIA and US
Census) on dCNN performance. Our analyses revealed significant socioeconomic
bias, where pretrained dCNNs demonstrated lower classification accuracy, lower
classification confidence, and a higher tendency to assign labels that could be
offensive when applied to homes (e.g., "ruin", "slum"), especially in images
from homes with lower socioeconomic status (SES). This trend is consistent
across two datasets of international images and within the diverse economic and
racial landscapes of the United States. This research contributes to
understanding biases in computer vision, emphasizing the need for more
inclusive and representative training datasets. By mitigating the bias in the
computer vision pipelines, we can ensure fairer and more equitable outcomes for
applied computer vision, including home valuation and smart home security
systems. There is urgency in addressing these biases, which can significantly
impact critical decisions in urban development and resource allocation. Our
findings also motivate the development of AI systems that better understand and
serve diverse communities, moving towards technology that equitably benefits
all sectors of society.
| [
{
"version": "v1",
"created": "Tue, 23 Jan 2024 21:22:06 GMT"
},
{
"version": "v2",
"created": "Wed, 5 Mar 2025 21:31:31 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Greene",
"Michelle R.",
""
],
[
"Josyula",
"Mariam",
""
],
[
"Si",
"Wentao",
""
],
[
"Hart",
"Jennifer A.",
""
]
]
| TITLE: Digital Divides in Scene Recognition: Uncovering Socioeconomic Biases in
Deep Learning Systems
ABSTRACT: Computer-based scene understanding has influenced fields ranging from urban
planning to autonomous vehicle performance, yet little is known about how well
these technologies work across social differences. We investigate the biases of
deep convolutional neural networks (dCNNs) in scene classification, using
nearly one million images from global and US sources, including user-submitted
home photographs and Airbnb listings. We applied statistical models to quantify
the impact of socioeconomic indicators such as family income, Human Development
Index (HDI), and demographic factors from public data sources (CIA and US
Census) on dCNN performance. Our analyses revealed significant socioeconomic
bias, where pretrained dCNNs demonstrated lower classification accuracy, lower
classification confidence, and a higher tendency to assign labels that could be
offensive when applied to homes (e.g., "ruin", "slum"), especially in images
from homes with lower socioeconomic status (SES). This trend is consistent
across two datasets of international images and within the diverse economic and
racial landscapes of the United States. This research contributes to
understanding biases in computer vision, emphasizing the need for more
inclusive and representative training datasets. By mitigating the bias in the
computer vision pipelines, we can ensure fairer and more equitable outcomes for
applied computer vision, including home valuation and smart home security
systems. There is urgency in addressing these biases, which can significantly
impact critical decisions in urban development and resource allocation. Our
findings also motivate the development of AI systems that better understand and
serve diverse communities, moving towards technology that equitably benefits
all sectors of society.
| no_new_dataset | 0.947672 |
2401.13898 | Huy Le Quang | Huy Q. Le, Chu Myaet Thwal, Yu Qiao, Ye Lin Tun, Minh N. H. Nguyen,
Eui-Nam Huh and Choong Seon Hong | Cross-Modal Prototype based Multimodal Federated Learning under Severely
Missing Modality | 14 pages, 8 figures, 11 tables | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multimodal federated learning (MFL) has emerged as a decentralized machine
learning paradigm, allowing multiple clients with different modalities to
collaborate on training a global model across diverse data sources without
sharing their private data. However, challenges, such as data heterogeneity and
severely missing modalities, pose crucial hindrances to the robustness of MFL,
significantly impacting the performance of global model. The occurrence of
missing modalities in real-world applications, such as autonomous driving,
often arises from factors like sensor failures, leading knowledge gaps during
the training process. Specifically, the absence of a modality introduces
misalignment during the local training phase, stemming from zero-filling in the
case of clients with missing modalities. Consequently, achieving robust
generalization in global model becomes imperative, especially when dealing with
clients that have incomplete data. In this paper, we propose
$\textbf{Multimodal Federated Cross Prototype Learning (MFCPL)}$, a novel
approach for MFL under severely missing modalities. Our MFCPL leverages the
complete prototypes to provide diverse modality knowledge in modality-shared
level with the cross-modal regularization and modality-specific level with
cross-modal contrastive mechanism. Additionally, our approach introduces the
cross-modal alignment to provide regularization for modality-specific features,
thereby enhancing the overall performance, particularly in scenarios involving
severely missing modalities. Through extensive experiments on three multimodal
datasets, we demonstrate the effectiveness of MFCPL in mitigating the
challenges of data heterogeneity and severely missing modalities while
improving the overall performance and robustness of MFL.
| [
{
"version": "v1",
"created": "Thu, 25 Jan 2024 02:25:23 GMT"
},
{
"version": "v2",
"created": "Thu, 6 Mar 2025 11:38:00 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Le",
"Huy Q.",
""
],
[
"Thwal",
"Chu Myaet",
""
],
[
"Qiao",
"Yu",
""
],
[
"Tun",
"Ye Lin",
""
],
[
"Nguyen",
"Minh N. H.",
""
],
[
"Huh",
"Eui-Nam",
""
],
[
"Hong",
"Choong Seon",
""
]
]
| TITLE: Cross-Modal Prototype based Multimodal Federated Learning under Severely
Missing Modality
ABSTRACT: Multimodal federated learning (MFL) has emerged as a decentralized machine
learning paradigm, allowing multiple clients with different modalities to
collaborate on training a global model across diverse data sources without
sharing their private data. However, challenges, such as data heterogeneity and
severely missing modalities, pose crucial hindrances to the robustness of MFL,
significantly impacting the performance of global model. The occurrence of
missing modalities in real-world applications, such as autonomous driving,
often arises from factors like sensor failures, leading knowledge gaps during
the training process. Specifically, the absence of a modality introduces
misalignment during the local training phase, stemming from zero-filling in the
case of clients with missing modalities. Consequently, achieving robust
generalization in global model becomes imperative, especially when dealing with
clients that have incomplete data. In this paper, we propose
$\textbf{Multimodal Federated Cross Prototype Learning (MFCPL)}$, a novel
approach for MFL under severely missing modalities. Our MFCPL leverages the
complete prototypes to provide diverse modality knowledge in modality-shared
level with the cross-modal regularization and modality-specific level with
cross-modal contrastive mechanism. Additionally, our approach introduces the
cross-modal alignment to provide regularization for modality-specific features,
thereby enhancing the overall performance, particularly in scenarios involving
severely missing modalities. Through extensive experiments on three multimodal
datasets, we demonstrate the effectiveness of MFCPL in mitigating the
challenges of data heterogeneity and severely missing modalities while
improving the overall performance and robustness of MFL.
| no_new_dataset | 0.951908 |
2402.01879 | Antonio Emanuele Cin\`a | Antonio Emanuele Cin\`a, Francesco Villani, Maura Pintor, Lea
Sch\"onherr, Battista Biggio, and Marcello Pelillo | $\sigma$-zero: Gradient-based Optimization of $\ell_0$-norm Adversarial
Examples | Paper accepted at International Conference on Learning
Representations (ICLR 2025). Code available at
https://github.com/sigma0-advx/sigma-zero | null | null | null | cs.LG cs.CR cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Evaluating the adversarial robustness of deep networks to gradient-based
attacks is challenging. While most attacks consider $\ell_2$- and
$\ell_\infty$-norm constraints to craft input perturbations, only a few
investigate sparse $\ell_1$- and $\ell_0$-norm attacks. In particular,
$\ell_0$-norm attacks remain the least studied due to the inherent complexity
of optimizing over a non-convex and non-differentiable constraint. However,
evaluating adversarial robustness under these attacks could reveal weaknesses
otherwise left untested with more conventional $\ell_2$- and $\ell_\infty$-norm
attacks. In this work, we propose a novel $\ell_0$-norm attack, called
$\sigma$-zero, which leverages a differentiable approximation of the $\ell_0$
norm to facilitate gradient-based optimization, and an adaptive projection
operator to dynamically adjust the trade-off between loss minimization and
perturbation sparsity. Extensive evaluations using MNIST, CIFAR10, and ImageNet
datasets, involving robust and non-robust models, show that
$\sigma$\texttt{-zero} finds minimum $\ell_0$-norm adversarial examples without
requiring any time-consuming hyperparameter tuning, and that it outperforms all
competing sparse attacks in terms of success rate, perturbation size, and
efficiency.
| [
{
"version": "v1",
"created": "Fri, 2 Feb 2024 20:08:11 GMT"
},
{
"version": "v2",
"created": "Wed, 2 Oct 2024 12:42:56 GMT"
},
{
"version": "v3",
"created": "Thu, 6 Mar 2025 11:05:33 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Cinà",
"Antonio Emanuele",
""
],
[
"Villani",
"Francesco",
""
],
[
"Pintor",
"Maura",
""
],
[
"Schönherr",
"Lea",
""
],
[
"Biggio",
"Battista",
""
],
[
"Pelillo",
"Marcello",
""
]
]
| TITLE: $\sigma$-zero: Gradient-based Optimization of $\ell_0$-norm Adversarial
Examples
ABSTRACT: Evaluating the adversarial robustness of deep networks to gradient-based
attacks is challenging. While most attacks consider $\ell_2$- and
$\ell_\infty$-norm constraints to craft input perturbations, only a few
investigate sparse $\ell_1$- and $\ell_0$-norm attacks. In particular,
$\ell_0$-norm attacks remain the least studied due to the inherent complexity
of optimizing over a non-convex and non-differentiable constraint. However,
evaluating adversarial robustness under these attacks could reveal weaknesses
otherwise left untested with more conventional $\ell_2$- and $\ell_\infty$-norm
attacks. In this work, we propose a novel $\ell_0$-norm attack, called
$\sigma$-zero, which leverages a differentiable approximation of the $\ell_0$
norm to facilitate gradient-based optimization, and an adaptive projection
operator to dynamically adjust the trade-off between loss minimization and
perturbation sparsity. Extensive evaluations using MNIST, CIFAR10, and ImageNet
datasets, involving robust and non-robust models, show that
$\sigma$\texttt{-zero} finds minimum $\ell_0$-norm adversarial examples without
requiring any time-consuming hyperparameter tuning, and that it outperforms all
competing sparse attacks in terms of success rate, perturbation size, and
efficiency.
| no_new_dataset | 0.93511 |
2402.02998 | Pierre Ablin | Yu-Guan Hsieh, James Thornton, Eugene Ndiaye, Michal Klein, Marco
Cuturi, Pierre Ablin | Careful with that Scalpel: Improving Gradient Surgery with an EMA | null | null | null | null | cs.LG stat.ML | http://creativecommons.org/licenses/by/4.0/ | Beyond minimizing a single training loss, many deep learning estimation
pipelines rely on an auxiliary objective to quantify and encourage desirable
properties of the model (e.g. performance on another dataset, robustness,
agreement with a prior). Although the simplest approach to incorporating an
auxiliary loss is to sum it with the training loss as a regularizer, recent
works have shown that one can improve performance by blending the gradients
beyond a simple sum; this is known as gradient surgery. We cast the problem as
a constrained minimization problem where the auxiliary objective is minimized
among the set of minimizers of the training loss. To solve this bilevel
problem, we follow a parameter update direction that combines the training loss
gradient and the orthogonal projection of the auxiliary gradient to the
training gradient. In a setting where gradients come from mini-batches, we
explain how, using a moving average of the training loss gradients, we can
carefully maintain this critical orthogonality property. We demonstrate that
our method, Bloop, can lead to much better performances on NLP and vision
experiments than other gradient surgery methods without EMA.
| [
{
"version": "v1",
"created": "Mon, 5 Feb 2024 13:37:00 GMT"
},
{
"version": "v2",
"created": "Thu, 6 Mar 2025 08:57:29 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Hsieh",
"Yu-Guan",
""
],
[
"Thornton",
"James",
""
],
[
"Ndiaye",
"Eugene",
""
],
[
"Klein",
"Michal",
""
],
[
"Cuturi",
"Marco",
""
],
[
"Ablin",
"Pierre",
""
]
]
| TITLE: Careful with that Scalpel: Improving Gradient Surgery with an EMA
ABSTRACT: Beyond minimizing a single training loss, many deep learning estimation
pipelines rely on an auxiliary objective to quantify and encourage desirable
properties of the model (e.g. performance on another dataset, robustness,
agreement with a prior). Although the simplest approach to incorporating an
auxiliary loss is to sum it with the training loss as a regularizer, recent
works have shown that one can improve performance by blending the gradients
beyond a simple sum; this is known as gradient surgery. We cast the problem as
a constrained minimization problem where the auxiliary objective is minimized
among the set of minimizers of the training loss. To solve this bilevel
problem, we follow a parameter update direction that combines the training loss
gradient and the orthogonal projection of the auxiliary gradient to the
training gradient. In a setting where gradients come from mini-batches, we
explain how, using a moving average of the training loss gradients, we can
carefully maintain this critical orthogonality property. We demonstrate that
our method, Bloop, can lead to much better performances on NLP and vision
experiments than other gradient surgery methods without EMA.
| no_new_dataset | 0.942665 |
2402.04722 | Kit Gallagher | Kit Gallagher, Richard Creswell, Ben Lambert, Martin Robinson, Chon
Lok Lei, Gary R. Mirams, David J. Gavaghan | Ten simple rules for training scientists to make better software | 20 pages, 7 figures | null | 10.1371/journal.pcbi.1012410 | null | cs.CY cs.SE | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Computational methods and associated software implementations are central to
every field of scientific investigation. Modern biological research,
particularly within systems biology, has relied heavily on the development of
software tools to process and organize increasingly large datasets, simulate
complex mechanistic models, provide tools for the analysis and management of
data, and visualize and organize outputs. However, developing high-quality
research software requires scientists to develop a host of software development
skills, and teaching these skills to students is challenging. There has been a
growing importance placed on ensuring reproducibility and good development
practices in computational research. However, less attention has been devoted
to informing the specific teaching strategies which are effective at nurturing
in researchers the complex skillset required to produce high-quality software
that, increasingly, is required to underpin both academic and industrial
biomedical research. Recent articles in the Ten Simple Rules collection have
discussed the teaching of foundational computer science and coding techniques
to biology students. We advance this discussion by describing the specific
steps for effectively teaching the necessary skills scientists need to develop
sustainable software packages which are fit for (re-)use in academic research
or more widely. Although our advice is likely to be applicable to all students
and researchers hoping to improve their software development skills, our
guidelines are directed towards an audience of students that have some
programming literacy but little formal training in software development or
engineering, typical of early doctoral students. These practices are also
applicable outside of doctoral training environments, and we believe they
should form a key part of postgraduate training schemes more generally in the
life sciences.
| [
{
"version": "v1",
"created": "Wed, 7 Feb 2024 10:16:20 GMT"
},
{
"version": "v2",
"created": "Thu, 6 Mar 2025 17:54:09 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Gallagher",
"Kit",
""
],
[
"Creswell",
"Richard",
""
],
[
"Lambert",
"Ben",
""
],
[
"Robinson",
"Martin",
""
],
[
"Lei",
"Chon Lok",
""
],
[
"Mirams",
"Gary R.",
""
],
[
"Gavaghan",
"David J.",
""
]
]
| TITLE: Ten simple rules for training scientists to make better software
ABSTRACT: Computational methods and associated software implementations are central to
every field of scientific investigation. Modern biological research,
particularly within systems biology, has relied heavily on the development of
software tools to process and organize increasingly large datasets, simulate
complex mechanistic models, provide tools for the analysis and management of
data, and visualize and organize outputs. However, developing high-quality
research software requires scientists to develop a host of software development
skills, and teaching these skills to students is challenging. There has been a
growing importance placed on ensuring reproducibility and good development
practices in computational research. However, less attention has been devoted
to informing the specific teaching strategies which are effective at nurturing
in researchers the complex skillset required to produce high-quality software
that, increasingly, is required to underpin both academic and industrial
biomedical research. Recent articles in the Ten Simple Rules collection have
discussed the teaching of foundational computer science and coding techniques
to biology students. We advance this discussion by describing the specific
steps for effectively teaching the necessary skills scientists need to develop
sustainable software packages which are fit for (re-)use in academic research
or more widely. Although our advice is likely to be applicable to all students
and researchers hoping to improve their software development skills, our
guidelines are directed towards an audience of students that have some
programming literacy but little formal training in software development or
engineering, typical of early doctoral students. These practices are also
applicable outside of doctoral training environments, and we believe they
should form a key part of postgraduate training schemes more generally in the
life sciences.
| no_new_dataset | 0.940079 |
2403.16182 | Yifei Huang | Yifei Huang, Guo Chen, Jilan Xu, Mingfang Zhang, Lijin Yang, Baoqi
Pei, Hongjie Zhang, Lu Dong, Yali Wang, Limin Wang, Yu Qiao | EgoExoLearn: A Dataset for Bridging Asynchronous Ego- and Exo-centric
View of Procedural Activities in Real World | CVPR 2024 | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Being able to map the activities of others into one's own point of view is
one fundamental human skill even from a very early age. Taking a step toward
understanding this human ability, we introduce EgoExoLearn, a large-scale
dataset that emulates the human demonstration following process, in which
individuals record egocentric videos as they execute tasks guided by
demonstration videos. Focusing on the potential applications in daily
assistance and professional support, EgoExoLearn contains egocentric and
demonstration video data spanning 120 hours captured in daily life scenarios
and specialized laboratories. Along with the videos we record high-quality gaze
data and provide detailed multimodal annotations, formulating a playground for
modeling the human ability to bridge asynchronous procedural actions from
different viewpoints. To this end, we present benchmarks such as cross-view
association, cross-view action planning, and cross-view referenced skill
assessment, along with detailed analysis. We expect EgoExoLearn can serve as an
important resource for bridging the actions across views, thus paving the way
for creating AI agents capable of seamlessly learning by observing humans in
the real world. Code and data can be found at:
https://github.com/OpenGVLab/EgoExoLearn
| [
{
"version": "v1",
"created": "Sun, 24 Mar 2024 15:00:44 GMT"
},
{
"version": "v2",
"created": "Wed, 5 Jun 2024 09:44:52 GMT"
},
{
"version": "v3",
"created": "Thu, 6 Mar 2025 02:46:51 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Huang",
"Yifei",
""
],
[
"Chen",
"Guo",
""
],
[
"Xu",
"Jilan",
""
],
[
"Zhang",
"Mingfang",
""
],
[
"Yang",
"Lijin",
""
],
[
"Pei",
"Baoqi",
""
],
[
"Zhang",
"Hongjie",
""
],
[
"Dong",
"Lu",
""
],
[
"Wang",
"Yali",
""
],
[
"Wang",
"Limin",
""
],
[
"Qiao",
"Yu",
""
]
]
| TITLE: EgoExoLearn: A Dataset for Bridging Asynchronous Ego- and Exo-centric
View of Procedural Activities in Real World
ABSTRACT: Being able to map the activities of others into one's own point of view is
one fundamental human skill even from a very early age. Taking a step toward
understanding this human ability, we introduce EgoExoLearn, a large-scale
dataset that emulates the human demonstration following process, in which
individuals record egocentric videos as they execute tasks guided by
demonstration videos. Focusing on the potential applications in daily
assistance and professional support, EgoExoLearn contains egocentric and
demonstration video data spanning 120 hours captured in daily life scenarios
and specialized laboratories. Along with the videos we record high-quality gaze
data and provide detailed multimodal annotations, formulating a playground for
modeling the human ability to bridge asynchronous procedural actions from
different viewpoints. To this end, we present benchmarks such as cross-view
association, cross-view action planning, and cross-view referenced skill
assessment, along with detailed analysis. We expect EgoExoLearn can serve as an
important resource for bridging the actions across views, thus paving the way
for creating AI agents capable of seamlessly learning by observing humans in
the real world. Code and data can be found at:
https://github.com/OpenGVLab/EgoExoLearn
| new_dataset | 0.96707 |
2404.05569 | Hao Li | Shen Gao, Hao Li, Chengrui Huang, Quan Tu, Zhiliang Tian, Minlie
Huang, Shuo Shang | 360$^\circ$REA: Towards A Reusable Experience Accumulation with
360{\deg} Assessment for Multi-Agent System | null | null | null | null | cs.AI cs.CL cs.MA | http://creativecommons.org/licenses/by/4.0/ | Large language model agents have demonstrated remarkable advancements across
various complex tasks. Recent works focus on optimizing the agent team or
employing self-reflection to iteratively solve complex tasks. Since these
agents are all based on the same LLM, only conducting self-evaluation or
removing underperforming agents does not substantively enhance the capability
of the agents. We argue that a comprehensive evaluation and accumulating
experience from evaluation feedback is an effective approach to improving
system performance. In this paper, we propose Reusable Experience Accumulation
with 360$^\circ$ Assessment (360$^\circ$REA), a hierarchical multi-agent
framework inspired by corporate organizational practices. The framework employs
a novel 360$^\circ$ performance assessment method for multi-perspective
performance evaluation with fine-grained assessment. To enhance the capability
of agents in addressing complex tasks, we introduce dual-level experience pool
for agents to accumulate experience through fine-grained assessment. Extensive
experiments on complex task datasets demonstrate the effectiveness of
360$^\circ$REA.
| [
{
"version": "v1",
"created": "Mon, 8 Apr 2024 14:43:13 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Jun 2024 11:42:10 GMT"
},
{
"version": "v3",
"created": "Thu, 6 Mar 2025 12:54:37 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Gao",
"Shen",
""
],
[
"Li",
"Hao",
""
],
[
"Huang",
"Chengrui",
""
],
[
"Tu",
"Quan",
""
],
[
"Tian",
"Zhiliang",
""
],
[
"Huang",
"Minlie",
""
],
[
"Shang",
"Shuo",
""
]
]
| TITLE: 360$^\circ$REA: Towards A Reusable Experience Accumulation with
360{\deg} Assessment for Multi-Agent System
ABSTRACT: Large language model agents have demonstrated remarkable advancements across
various complex tasks. Recent works focus on optimizing the agent team or
employing self-reflection to iteratively solve complex tasks. Since these
agents are all based on the same LLM, only conducting self-evaluation or
removing underperforming agents does not substantively enhance the capability
of the agents. We argue that a comprehensive evaluation and accumulating
experience from evaluation feedback is an effective approach to improving
system performance. In this paper, we propose Reusable Experience Accumulation
with 360$^\circ$ Assessment (360$^\circ$REA), a hierarchical multi-agent
framework inspired by corporate organizational practices. The framework employs
a novel 360$^\circ$ performance assessment method for multi-perspective
performance evaluation with fine-grained assessment. To enhance the capability
of agents in addressing complex tasks, we introduce dual-level experience pool
for agents to accumulate experience through fine-grained assessment. Extensive
experiments on complex task datasets demonstrate the effectiveness of
360$^\circ$REA.
| no_new_dataset | 0.949201 |
2405.01924 | Jiawei Zhou | Jiawei Zhou, Li Dong, Furu Wei, Lei Chen | Semi-Parametric Retrieval via Binary Bag-of-Tokens Index | null | null | null | null | cs.CL cs.AI cs.IR | http://creativecommons.org/licenses/by/4.0/ | Information retrieval has transitioned from standalone systems into essential
components across broader applications, with indexing efficiency,
cost-effectiveness, and freshness becoming increasingly critical yet often
overlooked. In this paper, we introduce SemI-parametric Disentangled Retrieval
(SiDR), a bi-encoder retrieval framework that decouples retrieval index from
neural parameters to enable efficient, low-cost, and parameter-agnostic
indexing for emerging use cases. Specifically, in addition to using embeddings
as indexes like existing neural retrieval methods, SiDR supports a
non-parametric tokenization index for search, achieving BM25-like indexing
complexity with significantly better effectiveness. Our comprehensive
evaluation across 16 retrieval benchmarks demonstrates that SiDR outperforms
both neural and term-based retrieval baselines under the same indexing
workload: (i) When using an embedding-based index, SiDR exceeds the performance
of conventional neural retrievers while maintaining similar training
complexity; (ii) When using a tokenization-based index, SiDR drastically
reduces indexing cost and time, matching the complexity of traditional
term-based retrieval, while consistently outperforming BM25 on all in-domain
datasets; (iii) Additionally, we introduce a late parametric mechanism that
matches BM25 index preparation time while outperforming other neural retrieval
baselines in effectiveness.
| [
{
"version": "v1",
"created": "Fri, 3 May 2024 08:34:13 GMT"
},
{
"version": "v2",
"created": "Thu, 6 Mar 2025 10:39:48 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Zhou",
"Jiawei",
""
],
[
"Dong",
"Li",
""
],
[
"Wei",
"Furu",
""
],
[
"Chen",
"Lei",
""
]
]
| TITLE: Semi-Parametric Retrieval via Binary Bag-of-Tokens Index
ABSTRACT: Information retrieval has transitioned from standalone systems into essential
components across broader applications, with indexing efficiency,
cost-effectiveness, and freshness becoming increasingly critical yet often
overlooked. In this paper, we introduce SemI-parametric Disentangled Retrieval
(SiDR), a bi-encoder retrieval framework that decouples retrieval index from
neural parameters to enable efficient, low-cost, and parameter-agnostic
indexing for emerging use cases. Specifically, in addition to using embeddings
as indexes like existing neural retrieval methods, SiDR supports a
non-parametric tokenization index for search, achieving BM25-like indexing
complexity with significantly better effectiveness. Our comprehensive
evaluation across 16 retrieval benchmarks demonstrates that SiDR outperforms
both neural and term-based retrieval baselines under the same indexing
workload: (i) When using an embedding-based index, SiDR exceeds the performance
of conventional neural retrievers while maintaining similar training
complexity; (ii) When using a tokenization-based index, SiDR drastically
reduces indexing cost and time, matching the complexity of traditional
term-based retrieval, while consistently outperforming BM25 on all in-domain
datasets; (iii) Additionally, we introduce a late parametric mechanism that
matches BM25 index preparation time while outperforming other neural retrieval
baselines in effectiveness.
| no_new_dataset | 0.949716 |
2405.02318 | Abhinav Lalwani | Abhinav Lalwani, Tasha Kim, Lovish Chopra, Christopher Hahn, Zhijing
Jin and Mrinmaya Sachan | Autoformalizing Natural Language to First-Order Logic: A Case Study in
Logical Fallacy Detection | null | null | null | null | cs.CL cs.AI cs.LG cs.LO | http://creativecommons.org/licenses/by/4.0/ | Translating natural language into formal language such as First-Order Logic
(FOL) is a foundational challenge in NLP with wide-ranging applications in
automated reasoning, misinformation tracking, and knowledge validation. In this
paper, we introduce Natural Language to First-Order Logic (NL2FOL), a framework
to autoformalize natural language to FOL step by step using Large Language
Models (LLMs). Our approach addresses key challenges in this translation
process, including the integration of implicit background knowledge. By
leveraging structured representations generated by NL2FOL, we use
Satisfiability Modulo Theory (SMT) solvers to reason about the logical validity
of natural language statements. We present logical fallacy detection as a case
study to evaluate the efficacy of NL2FOL. Being neurosymbolic, our approach
also provides interpretable insights into the reasoning process and
demonstrates robustness without requiring model fine-tuning or labeled training
data. Our framework achieves strong performance on multiple datasets. On the
LOGIC dataset, NL2FOL achieves an F1-score of 78%, while generalizing
effectively to the LOGICCLIMATE dataset with an F1-score of 80%.
| [
{
"version": "v1",
"created": "Thu, 18 Apr 2024 00:20:48 GMT"
},
{
"version": "v2",
"created": "Mon, 3 Mar 2025 00:38:48 GMT"
},
{
"version": "v3",
"created": "Thu, 6 Mar 2025 07:29:44 GMT"
}
]
| 2025-03-07T00:00:00 | [
[
"Lalwani",
"Abhinav",
""
],
[
"Kim",
"Tasha",
""
],
[
"Chopra",
"Lovish",
""
],
[
"Hahn",
"Christopher",
""
],
[
"Jin",
"Zhijing",
""
],
[
"Sachan",
"Mrinmaya",
""
]
]
| TITLE: Autoformalizing Natural Language to First-Order Logic: A Case Study in
Logical Fallacy Detection
ABSTRACT: Translating natural language into formal language such as First-Order Logic
(FOL) is a foundational challenge in NLP with wide-ranging applications in
automated reasoning, misinformation tracking, and knowledge validation. In this
paper, we introduce Natural Language to First-Order Logic (NL2FOL), a framework
to autoformalize natural language to FOL step by step using Large Language
Models (LLMs). Our approach addresses key challenges in this translation
process, including the integration of implicit background knowledge. By
leveraging structured representations generated by NL2FOL, we use
Satisfiability Modulo Theory (SMT) solvers to reason about the logical validity
of natural language statements. We present logical fallacy detection as a case
study to evaluate the efficacy of NL2FOL. Being neurosymbolic, our approach
also provides interpretable insights into the reasoning process and
demonstrates robustness without requiring model fine-tuning or labeled training
data. Our framework achieves strong performance on multiple datasets. On the
LOGIC dataset, NL2FOL achieves an F1-score of 78%, while generalizing
effectively to the LOGICCLIMATE dataset with an F1-score of 80%.
| no_new_dataset | 0.94625 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.