id
large_stringlengths 9
16
| submitter
large_stringlengths 3
64
⌀ | authors
large_stringlengths 5
20.2k
| title
large_stringlengths 5
245
| comments
large_stringlengths 1
482
⌀ | journal-ref
large_stringlengths 4
382
⌀ | doi
large_stringlengths 9
151
⌀ | report-no
large_stringlengths 2
509
⌀ | categories
large_stringlengths 5
78
| license
large_stringclasses 9
values | abstract
large_stringlengths 83
3.55k
| versions
large listlengths 1
21
| update_date
timestamp[ms]date 2007-05-23 00:00:00
2025-05-30 00:00:00
| authors_parsed
large listlengths 1
1.35k
| prompt
large_stringlengths 166
3.64k
| label
stringclasses 2
values | prob
float64 0.5
0.73
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2304.14190 | Mingyu Ding | Chao Xia, Chenfeng Xu, Patrick Rim, Mingyu Ding, Nanning Zheng, Kurt
Keutzer, Masayoshi Tomizuka, Wei Zhan | Quadric Representations for LiDAR Odometry, Mapping and Localization | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Current LiDAR odometry, mapping and localization methods leverage point-wise
representations of 3D scenes and achieve high accuracy in autonomous driving
tasks. However, the space-inefficiency of methods that use point-wise
representations limits their development and usage in practical applications.
In particular, scan-submap matching and global map representation methods are
restricted by the inefficiency of nearest neighbor searching (NNS) for
large-volume point clouds. To improve space-time efficiency, we propose a novel
method of describing scenes using quadric surfaces, which are far more compact
representations of 3D objects than conventional point clouds. In contrast to
point cloud-based methods, our quadric representation-based method decomposes a
3D scene into a collection of sparse quadric patches, which improves storage
efficiency and avoids the slow point-wise NNS process. Our method first
segments a given point cloud into patches and fits each of them to a quadric
implicit function. Each function is then coupled with other geometric
descriptors of the patch, such as its center position and covariance matrix.
Collectively, these patch representations fully describe a 3D scene, which can
be used in place of the original point cloud and employed in LiDAR odometry,
mapping and localization algorithms. We further design a novel incremental
growing method for quadric representations, which eliminates the need to
repeatedly re-fit quadric surfaces from the original point cloud. Extensive
odometry, mapping and localization experiments on large-volume point clouds in
the KITTI and UrbanLoco datasets demonstrate that our method maintains low
latency and memory utility while achieving competitive, and even superior,
accuracy.
| [
{
"version": "v1",
"created": "Thu, 27 Apr 2023 13:52:01 GMT"
}
] | 2023-04-28T00:00:00 | [
[
"Xia",
"Chao",
""
],
[
"Xu",
"Chenfeng",
""
],
[
"Rim",
"Patrick",
""
],
[
"Ding",
"Mingyu",
""
],
[
"Zheng",
"Nanning",
""
],
[
"Keutzer",
"Kurt",
""
],
[
"Tomizuka",
"Masayoshi",
""
],
[
"Zhan",
"Wei",
""
]
] | TITLE: Quadric Representations for LiDAR Odometry, Mapping and Localization
ABSTRACT: Current LiDAR odometry, mapping and localization methods leverage point-wise
representations of 3D scenes and achieve high accuracy in autonomous driving
tasks. However, the space-inefficiency of methods that use point-wise
representations limits their development and usage in practical applications.
In particular, scan-submap matching and global map representation methods are
restricted by the inefficiency of nearest neighbor searching (NNS) for
large-volume point clouds. To improve space-time efficiency, we propose a novel
method of describing scenes using quadric surfaces, which are far more compact
representations of 3D objects than conventional point clouds. In contrast to
point cloud-based methods, our quadric representation-based method decomposes a
3D scene into a collection of sparse quadric patches, which improves storage
efficiency and avoids the slow point-wise NNS process. Our method first
segments a given point cloud into patches and fits each of them to a quadric
implicit function. Each function is then coupled with other geometric
descriptors of the patch, such as its center position and covariance matrix.
Collectively, these patch representations fully describe a 3D scene, which can
be used in place of the original point cloud and employed in LiDAR odometry,
mapping and localization algorithms. We further design a novel incremental
growing method for quadric representations, which eliminates the need to
repeatedly re-fit quadric surfaces from the original point cloud. Extensive
odometry, mapping and localization experiments on large-volume point clouds in
the KITTI and UrbanLoco datasets demonstrate that our method maintains low
latency and memory utility while achieving competitive, and even superior,
accuracy.
| no_new_dataset | 0.711005 |
2304.14204 | Bingqian Lin | Bingqian Lin, Zicong Chen, Mingjie Li, Haokun Lin, Hang Xu, Yi Zhu,
Jianzhuang Liu, Wenjia Cai, Lei Yang, Shen Zhao, Chenfei Wu, Ling Chen,
Xiaojun Chang, Yi Yang, Lei Xing, Xiaodan Liang | Towards Medical Artificial General Intelligence via Knowledge-Enhanced
Multimodal Pretraining | Project page: https://github.com/chenzcv7/MOTOR | null | null | null | cs.AI cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Medical artificial general intelligence (MAGI) enables one foundation model
to solve different medical tasks, which is very practical in the medical
domain. It can significantly reduce the requirement of large amounts of
task-specific data by sufficiently sharing medical knowledge among different
tasks. However, due to the challenges of designing strongly generalizable
models with limited and complex medical data, most existing approaches tend to
develop task-specific models. To take a step towards MAGI, we propose a new
paradigm called Medical-knOwledge-enhanced mulTimOdal pretRaining (MOTOR). In
MOTOR, we combine two kinds of basic medical knowledge, i.e., general and
specific knowledge, in a complementary manner to boost the general pretraining
process. As a result, the foundation model with comprehensive basic knowledge
can learn compact representations from pretraining radiographic data for better
cross-modal alignment. MOTOR unifies the understanding and generation, which
are two kinds of core intelligence of an AI system, into a single medical
foundation model, to flexibly handle more diverse medical tasks. To enable a
comprehensive evaluation and facilitate further research, we construct a
medical multimodal benchmark including a wide range of downstream tasks, such
as chest x-ray report generation and medical visual question answering.
Extensive experiments on our benchmark show that MOTOR obtains promising
results through simple task-oriented adaptation. The visualization shows that
the injected knowledge successfully highlights key information in the medical
data, demonstrating the excellent interpretability of MOTOR. Our MOTOR
successfully mimics the human practice of fulfilling a "medical student" to
accelerate the process of becoming a "specialist". We believe that our work
makes a significant stride in realizing MAGI.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 01:26:19 GMT"
}
] | 2023-04-28T00:00:00 | [
[
"Lin",
"Bingqian",
""
],
[
"Chen",
"Zicong",
""
],
[
"Li",
"Mingjie",
""
],
[
"Lin",
"Haokun",
""
],
[
"Xu",
"Hang",
""
],
[
"Zhu",
"Yi",
""
],
[
"Liu",
"Jianzhuang",
""
],
[
"Cai",
"Wenjia",
""
],
[
"Yang",
"Lei",
""
],
[
"Zhao",
"Shen",
""
],
[
"Wu",
"Chenfei",
""
],
[
"Chen",
"Ling",
""
],
[
"Chang",
"Xiaojun",
""
],
[
"Yang",
"Yi",
""
],
[
"Xing",
"Lei",
""
],
[
"Liang",
"Xiaodan",
""
]
] | TITLE: Towards Medical Artificial General Intelligence via Knowledge-Enhanced
Multimodal Pretraining
ABSTRACT: Medical artificial general intelligence (MAGI) enables one foundation model
to solve different medical tasks, which is very practical in the medical
domain. It can significantly reduce the requirement of large amounts of
task-specific data by sufficiently sharing medical knowledge among different
tasks. However, due to the challenges of designing strongly generalizable
models with limited and complex medical data, most existing approaches tend to
develop task-specific models. To take a step towards MAGI, we propose a new
paradigm called Medical-knOwledge-enhanced mulTimOdal pretRaining (MOTOR). In
MOTOR, we combine two kinds of basic medical knowledge, i.e., general and
specific knowledge, in a complementary manner to boost the general pretraining
process. As a result, the foundation model with comprehensive basic knowledge
can learn compact representations from pretraining radiographic data for better
cross-modal alignment. MOTOR unifies the understanding and generation, which
are two kinds of core intelligence of an AI system, into a single medical
foundation model, to flexibly handle more diverse medical tasks. To enable a
comprehensive evaluation and facilitate further research, we construct a
medical multimodal benchmark including a wide range of downstream tasks, such
as chest x-ray report generation and medical visual question answering.
Extensive experiments on our benchmark show that MOTOR obtains promising
results through simple task-oriented adaptation. The visualization shows that
the injected knowledge successfully highlights key information in the medical
data, demonstrating the excellent interpretability of MOTOR. Our MOTOR
successfully mimics the human practice of fulfilling a "medical student" to
accelerate the process of becoming a "specialist". We believe that our work
makes a significant stride in realizing MAGI.
| no_new_dataset | 0.706823 |
2304.14216 | Alexander Lobbe | Bertrand Chapron, Dan Crisan, Darryl Holm, Oana Lang, Alexander Lobbe,
Etienne M\'emin | Comparison of Stochastic Parametrization Schemes using Data Assimilation
on Triad Models | Submitted to Proceedings of the 3rd Stochastic Transport in Upper
Ocean Dynamics (STUOD) Annual Workshop, Springer Verlag | null | null | null | math.DS cs.NA math.NA math.PR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In recent years, stochastic parametrizations have been ubiquitous in
modelling uncertainty in fluid dynamics models. One source of model uncertainty
comes from the coarse graining of the fine-scale data and is in common usage in
computational simulations at coarser scales. In this paper, we look at two such
stochastic parametrizations: the Stochastic Advection by Lie Transport (SALT)
parametrization introduced by Holm and the Location Uncertainty (LU)
parametrization introduced by M\'emin. Whilst both parametrizations are
available for full-scale models, we study their reduced order versions obtained
by projecting them on a complex vector Fourier mode triad of eigenfunctions of
the curl. Remarkably, these two parametrizations lead to the same reduced order
model, which we term the helicity-preserving stochastic triad (HST). This
reduced order model is then compared with an alternative model which preserves
the energy of the system, and which is termed the energy preserving stochastic
triad (EST). These low-dimensional models are ideal benchmark models for
testing new Data Assimilation algorithms: they are easy to implement, exhibit
diverse behaviours depending on the choice of the coefficients and come with
natural physical properties such as the conservation of energy and helicity.
| [
{
"version": "v1",
"created": "Thu, 27 Apr 2023 14:24:48 GMT"
}
] | 2023-04-28T00:00:00 | [
[
"Chapron",
"Bertrand",
""
],
[
"Crisan",
"Dan",
""
],
[
"Holm",
"Darryl",
""
],
[
"Lang",
"Oana",
""
],
[
"Lobbe",
"Alexander",
""
],
[
"Mémin",
"Etienne",
""
]
] | TITLE: Comparison of Stochastic Parametrization Schemes using Data Assimilation
on Triad Models
ABSTRACT: In recent years, stochastic parametrizations have been ubiquitous in
modelling uncertainty in fluid dynamics models. One source of model uncertainty
comes from the coarse graining of the fine-scale data and is in common usage in
computational simulations at coarser scales. In this paper, we look at two such
stochastic parametrizations: the Stochastic Advection by Lie Transport (SALT)
parametrization introduced by Holm and the Location Uncertainty (LU)
parametrization introduced by M\'emin. Whilst both parametrizations are
available for full-scale models, we study their reduced order versions obtained
by projecting them on a complex vector Fourier mode triad of eigenfunctions of
the curl. Remarkably, these two parametrizations lead to the same reduced order
model, which we term the helicity-preserving stochastic triad (HST). This
reduced order model is then compared with an alternative model which preserves
the energy of the system, and which is termed the energy preserving stochastic
triad (EST). These low-dimensional models are ideal benchmark models for
testing new Data Assimilation algorithms: they are easy to implement, exhibit
diverse behaviours depending on the choice of the coefficients and come with
natural physical properties such as the conservation of energy and helicity.
| no_new_dataset | 0.710848 |
2304.14224 | Liang Luo | Jiutian Zhao, Liang Luo, Hao Wang | Self-discipline on multiple channels | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Self-distillation relies on its own information to improve the generalization
ability of the model and has a bright future. Existing self-distillation
methods either require additional models, model modification, or batch size
expansion for training, which increases the difficulty of use, memory
consumption, and computational cost. This paper developed Self-discipline on
multiple channels(SMC), which combines consistency regularization with
self-distillation using the concept of multiple channels. Conceptually, SMC
consists of two steps: 1) each channel data is simultaneously passed through
the model to obtain its corresponding soft label, and 2) the soft label saved
in the previous step is read together with the soft label obtained from the
current channel data through the model to calculate the loss function. SMC uses
consistent regularization and self-distillation to improve the generalization
ability of the model and the robustness of the model to noisy labels. We named
the SMC containing only two channels as SMC-2. Comparative experimental results
on both datasets show that SMC-2 outperforms Label Smoothing Regularizaion and
Self-distillation From The Last Mini-batch on all models, and outperforms the
state-of-the-art Sharpness-Aware Minimization method on 83% of the
models.Compatibility of SMC-2 and data augmentation experimental results show
that using both SMC-2 and data augmentation improves the generalization ability
of the model between 0.28% and 1.80% compared to using only data augmentation.
Ultimately, the results of the label noise interference experiments show that
SMC-2 curbs the tendency that the model's generalization ability decreases in
the late training period due to the interference of label noise. The code is
available at
https://github.com/JiuTiannn/SMC-Self-discipline-on-multiple-channels.
| [
{
"version": "v1",
"created": "Thu, 27 Apr 2023 14:34:41 GMT"
}
] | 2023-04-28T00:00:00 | [
[
"Zhao",
"Jiutian",
""
],
[
"Luo",
"Liang",
""
],
[
"Wang",
"Hao",
""
]
] | TITLE: Self-discipline on multiple channels
ABSTRACT: Self-distillation relies on its own information to improve the generalization
ability of the model and has a bright future. Existing self-distillation
methods either require additional models, model modification, or batch size
expansion for training, which increases the difficulty of use, memory
consumption, and computational cost. This paper developed Self-discipline on
multiple channels(SMC), which combines consistency regularization with
self-distillation using the concept of multiple channels. Conceptually, SMC
consists of two steps: 1) each channel data is simultaneously passed through
the model to obtain its corresponding soft label, and 2) the soft label saved
in the previous step is read together with the soft label obtained from the
current channel data through the model to calculate the loss function. SMC uses
consistent regularization and self-distillation to improve the generalization
ability of the model and the robustness of the model to noisy labels. We named
the SMC containing only two channels as SMC-2. Comparative experimental results
on both datasets show that SMC-2 outperforms Label Smoothing Regularizaion and
Self-distillation From The Last Mini-batch on all models, and outperforms the
state-of-the-art Sharpness-Aware Minimization method on 83% of the
models.Compatibility of SMC-2 and data augmentation experimental results show
that using both SMC-2 and data augmentation improves the generalization ability
of the model between 0.28% and 1.80% compared to using only data augmentation.
Ultimately, the results of the label noise interference experiments show that
SMC-2 curbs the tendency that the model's generalization ability decreases in
the late training period due to the interference of label noise. The code is
available at
https://github.com/JiuTiannn/SMC-Self-discipline-on-multiple-channels.
| no_new_dataset | 0.710434 |
2304.14241 | Egil R{\o}nningstad | Egil R{\o}nningstad and Erik Velldal and Lilja {\O}vrelid | Entity-Level Sentiment Analysis (ELSA): An exploratory task survey | null | Proceedings of the 29th International Conference on Computational
Linguistics, 2022, pages 6773-6783 | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper explores the task of identifying the overall sentiment expressed
towards volitional entities (persons and organizations) in a document -- what
we refer to as Entity-Level Sentiment Analysis (ELSA). While identifying
sentiment conveyed towards an entity is well researched for shorter texts like
tweets, we find little to no research on this specific task for longer texts
with multiple mentions and opinions towards the same entity. This lack of
research would be understandable if ELSA can be derived from existing tasks and
models. To assess this, we annotate a set of professional reviews for their
overall sentiment towards each volitional entity in the text. We sample from
data already annotated for document-level, sentence-level, and target-level
sentiment in a multi-domain review corpus, and our results indicate that there
is no single proxy task that provides this overall sentiment we seek for the
entities at a satisfactory level of performance. We present a suite of
experiments aiming to assess the contribution towards ELSA provided by
document-, sentence-, and target-level sentiment analysis, and provide a
discussion of their shortcomings. We show that sentiment in our dataset is
expressed not only with an entity mention as target, but also towards targets
with a sentiment-relevant relation to a volitional entity. In our data, these
relations extend beyond anaphoric coreference resolution, and our findings call
for further research of the topic. Finally, we also present a survey of
previous relevant work.
| [
{
"version": "v1",
"created": "Thu, 27 Apr 2023 15:01:20 GMT"
}
] | 2023-04-28T00:00:00 | [
[
"Rønningstad",
"Egil",
""
],
[
"Velldal",
"Erik",
""
],
[
"Øvrelid",
"Lilja",
""
]
] | TITLE: Entity-Level Sentiment Analysis (ELSA): An exploratory task survey
ABSTRACT: This paper explores the task of identifying the overall sentiment expressed
towards volitional entities (persons and organizations) in a document -- what
we refer to as Entity-Level Sentiment Analysis (ELSA). While identifying
sentiment conveyed towards an entity is well researched for shorter texts like
tweets, we find little to no research on this specific task for longer texts
with multiple mentions and opinions towards the same entity. This lack of
research would be understandable if ELSA can be derived from existing tasks and
models. To assess this, we annotate a set of professional reviews for their
overall sentiment towards each volitional entity in the text. We sample from
data already annotated for document-level, sentence-level, and target-level
sentiment in a multi-domain review corpus, and our results indicate that there
is no single proxy task that provides this overall sentiment we seek for the
entities at a satisfactory level of performance. We present a suite of
experiments aiming to assess the contribution towards ELSA provided by
document-, sentence-, and target-level sentiment analysis, and provide a
discussion of their shortcomings. We show that sentiment in our dataset is
expressed not only with an entity mention as target, but also towards targets
with a sentiment-relevant relation to a volitional entity. In our data, these
relations extend beyond anaphoric coreference resolution, and our findings call
for further research of the topic. Finally, we also present a survey of
previous relevant work.
| new_dataset | 0.716051 |
2304.14275 | Joseph G Lambourne | Peter Meltzer, Joseph G. Lambourne, Daniele Grandi | What's in a Name? Evaluating Assembly-Part Semantic Knowledge in
Language Models through User-Provided Names in CAD Files | null | null | null | null | cs.CL cs.LG | http://creativecommons.org/licenses/by-sa/4.0/ | Semantic knowledge of part-part and part-whole relationships in assemblies is
useful for a variety of tasks from searching design repositories to the
construction of engineering knowledge bases. In this work we propose that the
natural language names designers use in Computer Aided Design (CAD) software
are a valuable source of such knowledge, and that Large Language Models (LLMs)
contain useful domain-specific information for working with this data as well
as other CAD and engineering-related tasks.
In particular we extract and clean a large corpus of natural language part,
feature and document names and use this to quantitatively demonstrate that a
pre-trained language model can outperform numerous benchmarks on three
self-supervised tasks, without ever having seen this data before. Moreover, we
show that fine-tuning on the text data corpus further boosts the performance on
all tasks, thus demonstrating the value of the text data which until now has
been largely ignored. We also identify key limitations to using LLMs with text
data alone, and our findings provide a strong motivation for further work into
multi-modal text-geometry models.
To aid and encourage further work in this area we make all our data and code
publicly available.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 12:30:01 GMT"
}
] | 2023-04-28T00:00:00 | [
[
"Meltzer",
"Peter",
""
],
[
"Lambourne",
"Joseph G.",
""
],
[
"Grandi",
"Daniele",
""
]
] | TITLE: What's in a Name? Evaluating Assembly-Part Semantic Knowledge in
Language Models through User-Provided Names in CAD Files
ABSTRACT: Semantic knowledge of part-part and part-whole relationships in assemblies is
useful for a variety of tasks from searching design repositories to the
construction of engineering knowledge bases. In this work we propose that the
natural language names designers use in Computer Aided Design (CAD) software
are a valuable source of such knowledge, and that Large Language Models (LLMs)
contain useful domain-specific information for working with this data as well
as other CAD and engineering-related tasks.
In particular we extract and clean a large corpus of natural language part,
feature and document names and use this to quantitatively demonstrate that a
pre-trained language model can outperform numerous benchmarks on three
self-supervised tasks, without ever having seen this data before. Moreover, we
show that fine-tuning on the text data corpus further boosts the performance on
all tasks, thus demonstrating the value of the text data which until now has
been largely ignored. We also identify key limitations to using LLMs with text
data alone, and our findings provide a strong motivation for further work into
multi-modal text-geometry models.
To aid and encourage further work in this area we make all our data and code
publicly available.
| no_new_dataset | 0.711186 |
2304.14276 | Steffen Herbold | Steffen Herbold, Annette Hautli-Janisz, Ute Heuer, Zlata Kikteva,
Alexander Trautsch | AI, write an essay for me: A large-scale comparison of human-written
versus ChatGPT-generated essays | Submitted | null | null | null | cs.CL cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Background: Recently, ChatGPT and similar generative AI models have attracted
hundreds of millions of users and become part of the public discourse. Many
believe that such models will disrupt society and will result in a significant
change in the education system and information generation in the future. So
far, this belief is based on either colloquial evidence or benchmarks from the
owners of the models -- both lack scientific rigour.
Objective: Through a large-scale study comparing human-written versus
ChatGPT-generated argumentative student essays, we systematically assess the
quality of the AI-generated content.
Methods: A large corpus of essays was rated using standard criteria by a
large number of human experts (teachers). We augment the analysis with a
consideration of the linguistic characteristics of the generated essays.
Results: Our results demonstrate that ChatGPT generates essays that are rated
higher for quality than human-written essays. The writing style of the AI
models exhibits linguistic characteristics that are different from those of the
human-written essays, e.g., it is characterized by fewer discourse and
epistemic markers, but more nominalizations and greater lexical diversity.
Conclusions: Our results clearly demonstrate that models like ChatGPT
outperform humans in generating argumentative essays. Since the technology is
readily available for anyone to use, educators must act immediately. We must
re-invent homework and develop teaching concepts that utilize these AI models
in the same way as math utilized the calculator: teach the general concepts
first and then use AI tools to free up time for other learning objectives.
| [
{
"version": "v1",
"created": "Mon, 24 Apr 2023 12:58:28 GMT"
}
] | 2023-04-28T00:00:00 | [
[
"Herbold",
"Steffen",
""
],
[
"Hautli-Janisz",
"Annette",
""
],
[
"Heuer",
"Ute",
""
],
[
"Kikteva",
"Zlata",
""
],
[
"Trautsch",
"Alexander",
""
]
] | TITLE: AI, write an essay for me: A large-scale comparison of human-written
versus ChatGPT-generated essays
ABSTRACT: Background: Recently, ChatGPT and similar generative AI models have attracted
hundreds of millions of users and become part of the public discourse. Many
believe that such models will disrupt society and will result in a significant
change in the education system and information generation in the future. So
far, this belief is based on either colloquial evidence or benchmarks from the
owners of the models -- both lack scientific rigour.
Objective: Through a large-scale study comparing human-written versus
ChatGPT-generated argumentative student essays, we systematically assess the
quality of the AI-generated content.
Methods: A large corpus of essays was rated using standard criteria by a
large number of human experts (teachers). We augment the analysis with a
consideration of the linguistic characteristics of the generated essays.
Results: Our results demonstrate that ChatGPT generates essays that are rated
higher for quality than human-written essays. The writing style of the AI
models exhibits linguistic characteristics that are different from those of the
human-written essays, e.g., it is characterized by fewer discourse and
epistemic markers, but more nominalizations and greater lexical diversity.
Conclusions: Our results clearly demonstrate that models like ChatGPT
outperform humans in generating argumentative essays. Since the technology is
readily available for anyone to use, educators must act immediately. We must
re-invent homework and develop teaching concepts that utilize these AI models
in the same way as math utilized the calculator: teach the general concepts
first and then use AI tools to free up time for other learning objectives.
| no_new_dataset | 0.706418 |
2304.14281 | Michalis Lazarou Mr | Michalis Lazarou, Yannis Avrithis, Tania Stathaki | Adaptive manifold for imbalanced transductive few-shot learning | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Transductive few-shot learning algorithms have showed substantially superior
performance over their inductive counterparts by leveraging the unlabeled
queries. However, the vast majority of such methods are evaluated on perfectly
class-balanced benchmarks. It has been shown that they undergo remarkable drop
in performance under a more realistic, imbalanced setting. To this end, we
propose a novel algorithm to address imbalanced transductive few-shot learning,
named Adaptive Manifold. Our method exploits the underlying manifold of the
labeled support examples and unlabeled queries by using manifold similarity to
predict the class probability distribution per query. It is parameterized by
one centroid per class as well as a set of graph-specific parameters that
determine the manifold. All parameters are optimized through a loss function
that can be tuned towards class-balanced or imbalanced distributions. The
manifold similarity shows substantial improvement over Euclidean distance,
especially in the 1-shot setting. Our algorithm outperforms or is on par with
other state of the art methods in three benchmark datasets, namely
miniImageNet, tieredImageNet and CUB, and three different backbones, namely
ResNet-18, WideResNet-28-10 and DenseNet-121. In certain cases, our algorithm
outperforms the previous state of the art by as much as 4.2%.
| [
{
"version": "v1",
"created": "Thu, 27 Apr 2023 15:42:49 GMT"
}
] | 2023-04-28T00:00:00 | [
[
"Lazarou",
"Michalis",
""
],
[
"Avrithis",
"Yannis",
""
],
[
"Stathaki",
"Tania",
""
]
] | TITLE: Adaptive manifold for imbalanced transductive few-shot learning
ABSTRACT: Transductive few-shot learning algorithms have showed substantially superior
performance over their inductive counterparts by leveraging the unlabeled
queries. However, the vast majority of such methods are evaluated on perfectly
class-balanced benchmarks. It has been shown that they undergo remarkable drop
in performance under a more realistic, imbalanced setting. To this end, we
propose a novel algorithm to address imbalanced transductive few-shot learning,
named Adaptive Manifold. Our method exploits the underlying manifold of the
labeled support examples and unlabeled queries by using manifold similarity to
predict the class probability distribution per query. It is parameterized by
one centroid per class as well as a set of graph-specific parameters that
determine the manifold. All parameters are optimized through a loss function
that can be tuned towards class-balanced or imbalanced distributions. The
manifold similarity shows substantial improvement over Euclidean distance,
especially in the 1-shot setting. Our algorithm outperforms or is on par with
other state of the art methods in three benchmark datasets, namely
miniImageNet, tieredImageNet and CUB, and three different backbones, namely
ResNet-18, WideResNet-28-10 and DenseNet-121. In certain cases, our algorithm
outperforms the previous state of the art by as much as 4.2%.
| no_new_dataset | 0.708377 |
2304.14321 | Lucas Pascotti Valem | Lucas Pascotti Valem and Daniel Carlos Guimar\~aes Pedronette | Person Re-ID through Unsupervised Hypergraph Rank Selection and Fusion | null | Image and Vision Computing, 123, 104473 (2022) | 10.1016/j.imavis.2022.104473 | null | cs.CV cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Person Re-ID has been gaining a lot of attention and nowadays is of
fundamental importance in many camera surveillance applications. The task
consists of identifying individuals across multiple cameras that have no
overlapping views. Most of the approaches require labeled data, which is not
always available, given the huge amount of demanded data and the difficulty of
manually assigning a class for each individual. Recently, studies have shown
that re-ranking methods are capable of achieving significant gains, especially
in the absence of labeled data. Besides that, the fusion of feature extractors
and multiple-source training is another promising research direction not
extensively exploited. We aim to fill this gap through a manifold rank
aggregation approach capable of exploiting the complementarity of different
person Re-ID rankers. In this work, we perform a completely unsupervised
selection and fusion of diverse ranked lists obtained from multiple and diverse
feature extractors. Among the contributions, this work proposes a query
performance prediction measure that models the relationship among images
considering a hypergraph structure and does not require the use of any labeled
data. Expressive gains were obtained in four datasets commonly used for person
Re-ID. We achieved results competitive to the state-of-the-art in most of the
scenarios.
| [
{
"version": "v1",
"created": "Thu, 27 Apr 2023 16:47:27 GMT"
}
] | 2023-04-28T00:00:00 | [
[
"Valem",
"Lucas Pascotti",
""
],
[
"Pedronette",
"Daniel Carlos Guimarães",
""
]
] | TITLE: Person Re-ID through Unsupervised Hypergraph Rank Selection and Fusion
ABSTRACT: Person Re-ID has been gaining a lot of attention and nowadays is of
fundamental importance in many camera surveillance applications. The task
consists of identifying individuals across multiple cameras that have no
overlapping views. Most of the approaches require labeled data, which is not
always available, given the huge amount of demanded data and the difficulty of
manually assigning a class for each individual. Recently, studies have shown
that re-ranking methods are capable of achieving significant gains, especially
in the absence of labeled data. Besides that, the fusion of feature extractors
and multiple-source training is another promising research direction not
extensively exploited. We aim to fill this gap through a manifold rank
aggregation approach capable of exploiting the complementarity of different
person Re-ID rankers. In this work, we perform a completely unsupervised
selection and fusion of diverse ranked lists obtained from multiple and diverse
feature extractors. Among the contributions, this work proposes a query
performance prediction measure that models the relationship among images
considering a hypergraph structure and does not require the use of any labeled
data. Expressive gains were obtained in four datasets commonly used for person
Re-ID. We achieved results competitive to the state-of-the-art in most of the
scenarios.
| no_new_dataset | 0.709787 |
2304.14332 | Yuheng Bu | Yuheng Bu, Harsha Vardhan Tetali, Gholamali Aminian, Miguel Rodrigues
and Gregory Wornell | On the Generalization Error of Meta Learning for the Gibbs Algorithm | Accepted at ISIT 2023 | null | null | null | cs.LG cs.IT math.IT | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We analyze the generalization ability of joint-training meta learning
algorithms via the Gibbs algorithm. Our exact characterization of the expected
meta generalization error for the meta Gibbs algorithm is based on symmetrized
KL information, which measures the dependence between all meta-training
datasets and the output parameters, including task-specific and meta
parameters. Additionally, we derive an exact characterization of the meta
generalization error for the super-task Gibbs algorithm, in terms of
conditional symmetrized KL information within the super-sample and super-task
framework introduced in Steinke and Zakynthinou (2020) and Hellstrom and Durisi
(2022) respectively. Our results also enable us to provide novel
distribution-free generalization error upper bounds for these Gibbs algorithms
applicable to meta learning.
| [
{
"version": "v1",
"created": "Thu, 27 Apr 2023 17:04:36 GMT"
}
] | 2023-04-28T00:00:00 | [
[
"Bu",
"Yuheng",
""
],
[
"Tetali",
"Harsha Vardhan",
""
],
[
"Aminian",
"Gholamali",
""
],
[
"Rodrigues",
"Miguel",
""
],
[
"Wornell",
"Gregory",
""
]
] | TITLE: On the Generalization Error of Meta Learning for the Gibbs Algorithm
ABSTRACT: We analyze the generalization ability of joint-training meta learning
algorithms via the Gibbs algorithm. Our exact characterization of the expected
meta generalization error for the meta Gibbs algorithm is based on symmetrized
KL information, which measures the dependence between all meta-training
datasets and the output parameters, including task-specific and meta
parameters. Additionally, we derive an exact characterization of the meta
generalization error for the super-task Gibbs algorithm, in terms of
conditional symmetrized KL information within the super-sample and super-task
framework introduced in Steinke and Zakynthinou (2020) and Hellstrom and Durisi
(2022) respectively. Our results also enable us to provide novel
distribution-free generalization error upper bounds for these Gibbs algorithms
applicable to meta learning.
| no_new_dataset | 0.711619 |
2304.14333 | Filip Klubi\v{c}ka | Filip Klubi\v{c}ka, Vasudevan Nedumpozhimana, John D. Kelleher | Idioms, Probing and Dangerous Things: Towards Structural Probing for
Idiomaticity in Vector Space | 9 pages, 5 tables, In proceedings of the 19th Workshop on Multiword
Expressions @ EACL2023 | null | null | null | cs.CL cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | The goal of this paper is to learn more about how idiomatic information is
structurally encoded in embeddings, using a structural probing method. We
repurpose an existing English verbal multi-word expression (MWE) dataset to
suit the probing framework and perform a comparative probing study of static
(GloVe) and contextual (BERT) embeddings. Our experiments indicate that both
encode some idiomatic information to varying degrees, but yield conflicting
evidence as to whether idiomaticity is encoded in the vector norm, leaving this
an open question. We also identify some limitations of the used dataset and
highlight important directions for future work in improving its suitability for
a probing analysis.
| [
{
"version": "v1",
"created": "Thu, 27 Apr 2023 17:06:20 GMT"
}
] | 2023-04-28T00:00:00 | [
[
"Klubička",
"Filip",
""
],
[
"Nedumpozhimana",
"Vasudevan",
""
],
[
"Kelleher",
"John D.",
""
]
] | TITLE: Idioms, Probing and Dangerous Things: Towards Structural Probing for
Idiomaticity in Vector Space
ABSTRACT: The goal of this paper is to learn more about how idiomatic information is
structurally encoded in embeddings, using a structural probing method. We
repurpose an existing English verbal multi-word expression (MWE) dataset to
suit the probing framework and perform a comparative probing study of static
(GloVe) and contextual (BERT) embeddings. Our experiments indicate that both
encode some idiomatic information to varying degrees, but yield conflicting
evidence as to whether idiomaticity is encoded in the vector norm, leaving this
an open question. We also identify some limitations of the used dataset and
highlight important directions for future work in improving its suitability for
a probing analysis.
| no_new_dataset | 0.507568 |
2304.14340 | Yichen Xie | Yichen Xie, Chenfeng Xu, Marie-Julie Rakotosaona, Patrick Rim,
Federico Tombari, Kurt Keutzer, Masayoshi Tomizuka, Wei Zhan | SparseFusion: Fusing Multi-Modal Sparse Representations for Multi-Sensor
3D Object Detection | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | By identifying four important components of existing LiDAR-camera 3D object
detection methods (LiDAR and camera candidates, transformation, and fusion
outputs), we observe that all existing methods either find dense candidates or
yield dense representations of scenes. However, given that objects occupy only
a small part of a scene, finding dense candidates and generating dense
representations is noisy and inefficient. We propose SparseFusion, a novel
multi-sensor 3D detection method that exclusively uses sparse candidates and
sparse representations. Specifically, SparseFusion utilizes the outputs of
parallel detectors in the LiDAR and camera modalities as sparse candidates for
fusion. We transform the camera candidates into the LiDAR coordinate space by
disentangling the object representations. Then, we can fuse the multi-modality
candidates in a unified 3D space by a lightweight self-attention module. To
mitigate negative transfer between modalities, we propose novel semantic and
geometric cross-modality transfer modules that are applied prior to the
modality-specific detectors. SparseFusion achieves state-of-the-art performance
on the nuScenes benchmark while also running at the fastest speed, even
outperforming methods with stronger backbones. We perform extensive experiments
to demonstrate the effectiveness and efficiency of our modules and overall
method pipeline. Our code will be made publicly available at
https://github.com/yichen928/SparseFusion.
| [
{
"version": "v1",
"created": "Thu, 27 Apr 2023 17:17:39 GMT"
}
] | 2023-04-28T00:00:00 | [
[
"Xie",
"Yichen",
""
],
[
"Xu",
"Chenfeng",
""
],
[
"Rakotosaona",
"Marie-Julie",
""
],
[
"Rim",
"Patrick",
""
],
[
"Tombari",
"Federico",
""
],
[
"Keutzer",
"Kurt",
""
],
[
"Tomizuka",
"Masayoshi",
""
],
[
"Zhan",
"Wei",
""
]
] | TITLE: SparseFusion: Fusing Multi-Modal Sparse Representations for Multi-Sensor
3D Object Detection
ABSTRACT: By identifying four important components of existing LiDAR-camera 3D object
detection methods (LiDAR and camera candidates, transformation, and fusion
outputs), we observe that all existing methods either find dense candidates or
yield dense representations of scenes. However, given that objects occupy only
a small part of a scene, finding dense candidates and generating dense
representations is noisy and inefficient. We propose SparseFusion, a novel
multi-sensor 3D detection method that exclusively uses sparse candidates and
sparse representations. Specifically, SparseFusion utilizes the outputs of
parallel detectors in the LiDAR and camera modalities as sparse candidates for
fusion. We transform the camera candidates into the LiDAR coordinate space by
disentangling the object representations. Then, we can fuse the multi-modality
candidates in a unified 3D space by a lightweight self-attention module. To
mitigate negative transfer between modalities, we propose novel semantic and
geometric cross-modality transfer modules that are applied prior to the
modality-specific detectors. SparseFusion achieves state-of-the-art performance
on the nuScenes benchmark while also running at the fastest speed, even
outperforming methods with stronger backbones. We perform extensive experiments
to demonstrate the effectiveness and efficiency of our modules and overall
method pipeline. Our code will be made publicly available at
https://github.com/yichen928/SparseFusion.
| no_new_dataset | 0.710007 |
2304.14377 | Hengyi Wang | Hengyi Wang, Jingwen Wang, Lourdes Agapito | Co-SLAM: Joint Coordinate and Sparse Parametric Encodings for Neural
Real-Time SLAM | CVPR2023. First two authors contributed equally. Project page:
https://hengyiwang.github.io/projects/CoSLAM | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present Co-SLAM, a neural RGB-D SLAM system based on a hybrid
representation, that performs robust camera tracking and high-fidelity surface
reconstruction in real time. Co-SLAM represents the scene as a multi-resolution
hash-grid to exploit its high convergence speed and ability to represent
high-frequency local features. In addition, Co-SLAM incorporates one-blob
encoding, to encourage surface coherence and completion in unobserved areas.
This joint parametric-coordinate encoding enables real-time and robust
performance by bringing the best of both worlds: fast convergence and surface
hole filling. Moreover, our ray sampling strategy allows Co-SLAM to perform
global bundle adjustment over all keyframes instead of requiring keyframe
selection to maintain a small number of active keyframes as competing neural
SLAM approaches do. Experimental results show that Co-SLAM runs at 10-17Hz and
achieves state-of-the-art scene reconstruction results, and competitive
tracking performance in various datasets and benchmarks (ScanNet, TUM, Replica,
Synthetic RGBD). Project page: https://hengyiwang.github.io/projects/CoSLAM
| [
{
"version": "v1",
"created": "Thu, 27 Apr 2023 17:46:45 GMT"
}
] | 2023-04-28T00:00:00 | [
[
"Wang",
"Hengyi",
""
],
[
"Wang",
"Jingwen",
""
],
[
"Agapito",
"Lourdes",
""
]
] | TITLE: Co-SLAM: Joint Coordinate and Sparse Parametric Encodings for Neural
Real-Time SLAM
ABSTRACT: We present Co-SLAM, a neural RGB-D SLAM system based on a hybrid
representation, that performs robust camera tracking and high-fidelity surface
reconstruction in real time. Co-SLAM represents the scene as a multi-resolution
hash-grid to exploit its high convergence speed and ability to represent
high-frequency local features. In addition, Co-SLAM incorporates one-blob
encoding, to encourage surface coherence and completion in unobserved areas.
This joint parametric-coordinate encoding enables real-time and robust
performance by bringing the best of both worlds: fast convergence and surface
hole filling. Moreover, our ray sampling strategy allows Co-SLAM to perform
global bundle adjustment over all keyframes instead of requiring keyframe
selection to maintain a small number of active keyframes as competing neural
SLAM approaches do. Experimental results show that Co-SLAM runs at 10-17Hz and
achieves state-of-the-art scene reconstruction results, and competitive
tracking performance in various datasets and benchmarks (ScanNet, TUM, Replica,
Synthetic RGBD). Project page: https://hengyiwang.github.io/projects/CoSLAM
| no_new_dataset | 0.71027 |
2304.14401 | Jiteng Mu | Jiteng Mu, Shen Sang, Nuno Vasconcelos, Xiaolong Wang | ActorsNeRF: Animatable Few-shot Human Rendering with Generalizable NeRFs | Project Page : https://jitengmu.github.io/ActorsNeRF/ | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While NeRF-based human representations have shown impressive novel view
synthesis results, most methods still rely on a large number of images / views
for training. In this work, we propose a novel animatable NeRF called
ActorsNeRF. It is first pre-trained on diverse human subjects, and then adapted
with few-shot monocular video frames for a new actor with unseen poses.
Building on previous generalizable NeRFs with parameter sharing using a ConvNet
encoder, ActorsNeRF further adopts two human priors to capture the large human
appearance, shape, and pose variations. Specifically, in the encoded feature
space, we will first align different human subjects in a category-level
canonical space, and then align the same human from different frames in an
instance-level canonical space for rendering. We quantitatively and
qualitatively demonstrate that ActorsNeRF significantly outperforms the
existing state-of-the-art on few-shot generalization to new people and poses on
multiple datasets. Project Page: https://jitengmu.github.io/ActorsNeRF/
| [
{
"version": "v1",
"created": "Thu, 27 Apr 2023 17:58:48 GMT"
}
] | 2023-04-28T00:00:00 | [
[
"Mu",
"Jiteng",
""
],
[
"Sang",
"Shen",
""
],
[
"Vasconcelos",
"Nuno",
""
],
[
"Wang",
"Xiaolong",
""
]
] | TITLE: ActorsNeRF: Animatable Few-shot Human Rendering with Generalizable NeRFs
ABSTRACT: While NeRF-based human representations have shown impressive novel view
synthesis results, most methods still rely on a large number of images / views
for training. In this work, we propose a novel animatable NeRF called
ActorsNeRF. It is first pre-trained on diverse human subjects, and then adapted
with few-shot monocular video frames for a new actor with unseen poses.
Building on previous generalizable NeRFs with parameter sharing using a ConvNet
encoder, ActorsNeRF further adopts two human priors to capture the large human
appearance, shape, and pose variations. Specifically, in the encoded feature
space, we will first align different human subjects in a category-level
canonical space, and then align the same human from different frames in an
instance-level canonical space for rendering. We quantitatively and
qualitatively demonstrate that ActorsNeRF significantly outperforms the
existing state-of-the-art on few-shot generalization to new people and poses on
multiple datasets. Project Page: https://jitengmu.github.io/ActorsNeRF/
| no_new_dataset | 0.712038 |
2304.14406 | Sumith Kulal | Sumith Kulal, Tim Brooks, Alex Aiken, Jiajun Wu, Jimei Yang, Jingwan
Lu, Alexei A. Efros, Krishna Kumar Singh | Putting People in Their Place: Affordance-Aware Human Insertion into
Scenes | CVPR 2023. Project page with code:
https://sumith1896.github.io/affordance-insertion/ | null | null | null | cs.CV cs.AI cs.GR cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the problem of inferring scene affordances by presenting a method
for realistically inserting people into scenes. Given a scene image with a
marked region and an image of a person, we insert the person into the scene
while respecting the scene affordances. Our model can infer the set of
realistic poses given the scene context, re-pose the reference person, and
harmonize the composition. We set up the task in a self-supervised fashion by
learning to re-pose humans in video clips. We train a large-scale diffusion
model on a dataset of 2.4M video clips that produces diverse plausible poses
while respecting the scene context. Given the learned human-scene composition,
our model can also hallucinate realistic people and scenes when prompted
without conditioning and also enables interactive editing. A quantitative
evaluation shows that our method synthesizes more realistic human appearance
and more natural human-scene interactions than prior work.
| [
{
"version": "v1",
"created": "Thu, 27 Apr 2023 17:59:58 GMT"
}
] | 2023-04-28T00:00:00 | [
[
"Kulal",
"Sumith",
""
],
[
"Brooks",
"Tim",
""
],
[
"Aiken",
"Alex",
""
],
[
"Wu",
"Jiajun",
""
],
[
"Yang",
"Jimei",
""
],
[
"Lu",
"Jingwan",
""
],
[
"Efros",
"Alexei A.",
""
],
[
"Singh",
"Krishna Kumar",
""
]
] | TITLE: Putting People in Their Place: Affordance-Aware Human Insertion into
Scenes
ABSTRACT: We study the problem of inferring scene affordances by presenting a method
for realistically inserting people into scenes. Given a scene image with a
marked region and an image of a person, we insert the person into the scene
while respecting the scene affordances. Our model can infer the set of
realistic poses given the scene context, re-pose the reference person, and
harmonize the composition. We set up the task in a self-supervised fashion by
learning to re-pose humans in video clips. We train a large-scale diffusion
model on a dataset of 2.4M video clips that produces diverse plausible poses
while respecting the scene context. Given the learned human-scene composition,
our model can also hallucinate realistic people and scenes when prompted
without conditioning and also enables interactive editing. A quantitative
evaluation shows that our method synthesizes more realistic human appearance
and more natural human-scene interactions than prior work.
| new_dataset | 0.596198 |
1712.09752 | Abolfazl Asudeh | Abolfazl Asudeh, H. V. Jagadish, Julia Stoyanovich, Gautam Das | Designing Fair Ranking Schemes | null | null | 10.1145/3299869.3300079 | null | cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Items from a database are often ranked based on a combination of multiple
criteria. A user may have the flexibility to accept combinations that weigh
these criteria differently, within limits. On the other hand, this choice of
weights can greatly affect the fairness of the produced ranking. In this paper,
we develop a system that helps users choose criterion weights that lead to
greater fairness.
We consider ranking functions that compute the score of each item as a
weighted sum of (numeric) attribute values, and then sort items on their score.
Each ranking function can be expressed as a vector of weights, or as a point in
a multi-dimensional space. For a broad range of fairness criteria, we show how
to efficiently identify regions in this space that satisfy these criteria.
Using this identification method, our system is able to tell users whether
their proposed ranking function satisfies the desired fairness criteria and, if
it does not, to suggest the smallest modification that does. We develop
user-controllable approximation that and indexing techniques that are applied
during preprocessing, and support sub-second response times during the online
phase. Our extensive experiments on real datasets demonstrate that our methods
are able to find solutions that satisfy fairness criteria effectively and
efficiently.
| [
{
"version": "v1",
"created": "Thu, 28 Dec 2017 04:56:06 GMT"
},
{
"version": "v2",
"created": "Thu, 4 Jan 2018 16:35:55 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Asudeh",
"Abolfazl",
""
],
[
"Jagadish",
"H. V.",
""
],
[
"Stoyanovich",
"Julia",
""
],
[
"Das",
"Gautam",
""
]
] | TITLE: Designing Fair Ranking Schemes
ABSTRACT: Items from a database are often ranked based on a combination of multiple
criteria. A user may have the flexibility to accept combinations that weigh
these criteria differently, within limits. On the other hand, this choice of
weights can greatly affect the fairness of the produced ranking. In this paper,
we develop a system that helps users choose criterion weights that lead to
greater fairness.
We consider ranking functions that compute the score of each item as a
weighted sum of (numeric) attribute values, and then sort items on their score.
Each ranking function can be expressed as a vector of weights, or as a point in
a multi-dimensional space. For a broad range of fairness criteria, we show how
to efficiently identify regions in this space that satisfy these criteria.
Using this identification method, our system is able to tell users whether
their proposed ranking function satisfies the desired fairness criteria and, if
it does not, to suggest the smallest modification that does. We develop
user-controllable approximation that and indexing techniques that are applied
during preprocessing, and support sub-second response times during the online
phase. Our extensive experiments on real datasets demonstrate that our methods
are able to find solutions that satisfy fairness criteria effectively and
efficiently.
| no_new_dataset | 0.709837 |
1802.10303 | Abolfazl Asudeh | Abolfazl Asudeh and Azade Nazi and Nan Zhang and Gautam Das and H. V.
Jagadish | RRR: Rank-Regret Representative | null | null | 10.1145/3299869.3300080 | null | cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Selecting the best items in a dataset is a common task in data exploration.
However, the concept of "best" lies in the eyes of the beholder: different
users may consider different attributes more important, and hence arrive at
different rankings. Nevertheless, one can remove "dominated" items and create a
"representative" subset of the data set, comprising the "best items" in it. A
Pareto-optimal representative is guaranteed to contain the best item of each
possible ranking, but it can be almost as big as the full data. Representative
can be found if we relax the requirement to include the best item for every
possible user, and instead just limit the users' "regret". Existing work
defines regret as the loss in score by limiting consideration to the
representative instead of the full data set, for any chosen ranking function.
However, the score is often not a meaningful number and users may not
understand its absolute value. Sometimes small ranges in score can include
large fractions of the data set. In contrast, users do understand the notion of
rank ordering. Therefore, alternatively, we consider the position of the items
in the ranked list for defining the regret and propose the {\em rank-regret
representative} as the minimal subset of the data containing at least one of
the top-$k$ of any possible ranking function. This problem is NP-complete. We
use the geometric interpretation of items to bound their ranks on ranges of
functions and to utilize combinatorial geometry notions for developing
effective and efficient approximation algorithms for the problem. Experiments
on real datasets demonstrate that we can efficiently find small subsets with
small rank-regrets.
| [
{
"version": "v1",
"created": "Wed, 28 Feb 2018 08:24:02 GMT"
},
{
"version": "v2",
"created": "Sat, 3 Mar 2018 17:14:31 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Asudeh",
"Abolfazl",
""
],
[
"Nazi",
"Azade",
""
],
[
"Zhang",
"Nan",
""
],
[
"Das",
"Gautam",
""
],
[
"Jagadish",
"H. V.",
""
]
] | TITLE: RRR: Rank-Regret Representative
ABSTRACT: Selecting the best items in a dataset is a common task in data exploration.
However, the concept of "best" lies in the eyes of the beholder: different
users may consider different attributes more important, and hence arrive at
different rankings. Nevertheless, one can remove "dominated" items and create a
"representative" subset of the data set, comprising the "best items" in it. A
Pareto-optimal representative is guaranteed to contain the best item of each
possible ranking, but it can be almost as big as the full data. Representative
can be found if we relax the requirement to include the best item for every
possible user, and instead just limit the users' "regret". Existing work
defines regret as the loss in score by limiting consideration to the
representative instead of the full data set, for any chosen ranking function.
However, the score is often not a meaningful number and users may not
understand its absolute value. Sometimes small ranges in score can include
large fractions of the data set. In contrast, users do understand the notion of
rank ordering. Therefore, alternatively, we consider the position of the items
in the ranked list for defining the regret and propose the {\em rank-regret
representative} as the minimal subset of the data containing at least one of
the top-$k$ of any possible ranking function. This problem is NP-complete. We
use the geometric interpretation of items to bound their ranks on ranges of
functions and to utilize combinatorial geometry notions for developing
effective and efficient approximation algorithms for the problem. Experiments
on real datasets demonstrate that we can efficiently find small subsets with
small rank-regrets.
| no_new_dataset | 0.705214 |
1810.06742 | Abolfazl Asudeh | Abolfazl Asudeh, Zhongjun Jin, H. V. Jagadish | Assessing and Remedying Coverage for a Given Dataset | in ICDE 2019 | null | 10.1109/ICDE.2019.00056 | null | cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Data analysis impacts virtually every aspect of our society today. Often,
this analysis is performed on an existing dataset, possibly collected through a
process that the data scientists had limited control over. The existing data
analyzed may not include the complete universe, but it is expected to cover the
diversity of items in the universe. Lack of adequate coverage in the dataset
can result in undesirable outcomes such as biased decisions and algorithmic
racism, as well as creating vulnerabilities such as opening up room for
adversarial attacks.
In this paper, we assess the coverage of a given dataset over multiple
categorical attributes. We first provide efficient techniques for traversing
the combinatorial explosion of value combinations to identify any regions of
attribute space not adequately covered by the data. Then, we determine the
least amount of additional data that must be obtained to resolve this lack of
adequate coverage. We confirm the value of our proposal through both
theoretical analyses and comprehensive experiments on real data.
| [
{
"version": "v1",
"created": "Mon, 15 Oct 2018 22:59:14 GMT"
},
{
"version": "v2",
"created": "Sat, 23 Feb 2019 23:26:29 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Asudeh",
"Abolfazl",
""
],
[
"Jin",
"Zhongjun",
""
],
[
"Jagadish",
"H. V.",
""
]
] | TITLE: Assessing and Remedying Coverage for a Given Dataset
ABSTRACT: Data analysis impacts virtually every aspect of our society today. Often,
this analysis is performed on an existing dataset, possibly collected through a
process that the data scientists had limited control over. The existing data
analyzed may not include the complete universe, but it is expected to cover the
diversity of items in the universe. Lack of adequate coverage in the dataset
can result in undesirable outcomes such as biased decisions and algorithmic
racism, as well as creating vulnerabilities such as opening up room for
adversarial attacks.
In this paper, we assess the coverage of a given dataset over multiple
categorical attributes. We first provide efficient techniques for traversing
the combinatorial explosion of value combinations to identify any regions of
attribute space not adequately covered by the data. Then, we determine the
least amount of additional data that must be obtained to resolve this lack of
adequate coverage. We confirm the value of our proposal through both
theoretical analyses and comprehensive experiments on real data.
| no_new_dataset | 0.71222 |
2105.00082 | Haoyue Ping | Haoyue Ping, Julia Stoyanovich | Most Expected Winner: An Interpretation of Winners over Uncertain Voter
Preferences | This is the technical report of the following paper: Haoyue Ping and
Julia Stoyanovich. 2023. Most Expected Winner: An Interpretation of Winners
over Uncertain Voter Preferences. Proc. ACM Manag. Data, 1, N1, Article 22
(May 2023), 33 pages. https://doi.org/10.1145/3588702 | Proc. ACM Manag. Data, 1, N1, Article 22 (May 2023), 33 pages
(2023) | 10.1145/3588702 | null | cs.GT | http://creativecommons.org/licenses/by-nc-nd/4.0/ | It remains an open question how to determine the winner of an election when
voter preferences are incomplete or uncertain. One option is to assume some
probability space over the voting profile and select the Most Probable Winner
(MPW) -- the candidate or candidates with the best chance of winning. In this
paper, we propose an alternative winner interpretation, selecting the Most
Expected Winner (MEW) according to the expected performance of the candidates.
We separate the uncertainty in voter preferences into the generation step and
the observation step, which gives rise to a unified voting profile combining
both incomplete and probabilistic voting profiles. We use this framework to
establish the theoretical hardness of \mew over incomplete voter preferences,
and then identify a collection of tractable cases for a variety of voting
profiles, including those based on the popular Repeated Insertion Model (RIM)
and its special case, the Mallows model. We develop solvers customized for
various voter preference types to quantify the candidate performance for the
individual voters, and propose a pruning strategy that optimizes computation.
The performance of the proposed solvers and pruning strategy is evaluated
extensively on real and synthetic benchmarks, showing that our methods are
practical.
| [
{
"version": "v1",
"created": "Fri, 30 Apr 2021 20:44:49 GMT"
},
{
"version": "v2",
"created": "Sun, 9 Apr 2023 08:26:00 GMT"
},
{
"version": "v3",
"created": "Wed, 26 Apr 2023 02:56:17 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Ping",
"Haoyue",
""
],
[
"Stoyanovich",
"Julia",
""
]
] | TITLE: Most Expected Winner: An Interpretation of Winners over Uncertain Voter
Preferences
ABSTRACT: It remains an open question how to determine the winner of an election when
voter preferences are incomplete or uncertain. One option is to assume some
probability space over the voting profile and select the Most Probable Winner
(MPW) -- the candidate or candidates with the best chance of winning. In this
paper, we propose an alternative winner interpretation, selecting the Most
Expected Winner (MEW) according to the expected performance of the candidates.
We separate the uncertainty in voter preferences into the generation step and
the observation step, which gives rise to a unified voting profile combining
both incomplete and probabilistic voting profiles. We use this framework to
establish the theoretical hardness of \mew over incomplete voter preferences,
and then identify a collection of tractable cases for a variety of voting
profiles, including those based on the popular Repeated Insertion Model (RIM)
and its special case, the Mallows model. We develop solvers customized for
various voter preference types to quantify the candidate performance for the
individual voters, and propose a pruning strategy that optimizes computation.
The performance of the proposed solvers and pruning strategy is evaluated
extensively on real and synthetic benchmarks, showing that our methods are
practical.
| no_new_dataset | 0.710434 |
2112.07833 | Li-Hsiang Shen | Chia-Jou Ku, Li-Hsiang Shen, Kai-Ten Feng | Reconfigurable Intelligent Surface-Empowered Self-Interference
Cancellation for 6G Full-Duplex MIMO Communication Systems | null | IEEE 33rd Annual International Symposium on Personal, Indoor and
Mobile Radio Communications (PIMRC), 2022 | 10.1109/PIMRC54779.2022.9978133 | null | cs.IT cs.NI math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Substantially increasing wireless traffic and extending serving coverage is
required with the advent of sixth-generation (6G) wireless communication
networks. Reconfigurable intelligent surface (RIS) is widely considered as a
promising technique which is capable of improving the system sum rate and
energy efficiency. Moreover, full-duplex (FD) multi-input-multi-output (MIMO)
transmission provides simultaneous transmit and received signals, which
theoretically provides twice of spectrum efficiency. However, the
self-interference (SI) in FD system is a challenging task requiring
high-overhead cancellation, which can be resolved by configuring appropriate
phase shifts of RIS. This paper has proposed an RIS-empowered full-duplex
interference cancellation (RFIC) scheme in order to alleviate the severe
interference in an RIS-FD system. We consider the interference minimization of
RIS-FD MIMO while guaranteeing quality-of-service (QoS) of whole system. The
closed-form solution of RIS phase shifts is theoretically derived with the
discussion of different numbers of RIS elements and receiving antennas.
Simulation results reveal that the proposed RFIC scheme outperforms existing
benchmarks with more than 50% of performance gain of sum rate.
| [
{
"version": "v1",
"created": "Wed, 15 Dec 2021 02:03:06 GMT"
},
{
"version": "v2",
"created": "Wed, 29 Mar 2023 04:07:55 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Ku",
"Chia-Jou",
""
],
[
"Shen",
"Li-Hsiang",
""
],
[
"Feng",
"Kai-Ten",
""
]
] | TITLE: Reconfigurable Intelligent Surface-Empowered Self-Interference
Cancellation for 6G Full-Duplex MIMO Communication Systems
ABSTRACT: Substantially increasing wireless traffic and extending serving coverage is
required with the advent of sixth-generation (6G) wireless communication
networks. Reconfigurable intelligent surface (RIS) is widely considered as a
promising technique which is capable of improving the system sum rate and
energy efficiency. Moreover, full-duplex (FD) multi-input-multi-output (MIMO)
transmission provides simultaneous transmit and received signals, which
theoretically provides twice of spectrum efficiency. However, the
self-interference (SI) in FD system is a challenging task requiring
high-overhead cancellation, which can be resolved by configuring appropriate
phase shifts of RIS. This paper has proposed an RIS-empowered full-duplex
interference cancellation (RFIC) scheme in order to alleviate the severe
interference in an RIS-FD system. We consider the interference minimization of
RIS-FD MIMO while guaranteeing quality-of-service (QoS) of whole system. The
closed-form solution of RIS phase shifts is theoretically derived with the
discussion of different numbers of RIS elements and receiving antennas.
Simulation results reveal that the proposed RFIC scheme outperforms existing
benchmarks with more than 50% of performance gain of sum rate.
| no_new_dataset | 0.708036 |
2202.00640 | Francesco Fabbri | Francesco Fabbri, Yanhao Wang, Francesco Bonchi, Carlos Castillo,
Michael Mathioudakis | Rewiring What-to-Watch-Next Recommendations to Reduce Radicalization
Pathways | To appear in the Web conference 2022 (WWW '22) | null | 10.1145/3485447.3512143 | null | cs.CY cs.LG cs.SI | http://creativecommons.org/licenses/by/4.0/ | Recommender systems typically suggest to users content similar to what they
consumed in the past. If a user happens to be exposed to strongly polarized
content, she might subsequently receive recommendations which may steer her
towards more and more radicalized content, eventually being trapped in what we
call a "radicalization pathway". In this paper, we study the problem of
mitigating radicalization pathways using a graph-based approach. Specifically,
we model the set of recommendations of a "what-to-watch-next" recommender as a
d-regular directed graph where nodes correspond to content items, links to
recommendations, and paths to possible user sessions. We measure the
"segregation" score of a node representing radicalized content as the expected
length of a random walk from that node to any node representing non-radicalized
content. High segregation scores are associated to larger chances to get users
trapped in radicalization pathways. Hence, we define the problem of reducing
the prevalence of radicalization pathways by selecting a small number of edges
to "rewire", so to minimize the maximum of segregation scores among all
radicalized nodes, while maintaining the relevance of the recommendations. We
prove that the problem of finding the optimal set of recommendations to rewire
is NP-hard and NP-hard to approximate within any factor. Therefore, we turn our
attention to heuristics, and propose an efficient yet effective greedy
algorithm based on the absorbing random walk theory. Our experiments on
real-world datasets in the context of video and news recommendations confirm
the effectiveness of our proposal.
| [
{
"version": "v1",
"created": "Tue, 1 Feb 2022 18:27:34 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Fabbri",
"Francesco",
""
],
[
"Wang",
"Yanhao",
""
],
[
"Bonchi",
"Francesco",
""
],
[
"Castillo",
"Carlos",
""
],
[
"Mathioudakis",
"Michael",
""
]
] | TITLE: Rewiring What-to-Watch-Next Recommendations to Reduce Radicalization
Pathways
ABSTRACT: Recommender systems typically suggest to users content similar to what they
consumed in the past. If a user happens to be exposed to strongly polarized
content, she might subsequently receive recommendations which may steer her
towards more and more radicalized content, eventually being trapped in what we
call a "radicalization pathway". In this paper, we study the problem of
mitigating radicalization pathways using a graph-based approach. Specifically,
we model the set of recommendations of a "what-to-watch-next" recommender as a
d-regular directed graph where nodes correspond to content items, links to
recommendations, and paths to possible user sessions. We measure the
"segregation" score of a node representing radicalized content as the expected
length of a random walk from that node to any node representing non-radicalized
content. High segregation scores are associated to larger chances to get users
trapped in radicalization pathways. Hence, we define the problem of reducing
the prevalence of radicalization pathways by selecting a small number of edges
to "rewire", so to minimize the maximum of segregation scores among all
radicalized nodes, while maintaining the relevance of the recommendations. We
prove that the problem of finding the optimal set of recommendations to rewire
is NP-hard and NP-hard to approximate within any factor. Therefore, we turn our
attention to heuristics, and propose an efficient yet effective greedy
algorithm based on the absorbing random walk theory. Our experiments on
real-world datasets in the context of video and news recommendations confirm
the effectiveness of our proposal.
| no_new_dataset | 0.710208 |
2202.13402 | Yutong Ban | Yutong Ban, Jennifer A. Eckhoff, Thomas M. Ward, Daniel A. Hashimoto,
Ozanan R. Meireles, Daniela Rus, Guy Rosman | Concept Graph Neural Networks for Surgical Video Understanding | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We constantly integrate our knowledge and understanding of the world to
enhance our interpretation of what we see.
This ability is crucial in application domains which entail reasoning about
multiple entities and concepts, such as AI-augmented surgery. In this paper, we
propose a novel way of integrating conceptual knowledge into temporal analysis
tasks via temporal concept graph networks. In the proposed networks, a global
knowledge graph is incorporated into the temporal analysis of surgical
instances, learning the meaning of concepts and relations as they apply to the
data. We demonstrate our results in surgical video data for tasks such as
verification of critical view of safety, as well as estimation of Parkland
grading scale. The results show that our method improves the recognition and
detection of complex benchmarks as well as enables other analytic applications
of interest.
| [
{
"version": "v1",
"created": "Sun, 27 Feb 2022 17:32:49 GMT"
},
{
"version": "v2",
"created": "Tue, 25 Apr 2023 20:35:43 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Ban",
"Yutong",
""
],
[
"Eckhoff",
"Jennifer A.",
""
],
[
"Ward",
"Thomas M.",
""
],
[
"Hashimoto",
"Daniel A.",
""
],
[
"Meireles",
"Ozanan R.",
""
],
[
"Rus",
"Daniela",
""
],
[
"Rosman",
"Guy",
""
]
] | TITLE: Concept Graph Neural Networks for Surgical Video Understanding
ABSTRACT: We constantly integrate our knowledge and understanding of the world to
enhance our interpretation of what we see.
This ability is crucial in application domains which entail reasoning about
multiple entities and concepts, such as AI-augmented surgery. In this paper, we
propose a novel way of integrating conceptual knowledge into temporal analysis
tasks via temporal concept graph networks. In the proposed networks, a global
knowledge graph is incorporated into the temporal analysis of surgical
instances, learning the meaning of concepts and relations as they apply to the
data. We demonstrate our results in surgical video data for tasks such as
verification of critical view of safety, as well as estimation of Parkland
grading scale. The results show that our method improves the recognition and
detection of complex benchmarks as well as enables other analytic applications
of interest.
| no_new_dataset | 0.711406 |
2206.04656 | Jenny Seidenschwarz | Jenny Seidenschwarz, Guillem Bras\'o, Victor Castro Serrano, Ismail
Elezi, and Laura Leal-Taix\'e | Simple Cues Lead to a Strong Multi-Object Tracker | Accepted to CVPR2023! | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | For a long time, the most common paradigm in Multi-Object Tracking was
tracking-by-detection (TbD), where objects are first detected and then
associated over video frames. For association, most models resourced to motion
and appearance cues, e.g., re-identification networks. Recent approaches based
on attention propose to learn the cues in a data-driven manner, showing
impressive results. In this paper, we ask ourselves whether simple good old TbD
methods are also capable of achieving the performance of end-to-end models. To
this end, we propose two key ingredients that allow a standard
re-identification network to excel at appearance-based tracking. We extensively
analyse its failure cases, and show that a combination of our appearance
features with a simple motion model leads to strong tracking results. Our
tracker generalizes to four public datasets, namely MOT17, MOT20, BDD100k, and
DanceTrack, achieving state-of-the-art performance.
https://github.com/dvl-tum/GHOST.
| [
{
"version": "v1",
"created": "Thu, 9 Jun 2022 17:55:51 GMT"
},
{
"version": "v2",
"created": "Fri, 10 Jun 2022 07:30:22 GMT"
},
{
"version": "v3",
"created": "Mon, 13 Jun 2022 10:31:56 GMT"
},
{
"version": "v4",
"created": "Thu, 1 Dec 2022 18:46:03 GMT"
},
{
"version": "v5",
"created": "Tue, 28 Feb 2023 13:48:48 GMT"
},
{
"version": "v6",
"created": "Wed, 1 Mar 2023 10:49:17 GMT"
},
{
"version": "v7",
"created": "Wed, 26 Apr 2023 09:44:03 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Seidenschwarz",
"Jenny",
""
],
[
"Brasó",
"Guillem",
""
],
[
"Serrano",
"Victor Castro",
""
],
[
"Elezi",
"Ismail",
""
],
[
"Leal-Taixé",
"Laura",
""
]
] | TITLE: Simple Cues Lead to a Strong Multi-Object Tracker
ABSTRACT: For a long time, the most common paradigm in Multi-Object Tracking was
tracking-by-detection (TbD), where objects are first detected and then
associated over video frames. For association, most models resourced to motion
and appearance cues, e.g., re-identification networks. Recent approaches based
on attention propose to learn the cues in a data-driven manner, showing
impressive results. In this paper, we ask ourselves whether simple good old TbD
methods are also capable of achieving the performance of end-to-end models. To
this end, we propose two key ingredients that allow a standard
re-identification network to excel at appearance-based tracking. We extensively
analyse its failure cases, and show that a combination of our appearance
features with a simple motion model leads to strong tracking results. Our
tracker generalizes to four public datasets, namely MOT17, MOT20, BDD100k, and
DanceTrack, achieving state-of-the-art performance.
https://github.com/dvl-tum/GHOST.
| no_new_dataset | 0.710208 |
2208.00194 | Yanhao Wang | Yanhao Wang and Francesco Fabbri and Michael Mathioudakis | Streaming Algorithms for Diversity Maximization with Fairness
Constraints | 13 pages, 11 figures; published in ICDE 2022 | null | 10.1109/ICDE53745.2022.00008 | null | cs.DS cs.LG | http://creativecommons.org/licenses/by-sa/4.0/ | Diversity maximization is a fundamental problem with wide applications in
data summarization, web search, and recommender systems. Given a set $X$ of $n$
elements, it asks to select a subset $S$ of $k \ll n$ elements with maximum
\emph{diversity}, as quantified by the dissimilarities among the elements in
$S$. In this paper, we focus on the diversity maximization problem with
fairness constraints in the streaming setting. Specifically, we consider the
max-min diversity objective, which selects a subset $S$ that maximizes the
minimum distance (dissimilarity) between any pair of distinct elements within
it. Assuming that the set $X$ is partitioned into $m$ disjoint groups by some
sensitive attribute, e.g., sex or race, ensuring \emph{fairness} requires that
the selected subset $S$ contains $k_i$ elements from each group $i \in [1,m]$.
A streaming algorithm should process $X$ sequentially in one pass and return a
subset with maximum \emph{diversity} while guaranteeing the fairness
constraint. Although diversity maximization has been extensively studied, the
only known algorithms that can work with the max-min diversity objective and
fairness constraints are very inefficient for data streams. Since diversity
maximization is NP-hard in general, we propose two approximation algorithms for
fair diversity maximization in data streams, the first of which is
$\frac{1-\varepsilon}{4}$-approximate and specific for $m=2$, where
$\varepsilon \in (0,1)$, and the second of which achieves a
$\frac{1-\varepsilon}{3m+2}$-approximation for an arbitrary $m$. Experimental
results on real-world and synthetic datasets show that both algorithms provide
solutions of comparable quality to the state-of-the-art algorithms while
running several orders of magnitude faster in the streaming setting.
| [
{
"version": "v1",
"created": "Sat, 30 Jul 2022 11:47:31 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Wang",
"Yanhao",
""
],
[
"Fabbri",
"Francesco",
""
],
[
"Mathioudakis",
"Michael",
""
]
] | TITLE: Streaming Algorithms for Diversity Maximization with Fairness
Constraints
ABSTRACT: Diversity maximization is a fundamental problem with wide applications in
data summarization, web search, and recommender systems. Given a set $X$ of $n$
elements, it asks to select a subset $S$ of $k \ll n$ elements with maximum
\emph{diversity}, as quantified by the dissimilarities among the elements in
$S$. In this paper, we focus on the diversity maximization problem with
fairness constraints in the streaming setting. Specifically, we consider the
max-min diversity objective, which selects a subset $S$ that maximizes the
minimum distance (dissimilarity) between any pair of distinct elements within
it. Assuming that the set $X$ is partitioned into $m$ disjoint groups by some
sensitive attribute, e.g., sex or race, ensuring \emph{fairness} requires that
the selected subset $S$ contains $k_i$ elements from each group $i \in [1,m]$.
A streaming algorithm should process $X$ sequentially in one pass and return a
subset with maximum \emph{diversity} while guaranteeing the fairness
constraint. Although diversity maximization has been extensively studied, the
only known algorithms that can work with the max-min diversity objective and
fairness constraints are very inefficient for data streams. Since diversity
maximization is NP-hard in general, we propose two approximation algorithms for
fair diversity maximization in data streams, the first of which is
$\frac{1-\varepsilon}{4}$-approximate and specific for $m=2$, where
$\varepsilon \in (0,1)$, and the second of which achieves a
$\frac{1-\varepsilon}{3m+2}$-approximation for an arbitrary $m$. Experimental
results on real-world and synthetic datasets show that both algorithms provide
solutions of comparable quality to the state-of-the-art algorithms while
running several orders of magnitude faster in the streaming setting.
| no_new_dataset | 0.707815 |
2208.05540 | Mahdi Razzaghpour | Ahura Jami, Mahdi Razzaghpour, Hussein Alnuweiri, Yaser P. Fallah | Augmented Driver Behavior Models for High-Fidelity Simulation Study of
Crash Detection Algorithms | null | null | null | null | cs.MA cs.AI cs.RO | http://creativecommons.org/licenses/by/4.0/ | Developing safety and efficiency applications for Connected and Automated
Vehicles (CAVs) require a great deal of testing and evaluation. The need for
the operation of these systems in critical and dangerous situations makes the
burden of their evaluation very costly, possibly dangerous, and time-consuming.
As an alternative, researchers attempt to study and evaluate their algorithms
and designs using simulation platforms. Modeling the behavior of drivers or
human operators in CAVs or other vehicles interacting with them is one of the
main challenges of such simulations. While developing a perfect model for human
behavior is a challenging task and an open problem, we present a significant
augmentation of the current models used in simulators for driver behavior. In
this paper, we present a simulation platform for a hybrid transportation system
that includes both human-driven and automated vehicles. In addition, we
decompose the human driving task and offer a modular approach to simulating a
large-scale traffic scenario, allowing for a thorough investigation of
automated and active safety systems. Such representation through Interconnected
modules offers a human-interpretable system that can be tuned to represent
different classes of drivers. Additionally, we analyze a large driving dataset
to extract expressive parameters that would best describe different driving
characteristics. Finally, we recreate a similarly dense traffic scenario within
our simulator and conduct a thorough analysis of various human-specific and
system-specific factors, studying their effect on traffic network performance
and safety.
| [
{
"version": "v1",
"created": "Wed, 10 Aug 2022 19:59:16 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 14:23:14 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Jami",
"Ahura",
""
],
[
"Razzaghpour",
"Mahdi",
""
],
[
"Alnuweiri",
"Hussein",
""
],
[
"Fallah",
"Yaser P.",
""
]
] | TITLE: Augmented Driver Behavior Models for High-Fidelity Simulation Study of
Crash Detection Algorithms
ABSTRACT: Developing safety and efficiency applications for Connected and Automated
Vehicles (CAVs) require a great deal of testing and evaluation. The need for
the operation of these systems in critical and dangerous situations makes the
burden of their evaluation very costly, possibly dangerous, and time-consuming.
As an alternative, researchers attempt to study and evaluate their algorithms
and designs using simulation platforms. Modeling the behavior of drivers or
human operators in CAVs or other vehicles interacting with them is one of the
main challenges of such simulations. While developing a perfect model for human
behavior is a challenging task and an open problem, we present a significant
augmentation of the current models used in simulators for driver behavior. In
this paper, we present a simulation platform for a hybrid transportation system
that includes both human-driven and automated vehicles. In addition, we
decompose the human driving task and offer a modular approach to simulating a
large-scale traffic scenario, allowing for a thorough investigation of
automated and active safety systems. Such representation through Interconnected
modules offers a human-interpretable system that can be tuned to represent
different classes of drivers. Additionally, we analyze a large driving dataset
to extract expressive parameters that would best describe different driving
characteristics. Finally, we recreate a similarly dense traffic scenario within
our simulator and conduct a thorough analysis of various human-specific and
system-specific factors, studying their effect on traffic network performance
and safety.
| no_new_dataset | 0.70724 |
2208.06553 | Yanhao Wang | Jiping Zheng and Yuan Ma and Wei Ma and Yanhao Wang and Xiaoyang Wang | Happiness Maximizing Sets under Group Fairness Constraints (Technical
Report) | Technical report, a shorter version to appear in PVLDB 16(2) | null | 10.14778/3565816.3565830 | null | cs.DB cs.CY cs.DS | http://creativecommons.org/licenses/by-sa/4.0/ | Finding a happiness maximizing set (HMS) from a database, i.e., selecting a
small subset of tuples that preserves the best score with respect to any
nonnegative linear utility function, is an important problem in multi-criteria
decision-making. When an HMS is extracted from a set of individuals to assist
data-driven algorithmic decisions such as hiring and admission, it is crucial
to ensure that the HMS can fairly represent different groups of candidates
without bias and discrimination. However, although the HMS problem was
extensively studied in the database community, existing algorithms do not take
group fairness into account and may provide solutions that under-represent some
groups.
In this paper, we propose and investigate a fair variant of HMS (FairHMS)
that not only maximizes the minimum happiness ratio but also guarantees that
the number of tuples chosen from each group falls within predefined lower and
upper bounds. Similar to the vanilla HMS problem, we show that FairHMS is
NP-hard in three and higher dimensions. Therefore, we first propose an exact
interval cover-based algorithm called IntCov for FairHMS on two-dimensional
databases. Then, we propose a bicriteria approximation algorithm called
BiGreedy for FairHMS on multi-dimensional databases by transforming it into a
submodular maximization problem under a matroid constraint. We also design an
adaptive sampling strategy to improve the practical efficiency of BiGreedy.
Extensive experiments on real-world and synthetic datasets confirm the efficacy
and efficiency of our proposal.
| [
{
"version": "v1",
"created": "Sat, 13 Aug 2022 02:54:29 GMT"
},
{
"version": "v2",
"created": "Wed, 17 Aug 2022 07:35:38 GMT"
},
{
"version": "v3",
"created": "Sat, 8 Oct 2022 08:46:52 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Zheng",
"Jiping",
""
],
[
"Ma",
"Yuan",
""
],
[
"Ma",
"Wei",
""
],
[
"Wang",
"Yanhao",
""
],
[
"Wang",
"Xiaoyang",
""
]
] | TITLE: Happiness Maximizing Sets under Group Fairness Constraints (Technical
Report)
ABSTRACT: Finding a happiness maximizing set (HMS) from a database, i.e., selecting a
small subset of tuples that preserves the best score with respect to any
nonnegative linear utility function, is an important problem in multi-criteria
decision-making. When an HMS is extracted from a set of individuals to assist
data-driven algorithmic decisions such as hiring and admission, it is crucial
to ensure that the HMS can fairly represent different groups of candidates
without bias and discrimination. However, although the HMS problem was
extensively studied in the database community, existing algorithms do not take
group fairness into account and may provide solutions that under-represent some
groups.
In this paper, we propose and investigate a fair variant of HMS (FairHMS)
that not only maximizes the minimum happiness ratio but also guarantees that
the number of tuples chosen from each group falls within predefined lower and
upper bounds. Similar to the vanilla HMS problem, we show that FairHMS is
NP-hard in three and higher dimensions. Therefore, we first propose an exact
interval cover-based algorithm called IntCov for FairHMS on two-dimensional
databases. Then, we propose a bicriteria approximation algorithm called
BiGreedy for FairHMS on multi-dimensional databases by transforming it into a
submodular maximization problem under a matroid constraint. We also design an
adaptive sampling strategy to improve the practical efficiency of BiGreedy.
Extensive experiments on real-world and synthetic datasets confirm the efficacy
and efficiency of our proposal.
| no_new_dataset | 0.707657 |
2209.01307 | Changwen Xu | Changwen Xu, Yuyang Wang, Amir Barati Farimani | TransPolymer: a Transformer-based language model for polymer property
predictions | null | npj Comput Mater 9, 64 (2023) | 10.1038/s41524-023-01016-5 | null | cs.LG physics.chem-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Accurate and efficient prediction of polymer properties is of great
significance in polymer design. Conventionally, expensive and time-consuming
experiments or simulations are required to evaluate polymer functions.
Recently, Transformer models, equipped with self-attention mechanisms, have
exhibited superior performance in natural language processing. However, such
methods have not been investigated in polymer sciences. Herein, we report
TransPolymer, a Transformer-based language model for polymer property
prediction. Our proposed polymer tokenizer with chemical awareness enables
learning representations from polymer sequences. Rigorous experiments on ten
polymer property prediction benchmarks demonstrate the superior performance of
TransPolymer. Moreover, we show that TransPolymer benefits from pretraining on
large unlabeled dataset via Masked Language Modeling. Experimental results
further manifest the important role of self-attention in modeling polymer
sequences. We highlight this model as a promising computational tool for
promoting rational polymer design and understanding structure-property
relationships from a data science view.
| [
{
"version": "v1",
"created": "Sat, 3 Sep 2022 01:29:59 GMT"
},
{
"version": "v2",
"created": "Sun, 11 Sep 2022 03:05:19 GMT"
},
{
"version": "v3",
"created": "Mon, 6 Feb 2023 21:18:29 GMT"
},
{
"version": "v4",
"created": "Wed, 26 Apr 2023 01:52:00 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Xu",
"Changwen",
""
],
[
"Wang",
"Yuyang",
""
],
[
"Farimani",
"Amir Barati",
""
]
] | TITLE: TransPolymer: a Transformer-based language model for polymer property
predictions
ABSTRACT: Accurate and efficient prediction of polymer properties is of great
significance in polymer design. Conventionally, expensive and time-consuming
experiments or simulations are required to evaluate polymer functions.
Recently, Transformer models, equipped with self-attention mechanisms, have
exhibited superior performance in natural language processing. However, such
methods have not been investigated in polymer sciences. Herein, we report
TransPolymer, a Transformer-based language model for polymer property
prediction. Our proposed polymer tokenizer with chemical awareness enables
learning representations from polymer sequences. Rigorous experiments on ten
polymer property prediction benchmarks demonstrate the superior performance of
TransPolymer. Moreover, we show that TransPolymer benefits from pretraining on
large unlabeled dataset via Masked Language Modeling. Experimental results
further manifest the important role of self-attention in modeling polymer
sequences. We highlight this model as a promising computational tool for
promoting rational polymer design and understanding structure-property
relationships from a data science view.
| no_new_dataset | 0.710616 |
2210.04432 | Kavisha Vidanapathirana | Kavisha Vidanapathirana, Peyman Moghadam, Sridha Sridharan, Clinton
Fookes | Spectral Geometric Verification: Re-Ranking Point Cloud Retrieval for
Metric Localization | Accepted for publication in IEEE RA-L (2023) | null | 10.1109/LRA.2023.3255560 | null | cs.CV cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In large-scale metric localization, an incorrect result during retrieval will
lead to an incorrect pose estimate or loop closure. Re-ranking methods propose
to take into account all the top retrieval candidates and re-order them to
increase the likelihood of the top candidate being correct. However,
state-of-the-art re-ranking methods are inefficient when re-ranking many
potential candidates due to their need for resource intensive point cloud
registration between the query and each candidate. In this work, we propose an
efficient spectral method for geometric verification (named SpectralGV) that
does not require registration. We demonstrate how the optimal inter-cluster
score of the correspondence compatibility graph of two point clouds represents
a robust fitness score measuring their spatial consistency. This score takes
into account the subtle geometric differences between structurally similar
point clouds and therefore can be used to identify the correct candidate among
potential matches retrieved by global similarity search. SpectralGV is
deterministic, robust to outlier correspondences, and can be computed in
parallel for all potential candidates. We conduct extensive experiments on 5
large-scale datasets to demonstrate that SpectralGV outperforms other
state-of-the-art re-ranking methods and show that it consistently improves the
recall and pose estimation of 3 state-of-the-art metric localization
architectures while having a negligible effect on their runtime. The
open-source implementation and trained models are available at:
https://github.com/csiro-robotics/SpectralGV.
| [
{
"version": "v1",
"created": "Mon, 10 Oct 2022 04:36:50 GMT"
},
{
"version": "v2",
"created": "Mon, 6 Mar 2023 06:29:13 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Vidanapathirana",
"Kavisha",
""
],
[
"Moghadam",
"Peyman",
""
],
[
"Sridharan",
"Sridha",
""
],
[
"Fookes",
"Clinton",
""
]
] | TITLE: Spectral Geometric Verification: Re-Ranking Point Cloud Retrieval for
Metric Localization
ABSTRACT: In large-scale metric localization, an incorrect result during retrieval will
lead to an incorrect pose estimate or loop closure. Re-ranking methods propose
to take into account all the top retrieval candidates and re-order them to
increase the likelihood of the top candidate being correct. However,
state-of-the-art re-ranking methods are inefficient when re-ranking many
potential candidates due to their need for resource intensive point cloud
registration between the query and each candidate. In this work, we propose an
efficient spectral method for geometric verification (named SpectralGV) that
does not require registration. We demonstrate how the optimal inter-cluster
score of the correspondence compatibility graph of two point clouds represents
a robust fitness score measuring their spatial consistency. This score takes
into account the subtle geometric differences between structurally similar
point clouds and therefore can be used to identify the correct candidate among
potential matches retrieved by global similarity search. SpectralGV is
deterministic, robust to outlier correspondences, and can be computed in
parallel for all potential candidates. We conduct extensive experiments on 5
large-scale datasets to demonstrate that SpectralGV outperforms other
state-of-the-art re-ranking methods and show that it consistently improves the
recall and pose estimation of 3 state-of-the-art metric localization
architectures while having a negligible effect on their runtime. The
open-source implementation and trained models are available at:
https://github.com/csiro-robotics/SpectralGV.
| no_new_dataset | 0.711017 |
2210.06289 | Zhiying Song | Zhiying Song, Fuxi Wen, Hailiang Zhang, Jun Li | A Cooperative Perception System Robust to Localization Errors | Accepted by IEEE IV 2023 | null | null | null | cs.MA cs.CV cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Cooperative perception is challenging for safety-critical autonomous driving
applications.The errors in the shared position and pose cause an inaccurate
relative transform estimation and disrupt the robust mapping of the Ego
vehicle. We propose a distributed object-level cooperative perception system
called OptiMatch, in which the detected 3D bounding boxes and local state
information are shared between the connected vehicles. To correct the noisy
relative transform, the local measurements of both connected vehicles (bounding
boxes) are utilized, and an optimal transport theory-based algorithm is
developed to filter out those objects jointly detected by the vehicles along
with their correspondence, constructing an associated co-visible set. A
correction transform is estimated from the matched object pairs and further
applied to the noisy relative transform, followed by global fusion and dynamic
mapping. Experiment results show that robust performance is achieved for
different levels of location and heading errors, and the proposed framework
outperforms the state-of-the-art benchmark fusion schemes, including early,
late, and intermediate fusion, on average precision by a large margin when
location and/or heading errors occur.
| [
{
"version": "v1",
"created": "Wed, 12 Oct 2022 15:07:24 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 00:13:29 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Song",
"Zhiying",
""
],
[
"Wen",
"Fuxi",
""
],
[
"Zhang",
"Hailiang",
""
],
[
"Li",
"Jun",
""
]
] | TITLE: A Cooperative Perception System Robust to Localization Errors
ABSTRACT: Cooperative perception is challenging for safety-critical autonomous driving
applications.The errors in the shared position and pose cause an inaccurate
relative transform estimation and disrupt the robust mapping of the Ego
vehicle. We propose a distributed object-level cooperative perception system
called OptiMatch, in which the detected 3D bounding boxes and local state
information are shared between the connected vehicles. To correct the noisy
relative transform, the local measurements of both connected vehicles (bounding
boxes) are utilized, and an optimal transport theory-based algorithm is
developed to filter out those objects jointly detected by the vehicles along
with their correspondence, constructing an associated co-visible set. A
correction transform is estimated from the matched object pairs and further
applied to the noisy relative transform, followed by global fusion and dynamic
mapping. Experiment results show that robust performance is achieved for
different levels of location and heading errors, and the proposed framework
outperforms the state-of-the-art benchmark fusion schemes, including early,
late, and intermediate fusion, on average precision by a large margin when
location and/or heading errors occur.
| no_new_dataset | 0.710007 |
2210.06989 | Richa Upadhyay | Richa Upadhyay, Prakash Chandra Chhipa, Ronald Phlypo, Rajkumar Saini,
Marcus Liwicki | Multi-Task Meta Learning: learn how to adapt to unseen tasks | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-sa/4.0/ | This work proposes Multi-task Meta Learning (MTML), integrating two learning
paradigms Multi-Task Learning (MTL) and meta learning, to bring together the
best of both worlds. In particular, it focuses simultaneous learning of
multiple tasks, an element of MTL and promptly adapting to new tasks, a quality
of meta learning. It is important to highlight that we focus on heterogeneous
tasks, which are of distinct kind, in contrast to typically considered
homogeneous tasks (e.g., if all tasks are classification or if all tasks are
regression tasks). The fundamental idea is to train a multi-task model, such
that when an unseen task is introduced, it can learn in fewer steps whilst
offering a performance at least as good as conventional single task learning on
the new task or inclusion within the MTL. By conducting various experiments, we
demonstrate this paradigm on two datasets and four tasks: NYU-v2 and the
taskonomy dataset for which we perform semantic segmentation, depth estimation,
surface normal estimation, and edge detection. MTML achieves state-of-the-art
results for three out of four tasks for the NYU-v2 dataset and two out of four
for the taskonomy dataset. In the taskonomy dataset, it was discovered that
many pseudo-labeled segmentation masks lacked classes that were expected to be
present in the ground truth; however, our MTML approach was found to be
effective in detecting these missing classes, delivering good qualitative
results. While, quantitatively its performance was affected due to the presence
of incorrect ground truth labels. The the source code for reproducibility can
be found at https://github.com/ricupa/MTML-learn-how-to-adapt-to-unseen-tasks.
| [
{
"version": "v1",
"created": "Thu, 13 Oct 2022 12:59:54 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Nov 2022 15:03:51 GMT"
},
{
"version": "v3",
"created": "Sat, 11 Feb 2023 11:43:45 GMT"
},
{
"version": "v4",
"created": "Wed, 26 Apr 2023 08:09:10 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Upadhyay",
"Richa",
""
],
[
"Chhipa",
"Prakash Chandra",
""
],
[
"Phlypo",
"Ronald",
""
],
[
"Saini",
"Rajkumar",
""
],
[
"Liwicki",
"Marcus",
""
]
] | TITLE: Multi-Task Meta Learning: learn how to adapt to unseen tasks
ABSTRACT: This work proposes Multi-task Meta Learning (MTML), integrating two learning
paradigms Multi-Task Learning (MTL) and meta learning, to bring together the
best of both worlds. In particular, it focuses simultaneous learning of
multiple tasks, an element of MTL and promptly adapting to new tasks, a quality
of meta learning. It is important to highlight that we focus on heterogeneous
tasks, which are of distinct kind, in contrast to typically considered
homogeneous tasks (e.g., if all tasks are classification or if all tasks are
regression tasks). The fundamental idea is to train a multi-task model, such
that when an unseen task is introduced, it can learn in fewer steps whilst
offering a performance at least as good as conventional single task learning on
the new task or inclusion within the MTL. By conducting various experiments, we
demonstrate this paradigm on two datasets and four tasks: NYU-v2 and the
taskonomy dataset for which we perform semantic segmentation, depth estimation,
surface normal estimation, and edge detection. MTML achieves state-of-the-art
results for three out of four tasks for the NYU-v2 dataset and two out of four
for the taskonomy dataset. In the taskonomy dataset, it was discovered that
many pseudo-labeled segmentation masks lacked classes that were expected to be
present in the ground truth; however, our MTML approach was found to be
effective in detecting these missing classes, delivering good qualitative
results. While, quantitatively its performance was affected due to the presence
of incorrect ground truth labels. The the source code for reproducibility can
be found at https://github.com/ricupa/MTML-learn-how-to-adapt-to-unseen-tasks.
| no_new_dataset | 0.688298 |
2210.17475 | Sarath Shekkizhar | Carlos Hurtado, Sarath Shekkizhar, Javier Ruiz-Hidalgo, Antonio Ortega | Study of Manifold Geometry using Multiscale Non-Negative Kernel Graphs | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Modern machine learning systems are increasingly trained on large amounts of
data embedded in high-dimensional spaces. Often this is done without analyzing
the structure of the dataset. In this work, we propose a framework to study the
geometric structure of the data. We make use of our recently introduced
non-negative kernel (NNK) regression graphs to estimate the point density,
intrinsic dimension, and the linearity of the data manifold (curvature). We
further generalize the graph construction and geometric estimation to multiple
scale by iteratively merging neighborhoods in the input data. Our experiments
demonstrate the effectiveness of our proposed approach over other baselines in
estimating the local geometry of the data manifolds on synthetic and real
datasets.
| [
{
"version": "v1",
"created": "Mon, 31 Oct 2022 17:01:17 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 17:41:44 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Hurtado",
"Carlos",
""
],
[
"Shekkizhar",
"Sarath",
""
],
[
"Ruiz-Hidalgo",
"Javier",
""
],
[
"Ortega",
"Antonio",
""
]
] | TITLE: Study of Manifold Geometry using Multiscale Non-Negative Kernel Graphs
ABSTRACT: Modern machine learning systems are increasingly trained on large amounts of
data embedded in high-dimensional spaces. Often this is done without analyzing
the structure of the dataset. In this work, we propose a framework to study the
geometric structure of the data. We make use of our recently introduced
non-negative kernel (NNK) regression graphs to estimate the point density,
intrinsic dimension, and the linearity of the data manifold (curvature). We
further generalize the graph construction and geometric estimation to multiple
scale by iteratively merging neighborhoods in the input data. Our experiments
demonstrate the effectiveness of our proposed approach over other baselines in
estimating the local geometry of the data manifolds on synthetic and real
datasets.
| no_new_dataset | 0.712026 |
2211.01960 | Vladislav Lomtev | Vladislav Lomtev, Alexander Kovalev, Alexey Timchenko | FingerFlex: Inferring Finger Trajectories from ECoG signals | 6 pages, 3 figures, 4 tables. Preprint. Under review | null | null | null | q-bio.NC cs.HC cs.LG | http://creativecommons.org/licenses/by/4.0/ | Motor brain-computer interface (BCI) development relies critically on neural
time series decoding algorithms. Recent advances in deep learning architectures
allow for automatic feature selection to approximate higher-order dependencies
in data. This article presents the FingerFlex model - a convolutional
encoder-decoder architecture adapted for finger movement regression on
electrocorticographic (ECoG) brain data. State-of-the-art performance was
achieved on a publicly available BCI competition IV dataset 4 with a
correlation coefficient between true and predicted trajectories up to 0.74. The
presented method provides the opportunity for developing fully-functional
high-precision cortical motor brain-computer interfaces.
| [
{
"version": "v1",
"created": "Sun, 23 Oct 2022 16:26:01 GMT"
},
{
"version": "v2",
"created": "Tue, 25 Apr 2023 19:14:18 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Lomtev",
"Vladislav",
""
],
[
"Kovalev",
"Alexander",
""
],
[
"Timchenko",
"Alexey",
""
]
] | TITLE: FingerFlex: Inferring Finger Trajectories from ECoG signals
ABSTRACT: Motor brain-computer interface (BCI) development relies critically on neural
time series decoding algorithms. Recent advances in deep learning architectures
allow for automatic feature selection to approximate higher-order dependencies
in data. This article presents the FingerFlex model - a convolutional
encoder-decoder architecture adapted for finger movement regression on
electrocorticographic (ECoG) brain data. State-of-the-art performance was
achieved on a publicly available BCI competition IV dataset 4 with a
correlation coefficient between true and predicted trajectories up to 0.74. The
presented method provides the opportunity for developing fully-functional
high-precision cortical motor brain-computer interfaces.
| no_new_dataset | 0.710201 |
2211.05105 | Manuel Brack | Patrick Schramowski, Manuel Brack, Bj\"orn Deiseroth, Kristian
Kersting | Safe Latent Diffusion: Mitigating Inappropriate Degeneration in
Diffusion Models | Proceedings of the 22nd IEEE/CVF Conference on Computer Vision and
Pattern Recognition (CVPR), 2023 | null | null | null | cs.CV cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Text-conditioned image generation models have recently achieved astonishing
results in image quality and text alignment and are consequently employed in a
fast-growing number of applications. Since they are highly data-driven, relying
on billion-sized datasets randomly scraped from the internet, they also suffer,
as we demonstrate, from degenerated and biased human behavior. In turn, they
may even reinforce such biases. To help combat these undesired side effects, we
present safe latent diffusion (SLD). Specifically, to measure the inappropriate
degeneration due to unfiltered and imbalanced training sets, we establish a
novel image generation test bed-inappropriate image prompts (I2P)-containing
dedicated, real-world image-to-text prompts covering concepts such as nudity
and violence. As our exhaustive empirical evaluation demonstrates, the
introduced SLD removes and suppresses inappropriate image parts during the
diffusion process, with no additional training required and no adverse effect
on overall image quality or text alignment.
| [
{
"version": "v1",
"created": "Wed, 9 Nov 2022 18:54:25 GMT"
},
{
"version": "v2",
"created": "Sat, 19 Nov 2022 16:10:46 GMT"
},
{
"version": "v3",
"created": "Tue, 25 Apr 2023 16:27:56 GMT"
},
{
"version": "v4",
"created": "Wed, 26 Apr 2023 11:47:44 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Schramowski",
"Patrick",
""
],
[
"Brack",
"Manuel",
""
],
[
"Deiseroth",
"Björn",
""
],
[
"Kersting",
"Kristian",
""
]
] | TITLE: Safe Latent Diffusion: Mitigating Inappropriate Degeneration in
Diffusion Models
ABSTRACT: Text-conditioned image generation models have recently achieved astonishing
results in image quality and text alignment and are consequently employed in a
fast-growing number of applications. Since they are highly data-driven, relying
on billion-sized datasets randomly scraped from the internet, they also suffer,
as we demonstrate, from degenerated and biased human behavior. In turn, they
may even reinforce such biases. To help combat these undesired side effects, we
present safe latent diffusion (SLD). Specifically, to measure the inappropriate
degeneration due to unfiltered and imbalanced training sets, we establish a
novel image generation test bed-inappropriate image prompts (I2P)-containing
dedicated, real-world image-to-text prompts covering concepts such as nudity
and violence. As our exhaustive empirical evaluation demonstrates, the
introduced SLD removes and suppresses inappropriate image parts during the
diffusion process, with no additional training required and no adverse effect
on overall image quality or text alignment.
| no_new_dataset | 0.710415 |
2211.11158 | Yue Yang | Yue Yang, Artemis Panagopoulou, Shenghao Zhou, Daniel Jin, Chris
Callison-Burch, Mark Yatskar | Language in a Bottle: Language Model Guided Concept Bottlenecks for
Interpretable Image Classification | Published in CVPR 2023, 18 pages, 12 figures, 16 tables | null | null | null | cs.CV cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Concept Bottleneck Models (CBM) are inherently interpretable models that
factor model decisions into human-readable concepts. They allow people to
easily understand why a model is failing, a critical feature for high-stakes
applications. CBMs require manually specified concepts and often under-perform
their black box counterparts, preventing their broad adoption. We address these
shortcomings and are first to show how to construct high-performance CBMs
without manual specification of similar accuracy to black box models. Our
approach, Language Guided Bottlenecks (LaBo), leverages a language model,
GPT-3, to define a large space of possible bottlenecks. Given a problem domain,
LaBo uses GPT-3 to produce factual sentences about categories to form candidate
concepts. LaBo efficiently searches possible bottlenecks through a novel
submodular utility that promotes the selection of discriminative and diverse
information. Ultimately, GPT-3's sentential concepts can be aligned to images
using CLIP, to form a bottleneck layer. Experiments demonstrate that LaBo is a
highly effective prior for concepts important to visual recognition. In the
evaluation with 11 diverse datasets, LaBo bottlenecks excel at few-shot
classification: they are 11.7% more accurate than black box linear probes at 1
shot and comparable with more data. Overall, LaBo demonstrates that inherently
interpretable models can be widely applied at similar, or better, performance
than black box approaches.
| [
{
"version": "v1",
"created": "Mon, 21 Nov 2022 03:05:02 GMT"
},
{
"version": "v2",
"created": "Tue, 25 Apr 2023 22:06:42 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Yang",
"Yue",
""
],
[
"Panagopoulou",
"Artemis",
""
],
[
"Zhou",
"Shenghao",
""
],
[
"Jin",
"Daniel",
""
],
[
"Callison-Burch",
"Chris",
""
],
[
"Yatskar",
"Mark",
""
]
] | TITLE: Language in a Bottle: Language Model Guided Concept Bottlenecks for
Interpretable Image Classification
ABSTRACT: Concept Bottleneck Models (CBM) are inherently interpretable models that
factor model decisions into human-readable concepts. They allow people to
easily understand why a model is failing, a critical feature for high-stakes
applications. CBMs require manually specified concepts and often under-perform
their black box counterparts, preventing their broad adoption. We address these
shortcomings and are first to show how to construct high-performance CBMs
without manual specification of similar accuracy to black box models. Our
approach, Language Guided Bottlenecks (LaBo), leverages a language model,
GPT-3, to define a large space of possible bottlenecks. Given a problem domain,
LaBo uses GPT-3 to produce factual sentences about categories to form candidate
concepts. LaBo efficiently searches possible bottlenecks through a novel
submodular utility that promotes the selection of discriminative and diverse
information. Ultimately, GPT-3's sentential concepts can be aligned to images
using CLIP, to form a bottleneck layer. Experiments demonstrate that LaBo is a
highly effective prior for concepts important to visual recognition. In the
evaluation with 11 diverse datasets, LaBo bottlenecks excel at few-shot
classification: they are 11.7% more accurate than black box linear probes at 1
shot and comparable with more data. Overall, LaBo demonstrates that inherently
interpretable models can be widely applied at similar, or better, performance
than black box approaches.
| no_new_dataset | 0.711675 |
2211.11355 | Timo Kaiser | Timo Kaiser, Lukas Ehmann, Christoph Reinders and Bodo Rosenhahn | Blind Knowledge Distillation for Robust Image Classification | 8 pages, 4 figures, 2 tables. Submitted to the 1st Learning and
Mining with Noisy Labels Challenge on IJCAI22, see
http://competition.noisylabels.com/. Code is available
https://github.com/TimoK93/blind_knowledge_distillation | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Optimizing neural networks with noisy labels is a challenging task,
especially if the label set contains real-world noise. Networks tend to
generalize to reasonable patterns in the early training stages and overfit to
specific details of noisy samples in the latter ones. We introduce Blind
Knowledge Distillation - a novel teacher-student approach for learning with
noisy labels by masking the ground truth related teacher output to filter out
potentially corrupted knowledge and to estimate the tipping point from
generalizing to overfitting. Based on this, we enable the estimation of noise
in the training data with Otsus algorithm. With this estimation, we train the
network with a modified weighted cross-entropy loss function. We show in our
experiments that Blind Knowledge Distillation detects overfitting effectively
during training and improves the detection of clean and noisy labels on the
recently published CIFAR-N dataset. Code is available at GitHub.
| [
{
"version": "v1",
"created": "Mon, 21 Nov 2022 11:17:07 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Kaiser",
"Timo",
""
],
[
"Ehmann",
"Lukas",
""
],
[
"Reinders",
"Christoph",
""
],
[
"Rosenhahn",
"Bodo",
""
]
] | TITLE: Blind Knowledge Distillation for Robust Image Classification
ABSTRACT: Optimizing neural networks with noisy labels is a challenging task,
especially if the label set contains real-world noise. Networks tend to
generalize to reasonable patterns in the early training stages and overfit to
specific details of noisy samples in the latter ones. We introduce Blind
Knowledge Distillation - a novel teacher-student approach for learning with
noisy labels by masking the ground truth related teacher output to filter out
potentially corrupted knowledge and to estimate the tipping point from
generalizing to overfitting. Based on this, we enable the estimation of noise
in the training data with Otsus algorithm. With this estimation, we train the
network with a modified weighted cross-entropy loss function. We show in our
experiments that Blind Knowledge Distillation detects overfitting effectively
during training and improves the detection of clean and noisy labels on the
recently published CIFAR-N dataset. Code is available at GitHub.
| no_new_dataset | 0.7114 |
2212.01674 | Jihye Kim | Jihye Kim, Aristide Baratin, Yan Zhang, Simon Lacoste-Julien | CrossSplit: Mitigating Label Noise Memorization through Data Splitting | Accepted to ICML 2023 | null | null | null | cs.CV cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We approach the problem of improving robustness of deep learning algorithms
in the presence of label noise. Building upon existing label correction and
co-teaching methods, we propose a novel training procedure to mitigate the
memorization of noisy labels, called CrossSplit, which uses a pair of neural
networks trained on two disjoint parts of the labelled dataset. CrossSplit
combines two main ingredients: (i) Cross-split label correction. The idea is
that, since the model trained on one part of the data cannot memorize
example-label pairs from the other part, the training labels presented to each
network can be smoothly adjusted by using the predictions of its peer network;
(ii) Cross-split semi-supervised training. A network trained on one part of the
data also uses the unlabeled inputs of the other part. Extensive experiments on
CIFAR-10, CIFAR-100, Tiny-ImageNet and mini-WebVision datasets demonstrate that
our method can outperform the current state-of-the-art in a wide range of noise
ratios.
| [
{
"version": "v1",
"created": "Sat, 3 Dec 2022 19:09:56 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 15:33:27 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Kim",
"Jihye",
""
],
[
"Baratin",
"Aristide",
""
],
[
"Zhang",
"Yan",
""
],
[
"Lacoste-Julien",
"Simon",
""
]
] | TITLE: CrossSplit: Mitigating Label Noise Memorization through Data Splitting
ABSTRACT: We approach the problem of improving robustness of deep learning algorithms
in the presence of label noise. Building upon existing label correction and
co-teaching methods, we propose a novel training procedure to mitigate the
memorization of noisy labels, called CrossSplit, which uses a pair of neural
networks trained on two disjoint parts of the labelled dataset. CrossSplit
combines two main ingredients: (i) Cross-split label correction. The idea is
that, since the model trained on one part of the data cannot memorize
example-label pairs from the other part, the training labels presented to each
network can be smoothly adjusted by using the predictions of its peer network;
(ii) Cross-split semi-supervised training. A network trained on one part of the
data also uses the unlabeled inputs of the other part. Extensive experiments on
CIFAR-10, CIFAR-100, Tiny-ImageNet and mini-WebVision datasets demonstrate that
our method can outperform the current state-of-the-art in a wide range of noise
ratios.
| no_new_dataset | 0.70878 |
2212.11477 | Mertcan Cokbas | Mertcan Cokbas, Prakash Ishwar, Janusz Konrad | Spatio-Visual Fusion-Based Person Re-Identification for Overhead Fisheye
Images | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Person re-identification (PRID) has been thoroughly researched in typical
surveillance scenarios where various scenes are monitored by side-mounted,
rectilinear-lens cameras. To date, few methods have been proposed for fisheye
cameras mounted overhead and their performance is lacking. In order to close
this performance gap, we propose a multi-feature framework for fisheye PRID
where we combine deep-learning, color-based and location-based features by
means of novel feature fusion. We evaluate the performance of our framework for
various feature combinations on FRIDA, a public fisheye PRID dataset. The
results demonstrate that our multi-feature approach outperforms recent
appearance-based deep-learning methods by almost 18% points and location-based
methods by almost 3% points in matching accuracy. We also demonstrate the
potential application of the proposed PRID framework to people counting in
large, crowded indoor spaces.
| [
{
"version": "v1",
"created": "Thu, 22 Dec 2022 04:21:34 GMT"
},
{
"version": "v2",
"created": "Tue, 25 Apr 2023 20:16:01 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Cokbas",
"Mertcan",
""
],
[
"Ishwar",
"Prakash",
""
],
[
"Konrad",
"Janusz",
""
]
] | TITLE: Spatio-Visual Fusion-Based Person Re-Identification for Overhead Fisheye
Images
ABSTRACT: Person re-identification (PRID) has been thoroughly researched in typical
surveillance scenarios where various scenes are monitored by side-mounted,
rectilinear-lens cameras. To date, few methods have been proposed for fisheye
cameras mounted overhead and their performance is lacking. In order to close
this performance gap, we propose a multi-feature framework for fisheye PRID
where we combine deep-learning, color-based and location-based features by
means of novel feature fusion. We evaluate the performance of our framework for
various feature combinations on FRIDA, a public fisheye PRID dataset. The
results demonstrate that our multi-feature approach outperforms recent
appearance-based deep-learning methods by almost 18% points and location-based
methods by almost 3% points in matching accuracy. We also demonstrate the
potential application of the proposed PRID framework to people counting in
large, crowded indoor spaces.
| no_new_dataset | 0.706634 |
2212.13971 | Niloufar Delfan | Niloufar Delfan, Hamid Abrishami Moghaddam, Mohammadreza Modaresi,
Kimia Afshari, Kasra Nezamabadi, Neda Pak, Omid Ghaemi, Mohamad Forouzanfar | CT-LungNet: A Deep Learning Framework for Precise Lung Tissue
Segmentation in 3D Thoracic CT Scans | null | null | null | null | eess.IV cs.CV cs.NE | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Segmentation of lung tissue in computed tomography (CT) images is a precursor
to most pulmonary image analysis applications. Semantic segmentation methods
using deep learning have exhibited top-tier performance in recent years,
however designing accurate and robust segmentation models for lung tissue is
challenging due to the variations in shape, size, and orientation.
Additionally, medical image artifacts and noise can affect lung tissue
segmentation and degrade the accuracy of downstream analysis. The practicality
of current deep learning methods for lung tissue segmentation is limited as
they require significant computational resources and may not be easily
deployable in clinical settings. This paper presents a fully automatic method
that identifies the lungs in three-dimensional (3D) pulmonary CT images using
deep networks and transfer learning. We introduce (1) a novel 2.5-dimensional
image representation from consecutive CT slices that succinctly represents
volumetric information and (2) a U-Net architecture equipped with pre-trained
InceptionV3 blocks to segment 3D CT scans while maintaining the number of
learnable parameters as low as possible. Our method was quantitatively assessed
using one public dataset, LUNA16, for training and testing and two public
datasets, namely, VESSEL12 and CRPF, only for testing. Due to the low number of
learnable parameters, our method achieved high generalizability to the unseen
VESSEL12 and CRPF datasets while obtaining superior performance over Luna16
compared to existing methods (Dice coefficients of 99.7, 99.1, and 98.8 over
LUNA16, VESSEL12, and CRPF datasets, respectively). We made our method publicly
accessible via a graphical user interface at medvispy.ee.kntu.ac.ir.
| [
{
"version": "v1",
"created": "Wed, 28 Dec 2022 17:37:08 GMT"
},
{
"version": "v2",
"created": "Sat, 15 Apr 2023 20:54:40 GMT"
},
{
"version": "v3",
"created": "Tue, 18 Apr 2023 07:43:51 GMT"
},
{
"version": "v4",
"created": "Wed, 26 Apr 2023 14:57:11 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Delfan",
"Niloufar",
""
],
[
"Moghaddam",
"Hamid Abrishami",
""
],
[
"Modaresi",
"Mohammadreza",
""
],
[
"Afshari",
"Kimia",
""
],
[
"Nezamabadi",
"Kasra",
""
],
[
"Pak",
"Neda",
""
],
[
"Ghaemi",
"Omid",
""
],
[
"Forouzanfar",
"Mohamad",
""
]
] | TITLE: CT-LungNet: A Deep Learning Framework for Precise Lung Tissue
Segmentation in 3D Thoracic CT Scans
ABSTRACT: Segmentation of lung tissue in computed tomography (CT) images is a precursor
to most pulmonary image analysis applications. Semantic segmentation methods
using deep learning have exhibited top-tier performance in recent years,
however designing accurate and robust segmentation models for lung tissue is
challenging due to the variations in shape, size, and orientation.
Additionally, medical image artifacts and noise can affect lung tissue
segmentation and degrade the accuracy of downstream analysis. The practicality
of current deep learning methods for lung tissue segmentation is limited as
they require significant computational resources and may not be easily
deployable in clinical settings. This paper presents a fully automatic method
that identifies the lungs in three-dimensional (3D) pulmonary CT images using
deep networks and transfer learning. We introduce (1) a novel 2.5-dimensional
image representation from consecutive CT slices that succinctly represents
volumetric information and (2) a U-Net architecture equipped with pre-trained
InceptionV3 blocks to segment 3D CT scans while maintaining the number of
learnable parameters as low as possible. Our method was quantitatively assessed
using one public dataset, LUNA16, for training and testing and two public
datasets, namely, VESSEL12 and CRPF, only for testing. Due to the low number of
learnable parameters, our method achieved high generalizability to the unseen
VESSEL12 and CRPF datasets while obtaining superior performance over Luna16
compared to existing methods (Dice coefficients of 99.7, 99.1, and 98.8 over
LUNA16, VESSEL12, and CRPF datasets, respectively). We made our method publicly
accessible via a graphical user interface at medvispy.ee.kntu.ac.ir.
| no_new_dataset | 0.709837 |
2301.02053 | Yanhao Wang | Yanhao Wang and Michael Mathioudakis and Jia Li and Francesco Fabbri | Max-Min Diversification with Fairness Constraints: Exact and
Approximation Algorithms | 13 pages, 8 figures, to appear in SDM '23 | null | 10.1137/1.9781611977653.ch11 | null | cs.DS cs.CY cs.LG | http://creativecommons.org/licenses/by/4.0/ | Diversity maximization aims to select a diverse and representative subset of
items from a large dataset. It is a fundamental optimization task that finds
applications in data summarization, feature selection, web search, recommender
systems, and elsewhere. However, in a setting where data items are associated
with different groups according to sensitive attributes like sex or race, it is
possible that algorithmic solutions for this task, if left unchecked, will
under- or over-represent some of the groups. Therefore, we are motivated to
address the problem of \emph{max-min diversification with fairness
constraints}, aiming to select $k$ items to maximize the minimum distance
between any pair of selected items while ensuring that the number of items
selected from each group falls within predefined lower and upper bounds. In
this work, we propose an exact algorithm based on integer linear programming
that is suitable for small datasets as well as a
$\frac{1-\varepsilon}{5}$-approximation algorithm for any $\varepsilon \in (0,
1)$ that scales to large datasets. Extensive experiments on real-world datasets
demonstrate the superior performance of our proposed algorithms over existing
ones.
| [
{
"version": "v1",
"created": "Thu, 5 Jan 2023 13:02:35 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Wang",
"Yanhao",
""
],
[
"Mathioudakis",
"Michael",
""
],
[
"Li",
"Jia",
""
],
[
"Fabbri",
"Francesco",
""
]
] | TITLE: Max-Min Diversification with Fairness Constraints: Exact and
Approximation Algorithms
ABSTRACT: Diversity maximization aims to select a diverse and representative subset of
items from a large dataset. It is a fundamental optimization task that finds
applications in data summarization, feature selection, web search, recommender
systems, and elsewhere. However, in a setting where data items are associated
with different groups according to sensitive attributes like sex or race, it is
possible that algorithmic solutions for this task, if left unchecked, will
under- or over-represent some of the groups. Therefore, we are motivated to
address the problem of \emph{max-min diversification with fairness
constraints}, aiming to select $k$ items to maximize the minimum distance
between any pair of selected items while ensuring that the number of items
selected from each group falls within predefined lower and upper bounds. In
this work, we propose an exact algorithm based on integer linear programming
that is suitable for small datasets as well as a
$\frac{1-\varepsilon}{5}$-approximation algorithm for any $\varepsilon \in (0,
1)$ that scales to large datasets. Extensive experiments on real-world datasets
demonstrate the superior performance of our proposed algorithms over existing
ones.
| no_new_dataset | 0.706178 |
2301.07927 | Shuzhen Rao | Shuzhen Rao, Jun Huang, Zengming Tang | Exploiting Style Transfer-based Task Augmentation for Cross-Domain
Few-Shot Learning | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In cross-domain few-shot learning, the core issue is that the model trained
on source domains struggles to generalize to the target domain, especially when
the domain shift is large. Motivated by the observation that the domain shift
between training tasks and target tasks usually can reflect in their style
variation, we propose Task Augmented Meta-Learning (TAML) to conduct style
transfer-based task augmentation to improve the domain generalization ability.
Firstly, Multi-task Interpolation (MTI) is introduced to fuse features from
multiple tasks with different styles, which makes more diverse styles
available. Furthermore, a novel task-augmentation strategy called Multi-Task
Style Transfer (MTST) is proposed to perform style transfer on existing tasks
to learn discriminative style-independent features. We also introduce a Feature
Modulation module (FM) to add random styles and improve generalization of the
model. The proposed TAML increases the diversity of styles of training tasks,
and contributes to training a model with better domain generalization ability.
The effectiveness is demonstrated via theoretical analysis and thorough
experiments on two popular cross-domain few-shot benchmarks.
| [
{
"version": "v1",
"created": "Thu, 19 Jan 2023 07:32:23 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 01:44:28 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Rao",
"Shuzhen",
""
],
[
"Huang",
"Jun",
""
],
[
"Tang",
"Zengming",
""
]
] | TITLE: Exploiting Style Transfer-based Task Augmentation for Cross-Domain
Few-Shot Learning
ABSTRACT: In cross-domain few-shot learning, the core issue is that the model trained
on source domains struggles to generalize to the target domain, especially when
the domain shift is large. Motivated by the observation that the domain shift
between training tasks and target tasks usually can reflect in their style
variation, we propose Task Augmented Meta-Learning (TAML) to conduct style
transfer-based task augmentation to improve the domain generalization ability.
Firstly, Multi-task Interpolation (MTI) is introduced to fuse features from
multiple tasks with different styles, which makes more diverse styles
available. Furthermore, a novel task-augmentation strategy called Multi-Task
Style Transfer (MTST) is proposed to perform style transfer on existing tasks
to learn discriminative style-independent features. We also introduce a Feature
Modulation module (FM) to add random styles and improve generalization of the
model. The proposed TAML increases the diversity of styles of training tasks,
and contributes to training a model with better domain generalization ability.
The effectiveness is demonstrated via theoretical analysis and thorough
experiments on two popular cross-domain few-shot benchmarks.
| no_new_dataset | 0.708244 |
2301.12073 | Zhixuan Liu | Zhixuan Liu, Youeun Shin, Beverley-Claire Okogwu, Youngsik Yun, Lia
Coleman, Peter Schaldenbrand, Jihie Kim, Jean Oh | Towards Equitable Representation in Text-to-Image Synthesis Models with
the Cross-Cultural Understanding Benchmark (CCUB) Dataset | Still on going work | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | It has been shown that accurate representation in media improves the
well-being of the people who consume it. By contrast, inaccurate
representations can negatively affect viewers and lead to harmful perceptions
of other cultures. To achieve inclusive representation in generated images, we
propose a culturally-aware priming approach for text-to-image synthesis using a
small but culturally curated dataset that we collected, known here as
Cross-Cultural Understanding Benchmark (CCUB) Dataset, to fight the bias
prevalent in giant datasets. Our proposed approach is comprised of two
fine-tuning techniques: (1) Adding visual context via fine-tuning a pre-trained
text-to-image synthesis model, Stable Diffusion, on the CCUB text-image pairs,
and (2) Adding semantic context via automated prompt engineering using the
fine-tuned large language model, GPT-3, trained on our CCUB culturally-aware
text data. CCUB dataset is curated and our approach is evaluated by people who
have a personal relationship with that particular culture. Our experiments
indicate that priming using both text and image is effective in improving the
cultural relevance and decreasing the offensiveness of generated images while
maintaining quality.
| [
{
"version": "v1",
"created": "Sat, 28 Jan 2023 03:10:33 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 15:41:05 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Liu",
"Zhixuan",
""
],
[
"Shin",
"Youeun",
""
],
[
"Okogwu",
"Beverley-Claire",
""
],
[
"Yun",
"Youngsik",
""
],
[
"Coleman",
"Lia",
""
],
[
"Schaldenbrand",
"Peter",
""
],
[
"Kim",
"Jihie",
""
],
[
"Oh",
"Jean",
""
]
] | TITLE: Towards Equitable Representation in Text-to-Image Synthesis Models with
the Cross-Cultural Understanding Benchmark (CCUB) Dataset
ABSTRACT: It has been shown that accurate representation in media improves the
well-being of the people who consume it. By contrast, inaccurate
representations can negatively affect viewers and lead to harmful perceptions
of other cultures. To achieve inclusive representation in generated images, we
propose a culturally-aware priming approach for text-to-image synthesis using a
small but culturally curated dataset that we collected, known here as
Cross-Cultural Understanding Benchmark (CCUB) Dataset, to fight the bias
prevalent in giant datasets. Our proposed approach is comprised of two
fine-tuning techniques: (1) Adding visual context via fine-tuning a pre-trained
text-to-image synthesis model, Stable Diffusion, on the CCUB text-image pairs,
and (2) Adding semantic context via automated prompt engineering using the
fine-tuned large language model, GPT-3, trained on our CCUB culturally-aware
text data. CCUB dataset is curated and our approach is evaluated by people who
have a personal relationship with that particular culture. Our experiments
indicate that priming using both text and image is effective in improving the
cultural relevance and decreasing the offensiveness of generated images while
maintaining quality.
| new_dataset | 0.715374 |
2302.08635 | Shayan Shirahmad Gale Bagi | Shayan Shirahmad Gale Bagi, Zahra Gharaee, Oliver Schulte, Mark
Crowley | Generative Causal Representation Learning for Out-of-Distribution Motion
Forecasting | null | null | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Conventional supervised learning methods typically assume i.i.d samples and
are found to be sensitive to out-of-distribution (OOD) data. We propose
Generative Causal Representation Learning (GCRL) which leverages causality to
facilitate knowledge transfer under distribution shifts. While we evaluate the
effectiveness of our proposed method in human trajectory prediction models,
GCRL can be applied to other domains as well. First, we propose a novel causal
model that explains the generative factors in motion forecasting datasets using
features that are common across all environments and with features that are
specific to each environment. Selection variables are used to determine which
parts of the model can be directly transferred to a new environment without
fine-tuning. Second, we propose an end-to-end variational learning paradigm to
learn the causal mechanisms that generate observations from features. GCRL is
supported by strong theoretical results that imply identifiability of the
causal model under certain assumptions. Experimental results on synthetic and
real-world motion forecasting datasets show the robustness and effectiveness of
our proposed method for knowledge transfer under zero-shot and low-shot
settings by substantially outperforming the prior motion forecasting models on
out-of-distribution prediction. Our code is available at
https://github.com/sshirahmad/GCRL.
| [
{
"version": "v1",
"created": "Fri, 17 Feb 2023 00:30:44 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 02:49:06 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Bagi",
"Shayan Shirahmad Gale",
""
],
[
"Gharaee",
"Zahra",
""
],
[
"Schulte",
"Oliver",
""
],
[
"Crowley",
"Mark",
""
]
] | TITLE: Generative Causal Representation Learning for Out-of-Distribution Motion
Forecasting
ABSTRACT: Conventional supervised learning methods typically assume i.i.d samples and
are found to be sensitive to out-of-distribution (OOD) data. We propose
Generative Causal Representation Learning (GCRL) which leverages causality to
facilitate knowledge transfer under distribution shifts. While we evaluate the
effectiveness of our proposed method in human trajectory prediction models,
GCRL can be applied to other domains as well. First, we propose a novel causal
model that explains the generative factors in motion forecasting datasets using
features that are common across all environments and with features that are
specific to each environment. Selection variables are used to determine which
parts of the model can be directly transferred to a new environment without
fine-tuning. Second, we propose an end-to-end variational learning paradigm to
learn the causal mechanisms that generate observations from features. GCRL is
supported by strong theoretical results that imply identifiability of the
causal model under certain assumptions. Experimental results on synthetic and
real-world motion forecasting datasets show the robustness and effectiveness of
our proposed method for knowledge transfer under zero-shot and low-shot
settings by substantially outperforming the prior motion forecasting models on
out-of-distribution prediction. Our code is available at
https://github.com/sshirahmad/GCRL.
| no_new_dataset | 0.70978 |
2303.17202 | Kun-Ting Chen | Kun-Ting Chen, Arnaud Prouzeau, Joshua Langmead, Ryan T
Whitelock-Jones, Lee Lawrence, Tim Dwyer, Christophe Hurter, Daniel Weiskopf,
Sarah Goodwin | Gazealytics: A Unified and Flexible Visual Toolkit for Exploratory and
Comparative Gaze Analysis | null | null | 10.1145/3588015.3589844 | null | cs.HC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a novel, web-based visual eye-tracking analytics tool called
Gazealytics. Our open-source toolkit features a unified combination of gaze
analytics features that support flexible exploratory analysis, along with
annotation of areas of interest (AOI) and filter options based on multiple
criteria to visually analyse eye tracking data across time and space.
Gazealytics features coordinated views unifying spatiotemporal exploration of
fixations and scanpaths for various analytical tasks. A novel matrix
representation allows analysis of relationships between such spatial or
temporal features. Data can be grouped across samples, user-defined AOIs or
time windows of interest (TWIs) to support aggregate or filtered analysis of
gaze activity. This approach exceeds the capabilities of existing systems by
supporting flexible comparison between and within subjects, hypothesis
generation, data analysis and communication of insights. We demonstrate in a
walkthrough that Gazealytics supports multiple types of eye tracking datasets
and analytical tasks.
| [
{
"version": "v1",
"created": "Thu, 30 Mar 2023 07:52:38 GMT"
},
{
"version": "v2",
"created": "Fri, 7 Apr 2023 12:40:43 GMT"
},
{
"version": "v3",
"created": "Wed, 26 Apr 2023 15:25:33 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Chen",
"Kun-Ting",
""
],
[
"Prouzeau",
"Arnaud",
""
],
[
"Langmead",
"Joshua",
""
],
[
"Whitelock-Jones",
"Ryan T",
""
],
[
"Lawrence",
"Lee",
""
],
[
"Dwyer",
"Tim",
""
],
[
"Hurter",
"Christophe",
""
],
[
"Weiskopf",
"Daniel",
""
],
[
"Goodwin",
"Sarah",
""
]
] | TITLE: Gazealytics: A Unified and Flexible Visual Toolkit for Exploratory and
Comparative Gaze Analysis
ABSTRACT: We present a novel, web-based visual eye-tracking analytics tool called
Gazealytics. Our open-source toolkit features a unified combination of gaze
analytics features that support flexible exploratory analysis, along with
annotation of areas of interest (AOI) and filter options based on multiple
criteria to visually analyse eye tracking data across time and space.
Gazealytics features coordinated views unifying spatiotemporal exploration of
fixations and scanpaths for various analytical tasks. A novel matrix
representation allows analysis of relationships between such spatial or
temporal features. Data can be grouped across samples, user-defined AOIs or
time windows of interest (TWIs) to support aggregate or filtered analysis of
gaze activity. This approach exceeds the capabilities of existing systems by
supporting flexible comparison between and within subjects, hypothesis
generation, data analysis and communication of insights. We demonstrate in a
walkthrough that Gazealytics supports multiple types of eye tracking datasets
and analytical tasks.
| no_new_dataset | 0.703919 |
2303.17349 | Basuraj Bhowmik | Satyam Panda, Sanghamitra Das, Basuraj Bhowmik, and Budhaditya Hazra | Mastering Complex Modes: A New Method for Real-Time Modal Identification
of Vibrating Systems | 18 pages, journal article | null | null | null | eess.SY cs.SY | http://creativecommons.org/licenses/by/4.0/ | A novel algorithm for real-time modal identification in linear vibrating
systems with complex modes is introduced, utilizing a combination of first
order eigen-perturbation and second order separation techniques. In practical
settings, structures with complex modes are frequently encountered and their
presence often poses a challenge in accurately estimating the source signal in
real-time. The proposed methodology addresses this issue by incorporating the
right angle phase shift of the response in the sensor output and updating the
second order statistics of the complex response through first order
eigen-perturbation. Empirical evidence of the efficacy of the technique is
demonstrated through numerical case studies and validation using various
numerically modeled systems, as well as a standard ASCE-SHM benchmark problem
with complex modes, highlighting the capability of the proposed method to
achieve precise real-time modal property identification and online source
separation with a minimal number of initially required batch data.
| [
{
"version": "v1",
"created": "Thu, 30 Mar 2023 13:09:04 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 04:45:58 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Panda",
"Satyam",
""
],
[
"Das",
"Sanghamitra",
""
],
[
"Bhowmik",
"Basuraj",
""
],
[
"Hazra",
"Budhaditya",
""
]
] | TITLE: Mastering Complex Modes: A New Method for Real-Time Modal Identification
of Vibrating Systems
ABSTRACT: A novel algorithm for real-time modal identification in linear vibrating
systems with complex modes is introduced, utilizing a combination of first
order eigen-perturbation and second order separation techniques. In practical
settings, structures with complex modes are frequently encountered and their
presence often poses a challenge in accurately estimating the source signal in
real-time. The proposed methodology addresses this issue by incorporating the
right angle phase shift of the response in the sensor output and updating the
second order statistics of the complex response through first order
eigen-perturbation. Empirical evidence of the efficacy of the technique is
demonstrated through numerical case studies and validation using various
numerically modeled systems, as well as a standard ASCE-SHM benchmark problem
with complex modes, highlighting the capability of the proposed method to
achieve precise real-time modal property identification and online source
separation with a minimal number of initially required batch data.
| no_new_dataset | 0.709636 |
2304.02813 | Pengyuan Lu | Pengyuan Lu and Ivan Ruchkin and Matthew Cleaveland and Oleg Sokolsky
and Insup Lee | Causal Repair of Learning-enabled Cyber-physical Systems | null | null | null | null | eess.SY cs.LG cs.LO cs.SY | http://creativecommons.org/licenses/by/4.0/ | Models of actual causality leverage domain knowledge to generate convincing
diagnoses of events that caused an outcome. It is promising to apply these
models to diagnose and repair run-time property violations in cyber-physical
systems (CPS) with learning-enabled components (LEC). However, given the high
diversity and complexity of LECs, it is challenging to encode domain knowledge
(e.g., the CPS dynamics) in a scalable actual causality model that could
generate useful repair suggestions. In this paper, we focus causal diagnosis on
the input/output behaviors of LECs. Specifically, we aim to identify which
subset of I/O behaviors of the LEC is an actual cause for a property violation.
An important by-product is a counterfactual version of the LEC that repairs the
run-time property by fixing the identified problematic behaviors. Based on this
insights, we design a two-step diagnostic pipeline: (1) construct and
Halpern-Pearl causality model that reflects the dependency of property outcome
on the component's I/O behaviors, and (2) perform a search for an actual cause
and corresponding repair on the model. We prove that our pipeline has the
following guarantee: if an actual cause is found, the system is guaranteed to
be repaired; otherwise, we have high probabilistic confidence that the LEC
under analysis did not cause the property violation. We demonstrate that our
approach successfully repairs learned controllers on a standard OpenAI Gym
benchmark.
| [
{
"version": "v1",
"created": "Thu, 6 Apr 2023 01:26:58 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 14:15:21 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Lu",
"Pengyuan",
""
],
[
"Ruchkin",
"Ivan",
""
],
[
"Cleaveland",
"Matthew",
""
],
[
"Sokolsky",
"Oleg",
""
],
[
"Lee",
"Insup",
""
]
] | TITLE: Causal Repair of Learning-enabled Cyber-physical Systems
ABSTRACT: Models of actual causality leverage domain knowledge to generate convincing
diagnoses of events that caused an outcome. It is promising to apply these
models to diagnose and repair run-time property violations in cyber-physical
systems (CPS) with learning-enabled components (LEC). However, given the high
diversity and complexity of LECs, it is challenging to encode domain knowledge
(e.g., the CPS dynamics) in a scalable actual causality model that could
generate useful repair suggestions. In this paper, we focus causal diagnosis on
the input/output behaviors of LECs. Specifically, we aim to identify which
subset of I/O behaviors of the LEC is an actual cause for a property violation.
An important by-product is a counterfactual version of the LEC that repairs the
run-time property by fixing the identified problematic behaviors. Based on this
insights, we design a two-step diagnostic pipeline: (1) construct and
Halpern-Pearl causality model that reflects the dependency of property outcome
on the component's I/O behaviors, and (2) perform a search for an actual cause
and corresponding repair on the model. We prove that our pipeline has the
following guarantee: if an actual cause is found, the system is guaranteed to
be repaired; otherwise, we have high probabilistic confidence that the LEC
under analysis did not cause the property violation. We demonstrate that our
approach successfully repairs learned controllers on a standard OpenAI Gym
benchmark.
| no_new_dataset | 0.709057 |
2304.03274 | Jiawei Ren | Jiawei Ren, Cunjun Yu, Siwei Chen, Xiao Ma, Liang Pan, Ziwei Liu | DiffMimic: Efficient Motion Mimicking with Differentiable Physics | ICLR 2023 Code is at https://github.com/jiawei-ren/diffmimic Project
page is at https://diffmimic.github.io/ | null | null | null | cs.CV cs.AI cs.GR cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Motion mimicking is a foundational task in physics-based character animation.
However, most existing motion mimicking methods are built upon reinforcement
learning (RL) and suffer from heavy reward engineering, high variance, and slow
convergence with hard explorations. Specifically, they usually take tens of
hours or even days of training to mimic a simple motion sequence, resulting in
poor scalability. In this work, we leverage differentiable physics simulators
(DPS) and propose an efficient motion mimicking method dubbed DiffMimic. Our
key insight is that DPS casts a complex policy learning task to a much simpler
state matching problem. In particular, DPS learns a stable policy by analytical
gradients with ground-truth physical priors hence leading to significantly
faster and stabler convergence than RL-based methods. Moreover, to escape from
local optima, we utilize a Demonstration Replay mechanism to enable stable
gradient backpropagation in a long horizon. Extensive experiments on standard
benchmarks show that DiffMimic has a better sample efficiency and time
efficiency than existing methods (e.g., DeepMimic). Notably, DiffMimic allows a
physically simulated character to learn Backflip after 10 minutes of training
and be able to cycle it after 3 hours of training, while the existing approach
may require about a day of training to cycle Backflip. More importantly, we
hope DiffMimic can benefit more differentiable animation systems with
techniques like differentiable clothes simulation in future research.
| [
{
"version": "v1",
"created": "Thu, 6 Apr 2023 17:56:22 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 06:29:21 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Ren",
"Jiawei",
""
],
[
"Yu",
"Cunjun",
""
],
[
"Chen",
"Siwei",
""
],
[
"Ma",
"Xiao",
""
],
[
"Pan",
"Liang",
""
],
[
"Liu",
"Ziwei",
""
]
] | TITLE: DiffMimic: Efficient Motion Mimicking with Differentiable Physics
ABSTRACT: Motion mimicking is a foundational task in physics-based character animation.
However, most existing motion mimicking methods are built upon reinforcement
learning (RL) and suffer from heavy reward engineering, high variance, and slow
convergence with hard explorations. Specifically, they usually take tens of
hours or even days of training to mimic a simple motion sequence, resulting in
poor scalability. In this work, we leverage differentiable physics simulators
(DPS) and propose an efficient motion mimicking method dubbed DiffMimic. Our
key insight is that DPS casts a complex policy learning task to a much simpler
state matching problem. In particular, DPS learns a stable policy by analytical
gradients with ground-truth physical priors hence leading to significantly
faster and stabler convergence than RL-based methods. Moreover, to escape from
local optima, we utilize a Demonstration Replay mechanism to enable stable
gradient backpropagation in a long horizon. Extensive experiments on standard
benchmarks show that DiffMimic has a better sample efficiency and time
efficiency than existing methods (e.g., DeepMimic). Notably, DiffMimic allows a
physically simulated character to learn Backflip after 10 minutes of training
and be able to cycle it after 3 hours of training, while the existing approach
may require about a day of training to cycle Backflip. More importantly, we
hope DiffMimic can benefit more differentiable animation systems with
techniques like differentiable clothes simulation in future research.
| no_new_dataset | 0.707436 |
2304.04554 | Bin Liu | Luping Wang, Bin Liu | Use the Detection Transformer as a Data Augmenter | 15 pages | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Detection Transformer (DETR) is a Transformer architecture based object
detection model. In this paper, we demonstrate that it can also be used as a
data augmenter. We term our approach as DETR assisted CutMix, or DeMix for
short. DeMix builds on CutMix, a simple yet highly effective data augmentation
technique that has gained popularity in recent years. CutMix improves model
performance by cutting and pasting a patch from one image onto another,
yielding a new image. The corresponding label for this new example is specified
as the weighted average of the original labels, where the weight is
proportional to the area of the patch. CutMix selects a random patch to be cut.
In contrast, DeMix elaborately selects a semantically rich patch, located by a
pre-trained DETR. The label of the new image is specified in the same way as in
CutMix. Experimental results on benchmark datasets for image classification
demonstrate that DeMix significantly outperforms prior art data augmentation
methods including CutMix. Oue code is available at
https://github.com/ZJLAB-AMMI/DeMix.
| [
{
"version": "v1",
"created": "Mon, 10 Apr 2023 12:50:17 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 02:44:40 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Wang",
"Luping",
""
],
[
"Liu",
"Bin",
""
]
] | TITLE: Use the Detection Transformer as a Data Augmenter
ABSTRACT: Detection Transformer (DETR) is a Transformer architecture based object
detection model. In this paper, we demonstrate that it can also be used as a
data augmenter. We term our approach as DETR assisted CutMix, or DeMix for
short. DeMix builds on CutMix, a simple yet highly effective data augmentation
technique that has gained popularity in recent years. CutMix improves model
performance by cutting and pasting a patch from one image onto another,
yielding a new image. The corresponding label for this new example is specified
as the weighted average of the original labels, where the weight is
proportional to the area of the patch. CutMix selects a random patch to be cut.
In contrast, DeMix elaborately selects a semantically rich patch, located by a
pre-trained DETR. The label of the new image is specified in the same way as in
CutMix. Experimental results on benchmark datasets for image classification
demonstrate that DeMix significantly outperforms prior art data augmentation
methods including CutMix. Oue code is available at
https://github.com/ZJLAB-AMMI/DeMix.
| no_new_dataset | 0.71081 |
2304.04839 | Md Ishtyaq Mahmud | Jonayet Miah, Muntasir Mamun, Md Minhazur Rahman, Md Ishtyaq Mahmud,
Sabbir Ahmed, Md Hasan Bin Nasir | MHfit: Mobile Health Data for Predicting Athletics Fitness Using Machine
Learning | 6, Accepted by 2nd International Seminar on Machine Learning,
Optimization, and Data Science (ISMODE) | null | null | null | cs.LG cs.CY | http://creativecommons.org/licenses/by/4.0/ | Mobile phones and other electronic gadgets or devices have aided in
collecting data without the need for data entry. This paper will specifically
focus on Mobile health data. Mobile health data use mobile devices to gather
clinical health data and track patient vitals in real-time. Our study is aimed
to give decisions for small or big sports teams on whether one athlete good fit
or not for a particular game with the compare several machine learning
algorithms to predict human behavior and health using the data collected from
mobile devices and sensors placed on patients. In this study, we have obtained
the dataset from a similar study done on mhealth. The dataset contains vital
signs recordings of ten volunteers from different backgrounds. They had to
perform several physical activities with a sensor placed on their bodies. Our
study used 5 machine learning algorithms (XGBoost, Naive Bayes, Decision Tree,
Random Forest, and Logistic Regression) to analyze and predict human health
behavior. XGBoost performed better compared to the other machine learning
algorithms and achieved 95.2% accuracy, 99.5% in sensitivity, 99.5% in
specificity, and 99.66% in F1 score. Our research indicated a promising future
in mhealth being used to predict human behavior and further research and
exploration need to be done for it to be available for commercial use
specifically in the sports industry.
| [
{
"version": "v1",
"created": "Mon, 10 Apr 2023 19:53:03 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 17:11:22 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Miah",
"Jonayet",
""
],
[
"Mamun",
"Muntasir",
""
],
[
"Rahman",
"Md Minhazur",
""
],
[
"Mahmud",
"Md Ishtyaq",
""
],
[
"Ahmed",
"Sabbir",
""
],
[
"Nasir",
"Md Hasan Bin",
""
]
] | TITLE: MHfit: Mobile Health Data for Predicting Athletics Fitness Using Machine
Learning
ABSTRACT: Mobile phones and other electronic gadgets or devices have aided in
collecting data without the need for data entry. This paper will specifically
focus on Mobile health data. Mobile health data use mobile devices to gather
clinical health data and track patient vitals in real-time. Our study is aimed
to give decisions for small or big sports teams on whether one athlete good fit
or not for a particular game with the compare several machine learning
algorithms to predict human behavior and health using the data collected from
mobile devices and sensors placed on patients. In this study, we have obtained
the dataset from a similar study done on mhealth. The dataset contains vital
signs recordings of ten volunteers from different backgrounds. They had to
perform several physical activities with a sensor placed on their bodies. Our
study used 5 machine learning algorithms (XGBoost, Naive Bayes, Decision Tree,
Random Forest, and Logistic Regression) to analyze and predict human health
behavior. XGBoost performed better compared to the other machine learning
algorithms and achieved 95.2% accuracy, 99.5% in sensitivity, 99.5% in
specificity, and 99.66% in F1 score. Our research indicated a promising future
in mhealth being used to predict human behavior and further research and
exploration need to be done for it to be available for commercial use
specifically in the sports industry.
| no_new_dataset | 0.690311 |
2304.06033 | Akara Supratak | Akara Supratak, Peter Haddawy | Quantifying the Impact of Data Characteristics on the Transferability of
Sleep Stage Scoring Models | null | null | 10.1016/j.artmed.2023.102540 | null | eess.SP cs.LG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Deep learning models for scoring sleep stages based on single-channel EEG
have been proposed as a promising method for remote sleep monitoring. However,
applying these models to new datasets, particularly from wearable devices,
raises two questions. First, when annotations on a target dataset are
unavailable, which different data characteristics affect the sleep stage
scoring performance the most and by how much? Second, when annotations are
available, which dataset should be used as the source of transfer learning to
optimize performance? In this paper, we propose a novel method for
computationally quantifying the impact of different data characteristics on the
transferability of deep learning models. Quantification is accomplished by
training and evaluating two models with significant architectural differences,
TinySleepNet and U-Time, under various transfer configurations in which the
source and target datasets have different recording channels, recording
environments, and subject conditions. For the first question, the environment
had the highest impact on sleep stage scoring performance, with performance
degrading by over 14% when sleep annotations were unavailable. For the second
question, the most useful transfer sources for TinySleepNet and the U-Time
models were MASS-SS1 and ISRUC-SG1, containing a high percentage of N1 (the
rarest sleep stage) relative to the others. The frontal and central EEGs were
preferred for TinySleepNet. The proposed approach enables full utilization of
existing sleep datasets for training and planning model transfer to maximize
the sleep stage scoring performance on a target problem when sleep annotations
are limited or unavailable, supporting the realization of remote sleep
monitoring.
| [
{
"version": "v1",
"created": "Tue, 28 Mar 2023 07:57:21 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Supratak",
"Akara",
""
],
[
"Haddawy",
"Peter",
""
]
] | TITLE: Quantifying the Impact of Data Characteristics on the Transferability of
Sleep Stage Scoring Models
ABSTRACT: Deep learning models for scoring sleep stages based on single-channel EEG
have been proposed as a promising method for remote sleep monitoring. However,
applying these models to new datasets, particularly from wearable devices,
raises two questions. First, when annotations on a target dataset are
unavailable, which different data characteristics affect the sleep stage
scoring performance the most and by how much? Second, when annotations are
available, which dataset should be used as the source of transfer learning to
optimize performance? In this paper, we propose a novel method for
computationally quantifying the impact of different data characteristics on the
transferability of deep learning models. Quantification is accomplished by
training and evaluating two models with significant architectural differences,
TinySleepNet and U-Time, under various transfer configurations in which the
source and target datasets have different recording channels, recording
environments, and subject conditions. For the first question, the environment
had the highest impact on sleep stage scoring performance, with performance
degrading by over 14% when sleep annotations were unavailable. For the second
question, the most useful transfer sources for TinySleepNet and the U-Time
models were MASS-SS1 and ISRUC-SG1, containing a high percentage of N1 (the
rarest sleep stage) relative to the others. The frontal and central EEGs were
preferred for TinySleepNet. The proposed approach enables full utilization of
existing sleep datasets for training and planning model transfer to maximize
the sleep stage scoring performance on a target problem when sleep annotations
are limited or unavailable, supporting the realization of remote sleep
monitoring.
| no_new_dataset | 0.711819 |
2304.06460 | Jonathan Lambert Mr | Jonathan Lambert and Kevin Casey and Rosemary Monahan | Repositioning Tiered HotSpot Execution Performance Relative to the
Interpreter | 17 pages | null | null | null | cs.PL cs.PF | http://creativecommons.org/licenses/by/4.0/ | Although the advantages of just-in-time compilation over traditional
interpretive execution are widely recognised, there needs to be more current
research investigating and repositioning the performance differences between
these two execution models relative to contemporary workloads. Specifically,
there is a need to examine the performance differences between Java Runtime
Environment (JRE) Java Virtual Machine (JVM) tiered execution and JRE JVM
interpretive execution relative to modern multicore architectures and modern
concurrent and parallel benchmark workloads. This article aims to fill this
research gap by presenting the results of a study that compares the performance
of these two execution models under load from the Renaissance Benchmark Suite.
This research is relevant to anyone interested in understanding the performance
differences between just-in-time compiled code and interpretive execution. It
provides a contemporary assessment of the interpretive JVM core, the entry and
starting point for bytecode execution, relative to just-in-time tiered
execution. The study considers factors such as the JRE version, the GNU GCC
version used in the JRE build toolchain, and the garbage collector algorithm
specified at runtime, and their impact on the performance difference envelope
between interpretive and tiered execution. Our findings indicate that tiered
execution is considerably more efficient than interpretive execution, and the
performance gap has increased, ranging from 4 to 37 times more efficient. On
average, tiered execution is approximately 15 times more efficient than
interpretive execution. Additionally, the performance differences between
interpretive and tiered execution are influenced by workload category, with
narrower performance differences observed for web-based workloads and more
significant differences for Functional and Scala-type workloads.
| [
{
"version": "v1",
"created": "Thu, 13 Apr 2023 12:55:31 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 10:29:33 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Lambert",
"Jonathan",
""
],
[
"Casey",
"Kevin",
""
],
[
"Monahan",
"Rosemary",
""
]
] | TITLE: Repositioning Tiered HotSpot Execution Performance Relative to the
Interpreter
ABSTRACT: Although the advantages of just-in-time compilation over traditional
interpretive execution are widely recognised, there needs to be more current
research investigating and repositioning the performance differences between
these two execution models relative to contemporary workloads. Specifically,
there is a need to examine the performance differences between Java Runtime
Environment (JRE) Java Virtual Machine (JVM) tiered execution and JRE JVM
interpretive execution relative to modern multicore architectures and modern
concurrent and parallel benchmark workloads. This article aims to fill this
research gap by presenting the results of a study that compares the performance
of these two execution models under load from the Renaissance Benchmark Suite.
This research is relevant to anyone interested in understanding the performance
differences between just-in-time compiled code and interpretive execution. It
provides a contemporary assessment of the interpretive JVM core, the entry and
starting point for bytecode execution, relative to just-in-time tiered
execution. The study considers factors such as the JRE version, the GNU GCC
version used in the JRE build toolchain, and the garbage collector algorithm
specified at runtime, and their impact on the performance difference envelope
between interpretive and tiered execution. Our findings indicate that tiered
execution is considerably more efficient than interpretive execution, and the
performance gap has increased, ranging from 4 to 37 times more efficient. On
average, tiered execution is approximately 15 times more efficient than
interpretive execution. Additionally, the performance differences between
interpretive and tiered execution are influenced by workload category, with
narrower performance differences observed for web-based workloads and more
significant differences for Functional and Scala-type workloads.
| no_new_dataset | 0.70557 |
2304.07051 | Jaime Spencer | Jaime Spencer, C. Stella Qian, Michaela Trescakova, Chris Russell,
Simon Hadfield, Erich W. Graf, Wendy J. Adams, Andrew J. Schofield, James
Elder, Richard Bowden, Ali Anwar, Hao Chen, Xiaozhi Chen, Kai Cheng, Yuchao
Dai, Huynh Thai Hoa, Sadat Hossain, Jianmian Huang, Mohan Jing, Bo Li, Chao
Li, Baojun Li, Zhiwen Liu, Stefano Mattoccia, Siegfried Mercelis, Myungwoo
Nam, Matteo Poggi, Xiaohua Qi, Jiahui Ren, Yang Tang, Fabio Tosi, Linh Trinh,
S. M. Nadim Uddin, Khan Muhammad Umair, Kaixuan Wang, Yufei Wang, Yixing
Wang, Mochu Xiang, Guangkai Xu, Wei Yin, Jun Yu, Qi Zhang, Chaoqiang Zhao | The Second Monocular Depth Estimation Challenge | Published at CVPRW2023 | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by-sa/4.0/ | This paper discusses the results for the second edition of the Monocular
Depth Estimation Challenge (MDEC). This edition was open to methods using any
form of supervision, including fully-supervised, self-supervised, multi-task or
proxy depth. The challenge was based around the SYNS-Patches dataset, which
features a wide diversity of environments with high-quality dense ground-truth.
This includes complex natural environments, e.g. forests or fields, which are
greatly underrepresented in current benchmarks.
The challenge received eight unique submissions that outperformed the
provided SotA baseline on any of the pointcloud- or image-based metrics. The
top supervised submission improved relative F-Score by 27.62%, while the top
self-supervised improved it by 16.61%. Supervised submissions generally
leveraged large collections of datasets to improve data diversity.
Self-supervised submissions instead updated the network architecture and
pretrained backbones. These results represent a significant progress in the
field, while highlighting avenues for future research, such as reducing
interpolation artifacts at depth boundaries, improving self-supervised indoor
performance and overall natural image accuracy.
| [
{
"version": "v1",
"created": "Fri, 14 Apr 2023 11:10:07 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Apr 2023 09:35:40 GMT"
},
{
"version": "v3",
"created": "Wed, 26 Apr 2023 12:28:16 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Spencer",
"Jaime",
""
],
[
"Qian",
"C. Stella",
""
],
[
"Trescakova",
"Michaela",
""
],
[
"Russell",
"Chris",
""
],
[
"Hadfield",
"Simon",
""
],
[
"Graf",
"Erich W.",
""
],
[
"Adams",
"Wendy J.",
""
],
[
"Schofield",
"Andrew J.",
""
],
[
"Elder",
"James",
""
],
[
"Bowden",
"Richard",
""
],
[
"Anwar",
"Ali",
""
],
[
"Chen",
"Hao",
""
],
[
"Chen",
"Xiaozhi",
""
],
[
"Cheng",
"Kai",
""
],
[
"Dai",
"Yuchao",
""
],
[
"Hoa",
"Huynh Thai",
""
],
[
"Hossain",
"Sadat",
""
],
[
"Huang",
"Jianmian",
""
],
[
"Jing",
"Mohan",
""
],
[
"Li",
"Bo",
""
],
[
"Li",
"Chao",
""
],
[
"Li",
"Baojun",
""
],
[
"Liu",
"Zhiwen",
""
],
[
"Mattoccia",
"Stefano",
""
],
[
"Mercelis",
"Siegfried",
""
],
[
"Nam",
"Myungwoo",
""
],
[
"Poggi",
"Matteo",
""
],
[
"Qi",
"Xiaohua",
""
],
[
"Ren",
"Jiahui",
""
],
[
"Tang",
"Yang",
""
],
[
"Tosi",
"Fabio",
""
],
[
"Trinh",
"Linh",
""
],
[
"Uddin",
"S. M. Nadim",
""
],
[
"Umair",
"Khan Muhammad",
""
],
[
"Wang",
"Kaixuan",
""
],
[
"Wang",
"Yufei",
""
],
[
"Wang",
"Yixing",
""
],
[
"Xiang",
"Mochu",
""
],
[
"Xu",
"Guangkai",
""
],
[
"Yin",
"Wei",
""
],
[
"Yu",
"Jun",
""
],
[
"Zhang",
"Qi",
""
],
[
"Zhao",
"Chaoqiang",
""
]
] | TITLE: The Second Monocular Depth Estimation Challenge
ABSTRACT: This paper discusses the results for the second edition of the Monocular
Depth Estimation Challenge (MDEC). This edition was open to methods using any
form of supervision, including fully-supervised, self-supervised, multi-task or
proxy depth. The challenge was based around the SYNS-Patches dataset, which
features a wide diversity of environments with high-quality dense ground-truth.
This includes complex natural environments, e.g. forests or fields, which are
greatly underrepresented in current benchmarks.
The challenge received eight unique submissions that outperformed the
provided SotA baseline on any of the pointcloud- or image-based metrics. The
top supervised submission improved relative F-Score by 27.62%, while the top
self-supervised improved it by 16.61%. Supervised submissions generally
leveraged large collections of datasets to improve data diversity.
Self-supervised submissions instead updated the network architecture and
pretrained backbones. These results represent a significant progress in the
field, while highlighting avenues for future research, such as reducing
interpolation artifacts at depth boundaries, improving self-supervised indoor
performance and overall natural image accuracy.
| no_new_dataset | 0.707777 |
2304.07258 | Benjamin Towle | Benjamin Towle and Ke Zhou | Learn What Is Possible, Then Choose What Is Best: Disentangling
One-To-Many Relations in Language Through Text-based Games | EMNLP Findings 2022 | null | null | null | cs.CL cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Language models pre-trained on large self-supervised corpora, followed by
task-specific fine-tuning has become the dominant paradigm in NLP. These
pre-training datasets often have a one-to-many structure--e.g. in dialogue
there are many valid responses for a given context. However, only some of these
responses will be desirable in our downstream task. This raises the question of
how we should train the model such that it can emulate the desirable
behaviours, but not the undesirable ones. Current approaches train in a
one-to-one setup--only a single target response is given for a single dialogue
context--leading to models only learning to predict the average response, while
ignoring the full range of possible responses. Using text-based games as a
testbed, our approach, PASA, uses discrete latent variables to capture the
range of different behaviours represented in our larger pre-training dataset.
We then use knowledge distillation to distil the posterior probability
distribution into a student model. This probability distribution is far richer
than learning from only the hard targets of the dataset, and thus allows the
student model to benefit from the richer range of actions the teacher model has
learned. Results show up to 49% empirical improvement over the previous
state-of-the-art model on the Jericho Walkthroughs dataset.
| [
{
"version": "v1",
"created": "Fri, 14 Apr 2023 17:11:26 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 10:35:13 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Towle",
"Benjamin",
""
],
[
"Zhou",
"Ke",
""
]
] | TITLE: Learn What Is Possible, Then Choose What Is Best: Disentangling
One-To-Many Relations in Language Through Text-based Games
ABSTRACT: Language models pre-trained on large self-supervised corpora, followed by
task-specific fine-tuning has become the dominant paradigm in NLP. These
pre-training datasets often have a one-to-many structure--e.g. in dialogue
there are many valid responses for a given context. However, only some of these
responses will be desirable in our downstream task. This raises the question of
how we should train the model such that it can emulate the desirable
behaviours, but not the undesirable ones. Current approaches train in a
one-to-one setup--only a single target response is given for a single dialogue
context--leading to models only learning to predict the average response, while
ignoring the full range of possible responses. Using text-based games as a
testbed, our approach, PASA, uses discrete latent variables to capture the
range of different behaviours represented in our larger pre-training dataset.
We then use knowledge distillation to distil the posterior probability
distribution into a student model. This probability distribution is far richer
than learning from only the hard targets of the dataset, and thus allows the
student model to benefit from the richer range of actions the teacher model has
learned. Results show up to 49% empirical improvement over the previous
state-of-the-art model on the Jericho Walkthroughs dataset.
| no_new_dataset | 0.710377 |
2304.09325 | Xilai Li | Xilai Li, Goeric Huybrechts, Srikanth Ronanki, Jeff Farris, Sravan
Bodapati | Dynamic Chunk Convolution for Unified Streaming and Non-Streaming
Conformer ASR | 5 pages, 3 figures, 2023 IEEE International Conference on Acoustics,
Speech and Signal Processing (ICASSP 2023) | null | null | null | eess.AS cs.SD | http://creativecommons.org/licenses/by/4.0/ | Recently, there has been an increasing interest in unifying streaming and
non-streaming speech recognition models to reduce development, training and
deployment cost. The best-known approaches rely on either window-based or
dynamic chunk-based attention strategy and causal convolutions to minimize the
degradation due to streaming. However, the performance gap still remains
relatively large between non-streaming and a full-contextual model trained
independently. To address this, we propose a dynamic chunk-based convolution
replacing the causal convolution in a hybrid Connectionist Temporal
Classification (CTC)-Attention Conformer architecture. Additionally, we
demonstrate further improvements through initialization of weights from a
full-contextual model and parallelization of the convolution and self-attention
modules. We evaluate our models on the open-source Voxpopuli, LibriSpeech and
in-house conversational datasets. Overall, our proposed model reduces the
degradation of the streaming mode over the non-streaming full-contextual model
from 41.7% and 45.7% to 16.7% and 26.2% on the LibriSpeech test-clean and
test-other datasets respectively, while improving by a relative 15.5% WER over
the previous state-of-the-art unified model.
| [
{
"version": "v1",
"created": "Tue, 18 Apr 2023 22:18:40 GMT"
},
{
"version": "v2",
"created": "Tue, 25 Apr 2023 19:36:32 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Li",
"Xilai",
""
],
[
"Huybrechts",
"Goeric",
""
],
[
"Ronanki",
"Srikanth",
""
],
[
"Farris",
"Jeff",
""
],
[
"Bodapati",
"Sravan",
""
]
] | TITLE: Dynamic Chunk Convolution for Unified Streaming and Non-Streaming
Conformer ASR
ABSTRACT: Recently, there has been an increasing interest in unifying streaming and
non-streaming speech recognition models to reduce development, training and
deployment cost. The best-known approaches rely on either window-based or
dynamic chunk-based attention strategy and causal convolutions to minimize the
degradation due to streaming. However, the performance gap still remains
relatively large between non-streaming and a full-contextual model trained
independently. To address this, we propose a dynamic chunk-based convolution
replacing the causal convolution in a hybrid Connectionist Temporal
Classification (CTC)-Attention Conformer architecture. Additionally, we
demonstrate further improvements through initialization of weights from a
full-contextual model and parallelization of the convolution and self-attention
modules. We evaluate our models on the open-source Voxpopuli, LibriSpeech and
in-house conversational datasets. Overall, our proposed model reduces the
degradation of the streaming mode over the non-streaming full-contextual model
from 41.7% and 45.7% to 16.7% and 26.2% on the LibriSpeech test-clean and
test-other datasets respectively, while improving by a relative 15.5% WER over
the previous state-of-the-art unified model.
| no_new_dataset | 0.708843 |
2304.10335 | Giulia Castagnolo | Giulia Castagnolo, Concetto Spampinato, Francesco Rundo, Daniela
Giordano, Simone Palazzo | A baseline on continual learning methods for video action recognition | null | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | Continual learning has recently attracted attention from the research
community, as it aims to solve long-standing limitations of classic
supervisedly-trained models. However, most research on this subject has tackled
continual learning in simple image classification scenarios. In this paper, we
present a benchmark of state-of-the-art continual learning methods on video
action recognition. Besides the increased complexity due to the temporal
dimension, the video setting imposes stronger requirements on computing
resources for top-performing rehearsal methods. To counteract the increased
memory requirements, we present two method-agnostic variants for rehearsal
methods, exploiting measures of either model confidence or data information to
select memorable samples. Our experiments show that, as expected from the
literature, rehearsal methods outperform other approaches; moreover, the
proposed memory-efficient variants are shown to be effective at retaining a
certain level of performance with a smaller buffer size.
| [
{
"version": "v1",
"created": "Thu, 20 Apr 2023 14:20:43 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 09:49:18 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Castagnolo",
"Giulia",
""
],
[
"Spampinato",
"Concetto",
""
],
[
"Rundo",
"Francesco",
""
],
[
"Giordano",
"Daniela",
""
],
[
"Palazzo",
"Simone",
""
]
] | TITLE: A baseline on continual learning methods for video action recognition
ABSTRACT: Continual learning has recently attracted attention from the research
community, as it aims to solve long-standing limitations of classic
supervisedly-trained models. However, most research on this subject has tackled
continual learning in simple image classification scenarios. In this paper, we
present a benchmark of state-of-the-art continual learning methods on video
action recognition. Besides the increased complexity due to the temporal
dimension, the video setting imposes stronger requirements on computing
resources for top-performing rehearsal methods. To counteract the increased
memory requirements, we present two method-agnostic variants for rehearsal
methods, exploiting measures of either model confidence or data information to
select memorable samples. Our experiments show that, as expected from the
literature, rehearsal methods outperform other approaches; moreover, the
proposed memory-efficient variants are shown to be effective at retaining a
certain level of performance with a smaller buffer size.
| no_new_dataset | 0.708023 |
2304.11244 | Robero Xavier Jimenez Jimbo | Roberto X. Jim\'enez J | Bilevel optimisation with embedded neural networks: Application to
scheduling and control integration | 18 pages | null | null | null | eess.SY cs.SY | http://creativecommons.org/licenses/by/4.0/ | Scheduling problems requires to explicitly account for control considerations
in their optimisation. The literature proposes two traditional ways to solve
this integrated problem: hierarchical and monolithic. The monolithic approach
ignores the control level's objective and incorporates it as a constraint into
the upper level at the cost of suboptimality. The hierarchical approach
requires solving a mathematically complex bilevel problem with the scheduling
acting as the leader and control as the follower. The linking variables between
both levels belong to a small subset of scheduling and control decision
variables. For this subset of variables, data-driven surrogate models have been
used to learn follower responses to different leader decisions. In this work,
we propose to use ReLU neural networks for the control level. Consequently, the
bilevel problem is collapsed into a single-level MILP that is still able to
account for the control level's objective. This single-level MILP reformulation
is compared with the monolithic approach and benchmarked against embedding a
nonlinear expression of the neural networks into the optimisation. Moreover, a
neural network is used to predict control level feasibility. The case studies
involve batch reactor and sequential batch process scheduling problems. The
proposed methodology finds optimal solutions while largely outperforming both
approaches in terms of computational time. Additionally, due to well-developed
MILP solvers, adding ReLU neural networks in a MILP form marginally impacts the
computational time. The solution's error due to prediction accuracy is
correlated with the neural network training error. Overall, we expose how - by
using an existing big-M reformulation and being careful about integrating
machine learning and optimisation pipelines - we can more efficiently solve the
bilevel scheduling-control problem with high accuracy.
| [
{
"version": "v1",
"created": "Fri, 21 Apr 2023 20:43:25 GMT"
},
{
"version": "v2",
"created": "Tue, 25 Apr 2023 21:10:58 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"J",
"Roberto X. Jiménez",
""
]
] | TITLE: Bilevel optimisation with embedded neural networks: Application to
scheduling and control integration
ABSTRACT: Scheduling problems requires to explicitly account for control considerations
in their optimisation. The literature proposes two traditional ways to solve
this integrated problem: hierarchical and monolithic. The monolithic approach
ignores the control level's objective and incorporates it as a constraint into
the upper level at the cost of suboptimality. The hierarchical approach
requires solving a mathematically complex bilevel problem with the scheduling
acting as the leader and control as the follower. The linking variables between
both levels belong to a small subset of scheduling and control decision
variables. For this subset of variables, data-driven surrogate models have been
used to learn follower responses to different leader decisions. In this work,
we propose to use ReLU neural networks for the control level. Consequently, the
bilevel problem is collapsed into a single-level MILP that is still able to
account for the control level's objective. This single-level MILP reformulation
is compared with the monolithic approach and benchmarked against embedding a
nonlinear expression of the neural networks into the optimisation. Moreover, a
neural network is used to predict control level feasibility. The case studies
involve batch reactor and sequential batch process scheduling problems. The
proposed methodology finds optimal solutions while largely outperforming both
approaches in terms of computational time. Additionally, due to well-developed
MILP solvers, adding ReLU neural networks in a MILP form marginally impacts the
computational time. The solution's error due to prediction accuracy is
correlated with the neural network training error. Overall, we expose how - by
using an existing big-M reformulation and being careful about integrating
machine learning and optimisation pipelines - we can more efficiently solve the
bilevel scheduling-control problem with high accuracy.
| no_new_dataset | 0.708231 |
2304.11370 | Haitao Li | Haitao Li, Qingyao Ai, Jia Chen, Qian Dong, Yueyue Wu, Yiqun Liu,
Chong Chen, Qi Tian | SAILER: Structure-aware Pre-trained Language Model for Legal Case
Retrieval | 10 pages, accepted by SIGIR 2023 | null | null | null | cs.IR cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Legal case retrieval, which aims to find relevant cases for a query case,
plays a core role in the intelligent legal system. Despite the success that
pre-training has achieved in ad-hoc retrieval tasks, effective pre-training
strategies for legal case retrieval remain to be explored. Compared with
general documents, legal case documents are typically long text sequences with
intrinsic logical structures. However, most existing language models have
difficulty understanding the long-distance dependencies between different
structures. Moreover, in contrast to the general retrieval, the relevance in
the legal domain is sensitive to key legal elements. Even subtle differences in
key legal elements can significantly affect the judgement of relevance.
However, existing pre-trained language models designed for general purposes
have not been equipped to handle legal elements.
To address these issues, in this paper, we propose SAILER, a new
Structure-Aware pre-traIned language model for LEgal case Retrieval. It is
highlighted in the following three aspects: (1) SAILER fully utilizes the
structural information contained in legal case documents and pays more
attention to key legal elements, similar to how legal experts browse legal case
documents. (2) SAILER employs an asymmetric encoder-decoder architecture to
integrate several different pre-training objectives. In this way, rich semantic
information across tasks is encoded into dense vectors. (3) SAILER has powerful
discriminative ability, even without any legal annotation data. It can
distinguish legal cases with different charges accurately. Extensive
experiments over publicly available legal benchmarks demonstrate that our
approach can significantly outperform previous state-of-the-art methods in
legal case retrieval.
| [
{
"version": "v1",
"created": "Sat, 22 Apr 2023 10:47:01 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Li",
"Haitao",
""
],
[
"Ai",
"Qingyao",
""
],
[
"Chen",
"Jia",
""
],
[
"Dong",
"Qian",
""
],
[
"Wu",
"Yueyue",
""
],
[
"Liu",
"Yiqun",
""
],
[
"Chen",
"Chong",
""
],
[
"Tian",
"Qi",
""
]
] | TITLE: SAILER: Structure-aware Pre-trained Language Model for Legal Case
Retrieval
ABSTRACT: Legal case retrieval, which aims to find relevant cases for a query case,
plays a core role in the intelligent legal system. Despite the success that
pre-training has achieved in ad-hoc retrieval tasks, effective pre-training
strategies for legal case retrieval remain to be explored. Compared with
general documents, legal case documents are typically long text sequences with
intrinsic logical structures. However, most existing language models have
difficulty understanding the long-distance dependencies between different
structures. Moreover, in contrast to the general retrieval, the relevance in
the legal domain is sensitive to key legal elements. Even subtle differences in
key legal elements can significantly affect the judgement of relevance.
However, existing pre-trained language models designed for general purposes
have not been equipped to handle legal elements.
To address these issues, in this paper, we propose SAILER, a new
Structure-Aware pre-traIned language model for LEgal case Retrieval. It is
highlighted in the following three aspects: (1) SAILER fully utilizes the
structural information contained in legal case documents and pays more
attention to key legal elements, similar to how legal experts browse legal case
documents. (2) SAILER employs an asymmetric encoder-decoder architecture to
integrate several different pre-training objectives. In this way, rich semantic
information across tasks is encoded into dense vectors. (3) SAILER has powerful
discriminative ability, even without any legal annotation data. It can
distinguish legal cases with different charges accurately. Extensive
experiments over publicly available legal benchmarks demonstrate that our
approach can significantly outperform previous state-of-the-art methods in
legal case retrieval.
| no_new_dataset | 0.709799 |
2304.11943 | Haitao Li | Haitao Li, Qingyao Ai, Jingtao Zhan, Jiaxin Mao, Yiqun Liu, Zheng Liu,
Zhao Cao | Constructing Tree-based Index for Efficient and Effective Dense
Retrieval | 10 pages, accepted at SIGIR 2023 | null | null | null | cs.IR cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent studies have shown that Dense Retrieval (DR) techniques can
significantly improve the performance of first-stage retrieval in IR systems.
Despite its empirical effectiveness, the application of DR is still limited. In
contrast to statistic retrieval models that rely on highly efficient inverted
index solutions, DR models build dense embeddings that are difficult to be
pre-processed with most existing search indexing systems. To avoid the
expensive cost of brute-force search, the Approximate Nearest Neighbor (ANN)
algorithm and corresponding indexes are widely applied to speed up the
inference process of DR models. Unfortunately, while ANN can improve the
efficiency of DR models, it usually comes with a significant price on retrieval
performance.
To solve this issue, we propose JTR, which stands for Joint optimization of
TRee-based index and query encoding. Specifically, we design a new unified
contrastive learning loss to train tree-based index and query encoder in an
end-to-end manner. The tree-based negative sampling strategy is applied to make
the tree have the maximum heap property, which supports the effectiveness of
beam search well. Moreover, we treat the cluster assignment as an optimization
problem to update the tree-based index that allows overlapped clustering. We
evaluate JTR on numerous popular retrieval benchmarks. Experimental results
show that JTR achieves better retrieval performance while retaining high system
efficiency compared with widely-adopted baselines. It provides a potential
solution to balance efficiency and effectiveness in neural retrieval system
designs.
| [
{
"version": "v1",
"created": "Mon, 24 Apr 2023 09:25:39 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Li",
"Haitao",
""
],
[
"Ai",
"Qingyao",
""
],
[
"Zhan",
"Jingtao",
""
],
[
"Mao",
"Jiaxin",
""
],
[
"Liu",
"Yiqun",
""
],
[
"Liu",
"Zheng",
""
],
[
"Cao",
"Zhao",
""
]
] | TITLE: Constructing Tree-based Index for Efficient and Effective Dense
Retrieval
ABSTRACT: Recent studies have shown that Dense Retrieval (DR) techniques can
significantly improve the performance of first-stage retrieval in IR systems.
Despite its empirical effectiveness, the application of DR is still limited. In
contrast to statistic retrieval models that rely on highly efficient inverted
index solutions, DR models build dense embeddings that are difficult to be
pre-processed with most existing search indexing systems. To avoid the
expensive cost of brute-force search, the Approximate Nearest Neighbor (ANN)
algorithm and corresponding indexes are widely applied to speed up the
inference process of DR models. Unfortunately, while ANN can improve the
efficiency of DR models, it usually comes with a significant price on retrieval
performance.
To solve this issue, we propose JTR, which stands for Joint optimization of
TRee-based index and query encoding. Specifically, we design a new unified
contrastive learning loss to train tree-based index and query encoder in an
end-to-end manner. The tree-based negative sampling strategy is applied to make
the tree have the maximum heap property, which supports the effectiveness of
beam search well. Moreover, we treat the cluster assignment as an optimization
problem to update the tree-based index that allows overlapped clustering. We
evaluate JTR on numerous popular retrieval benchmarks. Experimental results
show that JTR achieves better retrieval performance while retaining high system
efficiency compared with widely-adopted baselines. It provides a potential
solution to balance efficiency and effectiveness in neural retrieval system
designs.
| no_new_dataset | 0.709422 |
2304.12454 | Manon Flageat | Manon Flageat and Luca Grillotti and Antoine Cully | Benchmark tasks for Quality-Diversity applied to Uncertain domains | null | null | null | null | cs.NE | http://creativecommons.org/licenses/by/4.0/ | While standard approaches to optimisation focus on producing a single
high-performing solution, Quality-Diversity (QD) algorithms allow large diverse
collections of such solutions to be found. If QD has proven promising across a
large variety of domains, it still struggles when faced with uncertain domains,
where quantification of performance and diversity are non-deterministic.
Previous work in Uncertain Quality-Diversity (UQD) has proposed methods and
metrics designed for such uncertain domains. In this paper, we propose a first
set of benchmark tasks to analyse and estimate the performance of UQD
algorithms. We identify the key uncertainty properties to easily define UQD
benchmark tasks: the uncertainty location, the type of distribution and its
parameters. By varying the nature of those key UQD components, we introduce a
set of 8 easy-to-implement and lightweight tasks, split into 3 main categories.
All our tasks build on the Redundant Arm: a common QD environment that is
lightweight and easily replicable. Each one of these tasks highlights one
specific limitation that arises when considering UQD domains. With this first
benchmark, we hope to facilitate later advances in UQD.
| [
{
"version": "v1",
"created": "Mon, 24 Apr 2023 21:23:26 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 17:46:21 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Flageat",
"Manon",
""
],
[
"Grillotti",
"Luca",
""
],
[
"Cully",
"Antoine",
""
]
] | TITLE: Benchmark tasks for Quality-Diversity applied to Uncertain domains
ABSTRACT: While standard approaches to optimisation focus on producing a single
high-performing solution, Quality-Diversity (QD) algorithms allow large diverse
collections of such solutions to be found. If QD has proven promising across a
large variety of domains, it still struggles when faced with uncertain domains,
where quantification of performance and diversity are non-deterministic.
Previous work in Uncertain Quality-Diversity (UQD) has proposed methods and
metrics designed for such uncertain domains. In this paper, we propose a first
set of benchmark tasks to analyse and estimate the performance of UQD
algorithms. We identify the key uncertainty properties to easily define UQD
benchmark tasks: the uncertainty location, the type of distribution and its
parameters. By varying the nature of those key UQD components, we introduce a
set of 8 easy-to-implement and lightweight tasks, split into 3 main categories.
All our tasks build on the Redundant Arm: a common QD environment that is
lightweight and easily replicable. Each one of these tasks highlights one
specific limitation that arises when considering UQD domains. With this first
benchmark, we hope to facilitate later advances in UQD.
| no_new_dataset | 0.698227 |
2304.12591 | Ganning Zhao | Ganning Zhao, Tingwei Shen, Suya You, and C.-C. Jay Kuo | Unsupervised Synthetic Image Refinement via Contrastive Learning and
Consistent Semantic-Structural Constraints | null | null | null | null | cs.CV cs.AI eess.IV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Ensuring the realism of computer-generated synthetic images is crucial to
deep neural network (DNN) training. Due to different semantic distributions
between synthetic and real-world captured datasets, there exists semantic
mismatch between synthetic and refined images, which in turn results in the
semantic distortion. Recently, contrastive learning (CL) has been successfully
used to pull correlated patches together and push uncorrelated ones apart. In
this work, we exploit semantic and structural consistency between synthetic and
refined images and adopt CL to reduce the semantic distortion. Besides, we
incorporate hard negative mining to improve the performance furthermore. We
compare the performance of our method with several other benchmarking methods
using qualitative and quantitative measures and show that our method offers the
state-of-the-art performance.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 05:55:28 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 06:03:25 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Zhao",
"Ganning",
""
],
[
"Shen",
"Tingwei",
""
],
[
"You",
"Suya",
""
],
[
"Kuo",
"C. -C. Jay",
""
]
] | TITLE: Unsupervised Synthetic Image Refinement via Contrastive Learning and
Consistent Semantic-Structural Constraints
ABSTRACT: Ensuring the realism of computer-generated synthetic images is crucial to
deep neural network (DNN) training. Due to different semantic distributions
between synthetic and real-world captured datasets, there exists semantic
mismatch between synthetic and refined images, which in turn results in the
semantic distortion. Recently, contrastive learning (CL) has been successfully
used to pull correlated patches together and push uncorrelated ones apart. In
this work, we exploit semantic and structural consistency between synthetic and
refined images and adopt CL to reduce the semantic distortion. Besides, we
incorporate hard negative mining to improve the performance furthermore. We
compare the performance of our method with several other benchmarking methods
using qualitative and quantitative measures and show that our method offers the
state-of-the-art performance.
| no_new_dataset | 0.711875 |
2304.12608 | Dengtian Lin | Lin Dengtian, Ma Yang, Li Yuhong, Song Xuemeng, Wu Jianlong, and Nie
Liqiang | OFAR: A Multimodal Evidence Retrieval Framework for Illegal
Live-streaming Identification | null | null | 10.1145/3539618.3591864 | null | cs.IR | http://creativecommons.org/licenses/by/4.0/ | Illegal live-streaming identification, which aims to help live-streaming
platforms immediately recognize the illegal behaviors in the live-streaming,
such as selling precious and endangered animals, plays a crucial role in
purifying the network environment. Traditionally, the live-streaming platform
needs to employ some professionals to manually identify the potential illegal
live-streaming. Specifically, the professional needs to search for related
evidence from a large-scale knowledge database for evaluating whether a given
live-streaming clip contains illegal behavior, which is time-consuming and
laborious. To address this issue, in this work, we propose a multimodal
evidence retrieval system, named OFAR, to facilitate the illegal live-streaming
identification. OFAR consists of three modules: Query Encoder, Document
Encoder, and MaxSim-based Contrastive Late Intersection. Both query encoder and
document encoder are implemented with the advanced OFA encoder, which is
pretrained on a large-scale multimodal dataset. In the last module, we
introduce contrastive learning on the basis of the MaxiSim-based late
intersection, to enhance the model's ability of query-document matching. The
proposed framework achieves significant improvement on our industrial dataset
TaoLive, demonstrating the advances of our scheme.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 06:48:35 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Apr 2023 09:23:09 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Dengtian",
"Lin",
""
],
[
"Yang",
"Ma",
""
],
[
"Yuhong",
"Li",
""
],
[
"Xuemeng",
"Song",
""
],
[
"Jianlong",
"Wu",
""
],
[
"Liqiang",
"Nie",
""
]
] | TITLE: OFAR: A Multimodal Evidence Retrieval Framework for Illegal
Live-streaming Identification
ABSTRACT: Illegal live-streaming identification, which aims to help live-streaming
platforms immediately recognize the illegal behaviors in the live-streaming,
such as selling precious and endangered animals, plays a crucial role in
purifying the network environment. Traditionally, the live-streaming platform
needs to employ some professionals to manually identify the potential illegal
live-streaming. Specifically, the professional needs to search for related
evidence from a large-scale knowledge database for evaluating whether a given
live-streaming clip contains illegal behavior, which is time-consuming and
laborious. To address this issue, in this work, we propose a multimodal
evidence retrieval system, named OFAR, to facilitate the illegal live-streaming
identification. OFAR consists of three modules: Query Encoder, Document
Encoder, and MaxSim-based Contrastive Late Intersection. Both query encoder and
document encoder are implemented with the advanced OFA encoder, which is
pretrained on a large-scale multimodal dataset. In the last module, we
introduce contrastive learning on the basis of the MaxiSim-based late
intersection, to enhance the model's ability of query-document matching. The
proposed framework achieves significant improvement on our industrial dataset
TaoLive, demonstrating the advances of our scheme.
| no_new_dataset | 0.711794 |
2304.13103 | Jafar Tahmoresnezhad | Mohammad Khodadadi, Jafar Tahmoresnezhad (1) ((1) Department of IT &
Computer Engineering, Urmia University of Technology, Or\=um\=iyeh, Iran) | HyMo: Vulnerability Detection in Smart Contracts using a Novel
Multi-Modal Hybrid Model | null | null | null | null | cs.CR cs.AI | http://creativecommons.org/licenses/by/4.0/ | With blockchain technology rapidly progress, the smart contracts have become
a common tool in a number of industries including finance, healthcare,
insurance and gaming. The number of smart contracts has multiplied, and at the
same time, the security of smart contracts has drawn considerable attention due
to the monetary losses brought on by smart contract vulnerabilities. Existing
analysis techniques are capable of identifying a large number of smart contract
security flaws, but they rely too much on rigid criteria established by
specialists, where the detection process takes much longer as the complexity of
the smart contract rises. In this paper, we propose HyMo as a multi-modal
hybrid deep learning model, which intelligently considers various input
representations to consider multimodality and FastText word embedding
technique, which represents each word as an n-gram of characters with BiGRU
deep learning technique, as a sequence processing model that consists of two
GRUs to achieve higher accuracy in smart contract vulnerability detection. The
model gathers features using various deep learning models to identify the smart
contract vulnerabilities. Through a series of studies on the currently publicly
accessible dataset such as ScrawlD, we show that our hybrid HyMo model has
excellent smart contract vulnerability detection performance. Therefore, HyMo
performs better detection of smart contract vulnerabilities against other
approaches.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 19:16:21 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Khodadadi",
"Mohammad",
""
],
[
"Tahmoresnezhad",
"Jafar",
""
]
] | TITLE: HyMo: Vulnerability Detection in Smart Contracts using a Novel
Multi-Modal Hybrid Model
ABSTRACT: With blockchain technology rapidly progress, the smart contracts have become
a common tool in a number of industries including finance, healthcare,
insurance and gaming. The number of smart contracts has multiplied, and at the
same time, the security of smart contracts has drawn considerable attention due
to the monetary losses brought on by smart contract vulnerabilities. Existing
analysis techniques are capable of identifying a large number of smart contract
security flaws, but they rely too much on rigid criteria established by
specialists, where the detection process takes much longer as the complexity of
the smart contract rises. In this paper, we propose HyMo as a multi-modal
hybrid deep learning model, which intelligently considers various input
representations to consider multimodality and FastText word embedding
technique, which represents each word as an n-gram of characters with BiGRU
deep learning technique, as a sequence processing model that consists of two
GRUs to achieve higher accuracy in smart contract vulnerability detection. The
model gathers features using various deep learning models to identify the smart
contract vulnerabilities. Through a series of studies on the currently publicly
accessible dataset such as ScrawlD, we show that our hybrid HyMo model has
excellent smart contract vulnerability detection performance. Therefore, HyMo
performs better detection of smart contract vulnerabilities against other
approaches.
| no_new_dataset | 0.708572 |
2304.13109 | Li-Hsiang Shen | Po-Chun Hsu, Li-Hsiang Shen, Chun-Hung Liu, and Kai-Ten Feng | Federated Deep Reinforcement Learning for THz-Beam Search with Limited
CSI | null | IEEE Vehicular Technology Conference (VTC-Fall), 2022 | 10.1109/VTC2022-Fall57202.2022.10012887 | null | cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Terahertz (THz) communication with ultra-wide available spectrum is a
promising technique that can achieve the stringent requirement of high data
rate in the next-generation wireless networks, yet its severe propagation
attenuation significantly hinders its implementation in practice. Finding beam
directions for a large-scale antenna array to effectively overcome severe
propagation attenuation of THz signals is a pressing need. This paper proposes
a novel approach of federated deep reinforcement learning (FDRL) to swiftly
perform THz-beam search for multiple base stations (BSs) coordinated by an edge
server in a cellular network. All the BSs conduct deep deterministic policy
gradient (DDPG)-based DRL to obtain THz beamforming policy with limited channel
state information (CSI). They update their DDPG models with hidden information
in order to mitigate inter-cell interference. We demonstrate that the cell
network can achieve higher throughput as more THz CSI and hidden neurons of
DDPG are adopted. We also show that FDRL with partial model update is able to
nearly achieve the same performance of FDRL with full model update, which
indicates an effective means to reduce communication load between the edge
server and the BSs by partial model uploading. Moreover, the proposed FDRL
outperforms conventional non-learning-based and existing non-FDRL benchmark
optimization methods.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 19:28:15 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Hsu",
"Po-Chun",
""
],
[
"Shen",
"Li-Hsiang",
""
],
[
"Liu",
"Chun-Hung",
""
],
[
"Feng",
"Kai-Ten",
""
]
] | TITLE: Federated Deep Reinforcement Learning for THz-Beam Search with Limited
CSI
ABSTRACT: Terahertz (THz) communication with ultra-wide available spectrum is a
promising technique that can achieve the stringent requirement of high data
rate in the next-generation wireless networks, yet its severe propagation
attenuation significantly hinders its implementation in practice. Finding beam
directions for a large-scale antenna array to effectively overcome severe
propagation attenuation of THz signals is a pressing need. This paper proposes
a novel approach of federated deep reinforcement learning (FDRL) to swiftly
perform THz-beam search for multiple base stations (BSs) coordinated by an edge
server in a cellular network. All the BSs conduct deep deterministic policy
gradient (DDPG)-based DRL to obtain THz beamforming policy with limited channel
state information (CSI). They update their DDPG models with hidden information
in order to mitigate inter-cell interference. We demonstrate that the cell
network can achieve higher throughput as more THz CSI and hidden neurons of
DDPG are adopted. We also show that FDRL with partial model update is able to
nearly achieve the same performance of FDRL with full model update, which
indicates an effective means to reduce communication load between the edge
server and the BSs by partial model uploading. Moreover, the proposed FDRL
outperforms conventional non-learning-based and existing non-FDRL benchmark
optimization methods.
| no_new_dataset | 0.709372 |
2304.13130 | Giacomo Nebbia | Giacomo Nebbia, Adriana Kovashka | Hypernymization of named entity-rich captions for grounding-based
multi-modal pretraining | null | null | 10.1145/3591106.3592223 | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Named entities are ubiquitous in text that naturally accompanies images,
especially in domains such as news or Wikipedia articles. In previous work,
named entities have been identified as a likely reason for low performance of
image-text retrieval models pretrained on Wikipedia and evaluated on named
entities-free benchmark datasets. Because they are rarely mentioned, named
entities could be challenging to model. They also represent missed learning
opportunities for self-supervised models: the link between named entity and
object in the image may be missed by the model, but it would not be if the
object were mentioned using a more common term. In this work, we investigate
hypernymization as a way to deal with named entities for pretraining
grounding-based multi-modal models and for fine-tuning on open-vocabulary
detection. We propose two ways to perform hypernymization: (1) a ``manual''
pipeline relying on a comprehensive ontology of concepts, and (2) a ``learned''
approach where we train a language model to learn to perform hypernymization.
We run experiments on data from Wikipedia and from The New York Times. We
report improved pretraining performance on objects of interest following
hypernymization, and we show the promise of hypernymization on open-vocabulary
detection, specifically on classes not seen during training.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 20:17:40 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Nebbia",
"Giacomo",
""
],
[
"Kovashka",
"Adriana",
""
]
] | TITLE: Hypernymization of named entity-rich captions for grounding-based
multi-modal pretraining
ABSTRACT: Named entities are ubiquitous in text that naturally accompanies images,
especially in domains such as news or Wikipedia articles. In previous work,
named entities have been identified as a likely reason for low performance of
image-text retrieval models pretrained on Wikipedia and evaluated on named
entities-free benchmark datasets. Because they are rarely mentioned, named
entities could be challenging to model. They also represent missed learning
opportunities for self-supervised models: the link between named entity and
object in the image may be missed by the model, but it would not be if the
object were mentioned using a more common term. In this work, we investigate
hypernymization as a way to deal with named entities for pretraining
grounding-based multi-modal models and for fine-tuning on open-vocabulary
detection. We propose two ways to perform hypernymization: (1) a ``manual''
pipeline relying on a comprehensive ontology of concepts, and (2) a ``learned''
approach where we train a language model to learn to perform hypernymization.
We run experiments on data from Wikipedia and from The New York Times. We
report improved pretraining performance on objects of interest following
hypernymization, and we show the promise of hypernymization on open-vocabulary
detection, specifically on classes not seen during training.
| no_new_dataset | 0.712282 |
2304.13134 | Ke Wu | Ke Wu, Ehsan Variani, Tom Bagby, Michael Riley | LAST: Scalable Lattice-Based Speech Modelling in JAX | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | We introduce LAST, a LAttice-based Speech Transducer library in JAX. With an
emphasis on flexibility, ease-of-use, and scalability, LAST implements
differentiable weighted finite state automaton (WFSA) algorithms needed for
training \& inference that scale to a large WFSA such as a recognition lattice
over the entire utterance. Despite these WFSA algorithms being well-known in
the literature, new challenges arise from performance characteristics of modern
architectures, and from nuances in automatic differentiation. We describe a
suite of generally applicable techniques employed in LAST to address these
challenges, and demonstrate their effectiveness with benchmarks on TPUv3 and
V100 GPU.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 20:25:37 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Wu",
"Ke",
""
],
[
"Variani",
"Ehsan",
""
],
[
"Bagby",
"Tom",
""
],
[
"Riley",
"Michael",
""
]
] | TITLE: LAST: Scalable Lattice-Based Speech Modelling in JAX
ABSTRACT: We introduce LAST, a LAttice-based Speech Transducer library in JAX. With an
emphasis on flexibility, ease-of-use, and scalability, LAST implements
differentiable weighted finite state automaton (WFSA) algorithms needed for
training \& inference that scale to a large WFSA such as a recognition lattice
over the entire utterance. Despite these WFSA algorithms being well-known in
the literature, new challenges arise from performance characteristics of modern
architectures, and from nuances in automatic differentiation. We describe a
suite of generally applicable techniques employed in LAST to address these
challenges, and demonstrate their effectiveness with benchmarks on TPUv3 and
V100 GPU.
| no_new_dataset | 0.707777 |
2304.13135 | Yudong Zhang | Lin Yang, Shuihua Wang, Yudong Zhang | MEDNC: Multi-ensemble deep neural network for COVID-19 diagnosis | null | null | null | null | eess.IV cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | Coronavirus disease 2019 (COVID-19) has spread all over the world for three
years, but medical facilities in many areas still aren't adequate. There is a
need for rapid COVID-19 diagnosis to identify high-risk patients and maximize
the use of limited medical resources. Motivated by this fact, we proposed the
deep learning framework MEDNC for automatic prediction and diagnosis of
COVID-19 using computed tomography (CT) images. Our model was trained using two
publicly available sets of COVID-19 data. And it was built with the inspiration
of transfer learning. Results indicated that the MEDNC greatly enhanced the
detection of COVID-19 infections, reaching an accuracy of 98.79% and 99.82%
respectively. We tested MEDNC on a brain tumor and a blood cell dataset to show
that our model applies to a wide range of problems. The outcomes demonstrated
that our proposed models attained an accuracy of 99.39% and 99.28%,
respectively. This COVID-19 recognition tool could help optimize healthcare
resources and reduce clinicians' workload when screening for the virus.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 20:26:05 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Yang",
"Lin",
""
],
[
"Wang",
"Shuihua",
""
],
[
"Zhang",
"Yudong",
""
]
] | TITLE: MEDNC: Multi-ensemble deep neural network for COVID-19 diagnosis
ABSTRACT: Coronavirus disease 2019 (COVID-19) has spread all over the world for three
years, but medical facilities in many areas still aren't adequate. There is a
need for rapid COVID-19 diagnosis to identify high-risk patients and maximize
the use of limited medical resources. Motivated by this fact, we proposed the
deep learning framework MEDNC for automatic prediction and diagnosis of
COVID-19 using computed tomography (CT) images. Our model was trained using two
publicly available sets of COVID-19 data. And it was built with the inspiration
of transfer learning. Results indicated that the MEDNC greatly enhanced the
detection of COVID-19 infections, reaching an accuracy of 98.79% and 99.82%
respectively. We tested MEDNC on a brain tumor and a blood cell dataset to show
that our model applies to a wide range of problems. The outcomes demonstrated
that our proposed models attained an accuracy of 99.39% and 99.28%,
respectively. This COVID-19 recognition tool could help optimize healthcare
resources and reduce clinicians' workload when screening for the virus.
| no_new_dataset | 0.710427 |
2304.13140 | Opprash Ruanlu | Ruan Lu, Zhou HangCheng, Ran Meng, Zhao Jin, Qin JiaoYu, Wei Feng,
Wang ChenZi | ESimCSE Unsupervised Contrastive Learning Jointly with UDA
Semi-Supervised Learning for Large Label System Text Classification Mode | This paper contains 14 pages,4 figures,4 tables | null | null | null | cs.LG cs.CL | http://creativecommons.org/licenses/by/4.0/ | The challenges faced by text classification with large tag systems in natural
language processing tasks include multiple tag systems, uneven data
distribution, and high noise. To address these problems, the ESimCSE
unsupervised comparative learning and UDA semi-supervised comparative learning
models are combined through the use of joint training techniques in the
models.The ESimCSE model efficiently learns text vector representations using
unlabeled data to achieve better classification results, while UDA is trained
using unlabeled data through semi-supervised learning methods to improve the
prediction performance of the models and stability, and further improve the
generalization ability of the model. In addition, adversarial training
techniques FGM and PGD are used in the model training process to improve the
robustness and reliability of the model. The experimental results show that
there is an 8% and 10% accuracy improvement relative to Baseline on the public
dataset Ruesters as well as on the operational dataset, respectively, and a 15%
improvement in manual validation accuracy can be achieved on the operational
dataset, indicating that the method is effective.
| [
{
"version": "v1",
"created": "Wed, 19 Apr 2023 03:44:23 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Lu",
"Ruan",
""
],
[
"HangCheng",
"Zhou",
""
],
[
"Meng",
"Ran",
""
],
[
"Jin",
"Zhao",
""
],
[
"JiaoYu",
"Qin",
""
],
[
"Feng",
"Wei",
""
],
[
"ChenZi",
"Wang",
""
]
] | TITLE: ESimCSE Unsupervised Contrastive Learning Jointly with UDA
Semi-Supervised Learning for Large Label System Text Classification Mode
ABSTRACT: The challenges faced by text classification with large tag systems in natural
language processing tasks include multiple tag systems, uneven data
distribution, and high noise. To address these problems, the ESimCSE
unsupervised comparative learning and UDA semi-supervised comparative learning
models are combined through the use of joint training techniques in the
models.The ESimCSE model efficiently learns text vector representations using
unlabeled data to achieve better classification results, while UDA is trained
using unlabeled data through semi-supervised learning methods to improve the
prediction performance of the models and stability, and further improve the
generalization ability of the model. In addition, adversarial training
techniques FGM and PGD are used in the model training process to improve the
robustness and reliability of the model. The experimental results show that
there is an 8% and 10% accuracy improvement relative to Baseline on the public
dataset Ruesters as well as on the operational dataset, respectively, and a 15%
improvement in manual validation accuracy can be achieved on the operational
dataset, indicating that the method is effective.
| no_new_dataset | 0.71022 |
2304.13148 | Martin Wessel | Martin Wessel, Tom\'a\v{s} Horych, Terry Ruas, Akiko Aizawa, Bela Gipp
and Timo Spinde | Introducing MBIB -- the first Media Bias Identification Benchmark Task
and Dataset Collection | To be published in Proceedings of the 46th International ACM SIGIR
Conference on Research and Development in Information Retrieval (SIGIR '23) | null | 10.1145/3539618.3591882 | null | cs.IR cs.AI | http://creativecommons.org/licenses/by-sa/4.0/ | Although media bias detection is a complex multi-task problem, there is, to
date, no unified benchmark grouping these evaluation tasks. We introduce the
Media Bias Identification Benchmark (MBIB), a comprehensive benchmark that
groups different types of media bias (e.g., linguistic, cognitive, political)
under a common framework to test how prospective detection techniques
generalize. After reviewing 115 datasets, we select nine tasks and carefully
propose 22 associated datasets for evaluating media bias detection techniques.
We evaluate MBIB using state-of-the-art Transformer techniques (e.g., T5,
BART). Our results suggest that while hate speech, racial bias, and gender bias
are easier to detect, models struggle to handle certain bias types, e.g.,
cognitive and political bias. However, our results show that no single
technique can outperform all the others significantly. We also find an uneven
distribution of research interest and resource allocation to the individual
tasks in media bias. A unified benchmark encourages the development of more
robust systems and shifts the current paradigm in media bias detection
evaluation towards solutions that tackle not one but multiple media bias types
simultaneously.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 20:49:55 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Wessel",
"Martin",
""
],
[
"Horych",
"Tomáš",
""
],
[
"Ruas",
"Terry",
""
],
[
"Aizawa",
"Akiko",
""
],
[
"Gipp",
"Bela",
""
],
[
"Spinde",
"Timo",
""
]
] | TITLE: Introducing MBIB -- the first Media Bias Identification Benchmark Task
and Dataset Collection
ABSTRACT: Although media bias detection is a complex multi-task problem, there is, to
date, no unified benchmark grouping these evaluation tasks. We introduce the
Media Bias Identification Benchmark (MBIB), a comprehensive benchmark that
groups different types of media bias (e.g., linguistic, cognitive, political)
under a common framework to test how prospective detection techniques
generalize. After reviewing 115 datasets, we select nine tasks and carefully
propose 22 associated datasets for evaluating media bias detection techniques.
We evaluate MBIB using state-of-the-art Transformer techniques (e.g., T5,
BART). Our results suggest that while hate speech, racial bias, and gender bias
are easier to detect, models struggle to handle certain bias types, e.g.,
cognitive and political bias. However, our results show that no single
technique can outperform all the others significantly. We also find an uneven
distribution of research interest and resource allocation to the individual
tasks in media bias. A unified benchmark encourages the development of more
robust systems and shifts the current paradigm in media bias detection
evaluation towards solutions that tackle not one but multiple media bias types
simultaneously.
| new_dataset | 0.700306 |
2304.13157 | Iain Mackie | Iain Mackie, Shubham Chatterjee, Jeffrey Dalton | Generative Relevance Feedback with Large Language Models | SIGIR 2023 Preprint, 6 pages | null | 10.1145/3539618.3591992 | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Current query expansion models use pseudo-relevance feedback to improve
first-pass retrieval effectiveness; however, this fails when the initial
results are not relevant. Instead of building a language model from retrieved
results, we propose Generative Relevance Feedback (GRF) that builds
probabilistic feedback models from long-form text generated from Large Language
Models. We study the effective methods for generating text by varying the
zero-shot generation subtasks: queries, entities, facts, news articles,
documents, and essays. We evaluate GRF on document retrieval benchmarks
covering a diverse set of queries and document collections, and the results
show that GRF methods significantly outperform previous PRF methods.
Specifically, we improve MAP between 5-19% and NDCG@10 17-24% compared to RM3
expansion, and achieve the best R@1k effectiveness on all datasets compared to
state-of-the-art sparse, dense, and expansion models.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 21:35:29 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Mackie",
"Iain",
""
],
[
"Chatterjee",
"Shubham",
""
],
[
"Dalton",
"Jeffrey",
""
]
] | TITLE: Generative Relevance Feedback with Large Language Models
ABSTRACT: Current query expansion models use pseudo-relevance feedback to improve
first-pass retrieval effectiveness; however, this fails when the initial
results are not relevant. Instead of building a language model from retrieved
results, we propose Generative Relevance Feedback (GRF) that builds
probabilistic feedback models from long-form text generated from Large Language
Models. We study the effective methods for generating text by varying the
zero-shot generation subtasks: queries, entities, facts, news articles,
documents, and essays. We evaluate GRF on document retrieval benchmarks
covering a diverse set of queries and document collections, and the results
show that GRF methods significantly outperform previous PRF methods.
Specifically, we improve MAP between 5-19% and NDCG@10 17-24% compared to RM3
expansion, and achieve the best R@1k effectiveness on all datasets compared to
state-of-the-art sparse, dense, and expansion models.
| no_new_dataset | 0.711469 |
2304.13164 | Massimo Caccia | Massimo Caccia, Alexandre Galashov, Arthur Douillard, Amal
Rannen-Triki, Dushyant Rao, Michela Paganini, Laurent Charlin, Marc'Aurelio
Ranzato, Razvan Pascanu | Towards Compute-Optimal Transfer Learning | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | The field of transfer learning is undergoing a significant shift with the
introduction of large pretrained models which have demonstrated strong
adaptability to a variety of downstream tasks. However, the high computational
and memory requirements to finetune or use these models can be a hindrance to
their widespread use. In this study, we present a solution to this issue by
proposing a simple yet effective way to trade computational efficiency for
asymptotic performance which we define as the performance a learning algorithm
achieves as compute tends to infinity. Specifically, we argue that zero-shot
structured pruning of pretrained models allows them to increase compute
efficiency with minimal reduction in performance. We evaluate our method on the
Nevis'22 continual learning benchmark that offers a diverse set of transfer
scenarios. Our results show that pruning convolutional filters of pretrained
models can lead to more than 20% performance improvement in low computational
regimes.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 21:49:09 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Caccia",
"Massimo",
""
],
[
"Galashov",
"Alexandre",
""
],
[
"Douillard",
"Arthur",
""
],
[
"Rannen-Triki",
"Amal",
""
],
[
"Rao",
"Dushyant",
""
],
[
"Paganini",
"Michela",
""
],
[
"Charlin",
"Laurent",
""
],
[
"Ranzato",
"Marc'Aurelio",
""
],
[
"Pascanu",
"Razvan",
""
]
] | TITLE: Towards Compute-Optimal Transfer Learning
ABSTRACT: The field of transfer learning is undergoing a significant shift with the
introduction of large pretrained models which have demonstrated strong
adaptability to a variety of downstream tasks. However, the high computational
and memory requirements to finetune or use these models can be a hindrance to
their widespread use. In this study, we present a solution to this issue by
proposing a simple yet effective way to trade computational efficiency for
asymptotic performance which we define as the performance a learning algorithm
achieves as compute tends to infinity. Specifically, we argue that zero-shot
structured pruning of pretrained models allows them to increase compute
efficiency with minimal reduction in performance. We evaluate our method on the
Nevis'22 continual learning benchmark that offers a diverse set of transfer
scenarios. Our results show that pruning convolutional filters of pretrained
models can lead to more than 20% performance improvement in low computational
regimes.
| no_new_dataset | 0.709177 |
2304.13166 | Sheng Liu | Sheng Liu, Cong Phuoc Huynh, Cong Chen, Maxim Arap, Raffay Hamid | LEMaRT: Label-Efficient Masked Region Transform for Image Harmonization | Accepted by CVPR'23, 19 pages | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a simple yet effective self-supervised pre-training method for
image harmonization which can leverage large-scale unannotated image datasets.
To achieve this goal, we first generate pre-training data online with our
Label-Efficient Masked Region Transform (LEMaRT) pipeline. Given an image,
LEMaRT generates a foreground mask and then applies a set of transformations to
perturb various visual attributes, e.g., defocus blur, contrast, saturation, of
the region specified by the generated mask. We then pre-train image
harmonization models by recovering the original image from the perturbed image.
Secondly, we introduce an image harmonization model, namely SwinIH, by
retrofitting the Swin Transformer [27] with a combination of local and global
self-attention mechanisms. Pre-training SwinIH with LEMaRT results in a new
state of the art for image harmonization, while being label-efficient, i.e.,
consuming less annotated data for fine-tuning than existing methods. Notably,
on iHarmony4 dataset [8], SwinIH outperforms the state of the art, i.e., SCS-Co
[16] by a margin of 0.4 dB when it is fine-tuned on only 50% of the training
data, and by 1.0 dB when it is trained on the full training dataset.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 21:51:22 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Liu",
"Sheng",
""
],
[
"Huynh",
"Cong Phuoc",
""
],
[
"Chen",
"Cong",
""
],
[
"Arap",
"Maxim",
""
],
[
"Hamid",
"Raffay",
""
]
] | TITLE: LEMaRT: Label-Efficient Masked Region Transform for Image Harmonization
ABSTRACT: We present a simple yet effective self-supervised pre-training method for
image harmonization which can leverage large-scale unannotated image datasets.
To achieve this goal, we first generate pre-training data online with our
Label-Efficient Masked Region Transform (LEMaRT) pipeline. Given an image,
LEMaRT generates a foreground mask and then applies a set of transformations to
perturb various visual attributes, e.g., defocus blur, contrast, saturation, of
the region specified by the generated mask. We then pre-train image
harmonization models by recovering the original image from the perturbed image.
Secondly, we introduce an image harmonization model, namely SwinIH, by
retrofitting the Swin Transformer [27] with a combination of local and global
self-attention mechanisms. Pre-training SwinIH with LEMaRT results in a new
state of the art for image harmonization, while being label-efficient, i.e.,
consuming less annotated data for fine-tuning than existing methods. Notably,
on iHarmony4 dataset [8], SwinIH outperforms the state of the art, i.e., SCS-Co
[16] by a margin of 0.4 dB when it is fine-tuned on only 50% of the training
data, and by 1.0 dB when it is trained on the full training dataset.
| no_new_dataset | 0.709818 |
2304.13172 | Yiwei Hu | Yiwei Hu, Paul Guerrero, Milo\v{s} Ha\v{s}an, Holly Rushmeier and
Valentin Deschaintre | Generating Procedural Materials from Text or Image Prompts | null | SIGGRAPH 2023 | 10.1145/3588432.3591520 | null | cs.GR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Node graph systems are used ubiquitously for material design in computer
graphics. They allow the use of visual programming to achieve desired effects
without writing code. As high-level design tools they provide convenience and
flexibility, but mastering the creation of node graphs usually requires
professional training. We propose an algorithm capable of generating multiple
node graphs from different types of prompts, significantly lowering the bar for
users to explore a specific design space. Previous work was limited to
unconditional generation of random node graphs, making the generation of an
envisioned material challenging. We propose a multi-modal node graph generation
neural architecture for high-quality procedural material synthesis which can be
conditioned on different inputs (text or image prompts), using a CLIP-based
encoder. We also create a substantially augmented material graph dataset, key
to improving the generation quality. Finally, we generate high-quality graph
samples using a regularized sampling process and improve the matching quality
by differentiable optimization for top-ranked samples. We compare our methods
to CLIP-based database search baselines (which are themselves novel) and
achieve superior or similar performance without requiring massive data storage.
We further show that our model can produce a set of material graphs
unconditionally, conditioned on images, text prompts or partial graphs, serving
as a tool for automatic visual programming completion.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 22:13:11 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Hu",
"Yiwei",
""
],
[
"Guerrero",
"Paul",
""
],
[
"Hašan",
"Miloš",
""
],
[
"Rushmeier",
"Holly",
""
],
[
"Deschaintre",
"Valentin",
""
]
] | TITLE: Generating Procedural Materials from Text or Image Prompts
ABSTRACT: Node graph systems are used ubiquitously for material design in computer
graphics. They allow the use of visual programming to achieve desired effects
without writing code. As high-level design tools they provide convenience and
flexibility, but mastering the creation of node graphs usually requires
professional training. We propose an algorithm capable of generating multiple
node graphs from different types of prompts, significantly lowering the bar for
users to explore a specific design space. Previous work was limited to
unconditional generation of random node graphs, making the generation of an
envisioned material challenging. We propose a multi-modal node graph generation
neural architecture for high-quality procedural material synthesis which can be
conditioned on different inputs (text or image prompts), using a CLIP-based
encoder. We also create a substantially augmented material graph dataset, key
to improving the generation quality. Finally, we generate high-quality graph
samples using a regularized sampling process and improve the matching quality
by differentiable optimization for top-ranked samples. We compare our methods
to CLIP-based database search baselines (which are themselves novel) and
achieve superior or similar performance without requiring massive data storage.
We further show that our model can produce a set of material graphs
unconditionally, conditioned on images, text prompts or partial graphs, serving
as a tool for automatic visual programming completion.
| no_new_dataset | 0.58818 |
2304.13174 | Jiechao Gao | Xiao-Yang Liu, Ziyi Xia, Hongyang Yang, Jiechao Gao, Daochen Zha, Ming
Zhu, Christina Dan Wang, Zhaoran Wang, Jian Guo | Dynamic Datasets and Market Environments for Financial Reinforcement
Learning | 49 pages, 15 figures. arXiv admin note: substantial text overlap with
arXiv:2211.03107 | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The financial market is a particularly challenging playground for deep
reinforcement learning due to its unique feature of dynamic datasets. Building
high-quality market environments for training financial reinforcement learning
(FinRL) agents is difficult due to major factors such as the low
signal-to-noise ratio of financial data, survivorship bias of historical data,
and model overfitting. In this paper, we present FinRL-Meta, a data-centric and
openly accessible library that processes dynamic datasets from real-world
markets into gym-style market environments and has been actively maintained by
the AI4Finance community. First, following a DataOps paradigm, we provide
hundreds of market environments through an automatic data curation pipeline.
Second, we provide homegrown examples and reproduce popular research papers as
stepping stones for users to design new trading strategies. We also deploy the
library on cloud platforms so that users can visualize their own results and
assess the relative performance via community-wise competitions. Third, we
provide dozens of Jupyter/Python demos organized into a curriculum and a
documentation website to serve the rapidly growing community. The open-source
codes for the data curation pipeline are available at
https://github.com/AI4Finance-Foundation/FinRL-Meta
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 22:17:31 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Liu",
"Xiao-Yang",
""
],
[
"Xia",
"Ziyi",
""
],
[
"Yang",
"Hongyang",
""
],
[
"Gao",
"Jiechao",
""
],
[
"Zha",
"Daochen",
""
],
[
"Zhu",
"Ming",
""
],
[
"Wang",
"Christina Dan",
""
],
[
"Wang",
"Zhaoran",
""
],
[
"Guo",
"Jian",
""
]
] | TITLE: Dynamic Datasets and Market Environments for Financial Reinforcement
Learning
ABSTRACT: The financial market is a particularly challenging playground for deep
reinforcement learning due to its unique feature of dynamic datasets. Building
high-quality market environments for training financial reinforcement learning
(FinRL) agents is difficult due to major factors such as the low
signal-to-noise ratio of financial data, survivorship bias of historical data,
and model overfitting. In this paper, we present FinRL-Meta, a data-centric and
openly accessible library that processes dynamic datasets from real-world
markets into gym-style market environments and has been actively maintained by
the AI4Finance community. First, following a DataOps paradigm, we provide
hundreds of market environments through an automatic data curation pipeline.
Second, we provide homegrown examples and reproduce popular research papers as
stepping stones for users to design new trading strategies. We also deploy the
library on cloud platforms so that users can visualize their own results and
assess the relative performance via community-wise competitions. Third, we
provide dozens of Jupyter/Python demos organized into a curriculum and a
documentation website to serve the rapidly growing community. The open-source
codes for the data curation pipeline are available at
https://github.com/AI4Finance-Foundation/FinRL-Meta
| no_new_dataset | 0.710653 |
2304.13188 | Dylan Slack | Dylan Slack and Sameer Singh | TABLET: Learning From Instructions For Tabular Data | Please find the TABLET demo and code at
https://dylanslacks.website/Tablet | null | null | null | cs.LG cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Acquiring high-quality data is often a significant challenge in training
machine learning (ML) models for tabular prediction, particularly in
privacy-sensitive and costly domains like medicine and finance. Providing
natural language instructions to large language models (LLMs) offers an
alternative solution. However, it is unclear how effectively instructions
leverage the knowledge in LLMs for solving tabular prediction problems. To
address this gap, we introduce TABLET, a benchmark of 20 diverse tabular
datasets annotated with instructions that vary in their phrasing, granularity,
and technicality. Additionally, TABLET includes the instructions' logic and
structured modifications to the instructions. We find in-context instructions
increase zero-shot F1 performance for Flan-T5 11b by 44% on average and 13% for
ChatGPT on TABLET. Also, we explore the limitations of using LLMs for tabular
prediction in our benchmark by evaluating instruction faithfulness. We find
LLMs often ignore instructions and fail to predict specific instances
correctly, even with examples. Our analysis on TABLET shows that, while
instructions help LLM performance, learning from instructions for tabular data
requires new capabilities.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 23:07:20 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Slack",
"Dylan",
""
],
[
"Singh",
"Sameer",
""
]
] | TITLE: TABLET: Learning From Instructions For Tabular Data
ABSTRACT: Acquiring high-quality data is often a significant challenge in training
machine learning (ML) models for tabular prediction, particularly in
privacy-sensitive and costly domains like medicine and finance. Providing
natural language instructions to large language models (LLMs) offers an
alternative solution. However, it is unclear how effectively instructions
leverage the knowledge in LLMs for solving tabular prediction problems. To
address this gap, we introduce TABLET, a benchmark of 20 diverse tabular
datasets annotated with instructions that vary in their phrasing, granularity,
and technicality. Additionally, TABLET includes the instructions' logic and
structured modifications to the instructions. We find in-context instructions
increase zero-shot F1 performance for Flan-T5 11b by 44% on average and 13% for
ChatGPT on TABLET. Also, we explore the limitations of using LLMs for tabular
prediction in our benchmark by evaluating instruction faithfulness. We find
LLMs often ignore instructions and fail to predict specific instances
correctly, even with examples. Our analysis on TABLET shows that, while
instructions help LLM performance, learning from instructions for tabular data
requires new capabilities.
| new_dataset | 0.716392 |
2304.13191 | Surjodeep Sarkar | Surjodeep Sarkar, Manas Gaur, L. Chen, Muskan Garg, Biplav Srivastava,
Bhaktee Dongaonkar | Towards Explainable and Safe Conversational Agents for Mental Health: A
Survey | 10 pages, 3 figures, 2 tables | null | null | null | cs.AI cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | Virtual Mental Health Assistants (VMHAs) are seeing continual advancements to
support the overburdened global healthcare system that gets 60 million primary
care visits, and 6 million Emergency Room (ER) visits annually. These systems
are built by clinical psychologists, psychiatrists, and Artificial Intelligence
(AI) researchers for Cognitive Behavioral Therapy (CBT). At present, the role
of VMHAs is to provide emotional support through information, focusing less on
developing a reflective conversation with the patient. A more comprehensive,
safe and explainable approach is required to build responsible VMHAs to ask
follow-up questions or provide a well-informed response. This survey offers a
systematic critical review of the existing conversational agents in mental
health, followed by new insights into the improvements of VMHAs with contextual
knowledge, datasets, and their emerging role in clinical decision support. We
also provide new directions toward enriching the user experience of VMHAs with
explainability, safety, and wholesome trustworthiness. Finally, we provide
evaluation metrics and practical considerations for VMHAs beyond the current
literature to build trust between VMHAs and patients in active communications.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 23:12:13 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Sarkar",
"Surjodeep",
""
],
[
"Gaur",
"Manas",
""
],
[
"Chen",
"L.",
""
],
[
"Garg",
"Muskan",
""
],
[
"Srivastava",
"Biplav",
""
],
[
"Dongaonkar",
"Bhaktee",
""
]
] | TITLE: Towards Explainable and Safe Conversational Agents for Mental Health: A
Survey
ABSTRACT: Virtual Mental Health Assistants (VMHAs) are seeing continual advancements to
support the overburdened global healthcare system that gets 60 million primary
care visits, and 6 million Emergency Room (ER) visits annually. These systems
are built by clinical psychologists, psychiatrists, and Artificial Intelligence
(AI) researchers for Cognitive Behavioral Therapy (CBT). At present, the role
of VMHAs is to provide emotional support through information, focusing less on
developing a reflective conversation with the patient. A more comprehensive,
safe and explainable approach is required to build responsible VMHAs to ask
follow-up questions or provide a well-informed response. This survey offers a
systematic critical review of the existing conversational agents in mental
health, followed by new insights into the improvements of VMHAs with contextual
knowledge, datasets, and their emerging role in clinical decision support. We
also provide new directions toward enriching the user experience of VMHAs with
explainability, safety, and wholesome trustworthiness. Finally, we provide
evaluation metrics and practical considerations for VMHAs beyond the current
literature to build trust between VMHAs and patients in active communications.
| no_new_dataset | 0.706608 |
2304.13201 | Negar Nejatishahidin | Negar Nejatishahidin, Will Hutchcroft, Manjunath Narayana, Ivaylo
Boyadzhiev, Yuguang Li, Naji Khosravan, Jana Kosecka, Sing Bing Kang | Graph-CoVis: GNN-based Multi-view Panorama Global Pose Estimation | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | In this paper, we address the problem of wide-baseline camera pose estimation
from a group of 360$^\circ$ panoramas under upright-camera assumption. Recent
work has demonstrated the merit of deep-learning for end-to-end direct relative
pose regression in 360$^\circ$ panorama pairs [11]. To exploit the benefits of
multi-view logic in a learning-based framework, we introduce Graph-CoVis, which
non-trivially extends CoVisPose [11] from relative two-view to global
multi-view spherical camera pose estimation. Graph-CoVis is a novel Graph
Neural Network based architecture that jointly learns the co-visible structure
and global motion in an end-to-end and fully-supervised approach. Using the
ZInD [4] dataset, which features real homes presenting wide-baselines,
occlusion, and limited visual overlap, we show that our model performs
competitively to state-of-the-art approaches.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 00:04:50 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Nejatishahidin",
"Negar",
""
],
[
"Hutchcroft",
"Will",
""
],
[
"Narayana",
"Manjunath",
""
],
[
"Boyadzhiev",
"Ivaylo",
""
],
[
"Li",
"Yuguang",
""
],
[
"Khosravan",
"Naji",
""
],
[
"Kosecka",
"Jana",
""
],
[
"Kang",
"Sing Bing",
""
]
] | TITLE: Graph-CoVis: GNN-based Multi-view Panorama Global Pose Estimation
ABSTRACT: In this paper, we address the problem of wide-baseline camera pose estimation
from a group of 360$^\circ$ panoramas under upright-camera assumption. Recent
work has demonstrated the merit of deep-learning for end-to-end direct relative
pose regression in 360$^\circ$ panorama pairs [11]. To exploit the benefits of
multi-view logic in a learning-based framework, we introduce Graph-CoVis, which
non-trivially extends CoVisPose [11] from relative two-view to global
multi-view spherical camera pose estimation. Graph-CoVis is a novel Graph
Neural Network based architecture that jointly learns the co-visible structure
and global motion in an end-to-end and fully-supervised approach. Using the
ZInD [4] dataset, which features real homes presenting wide-baselines,
occlusion, and limited visual overlap, we show that our model performs
competitively to state-of-the-art approaches.
| no_new_dataset | 0.707821 |
2304.13214 | Isaac Corley | Isaac Corley and Peyman Najafirad | Single-View Height Estimation with Conditional Diffusion Probabilistic
Models | null | null | null | null | cs.CV cs.LG eess.IV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Digital Surface Models (DSM) offer a wealth of height information for
understanding the Earth's surface as well as monitoring the existence or change
in natural and man-made structures. Classical height estimation requires
multi-view geospatial imagery or LiDAR point clouds which can be expensive to
acquire. Single-view height estimation using neural network based models shows
promise however it can struggle with reconstructing high resolution features.
The latest advancements in diffusion models for high resolution image synthesis
and editing have yet to be utilized for remote sensing imagery, particularly
height estimation. Our approach involves training a generative diffusion model
to learn the joint distribution of optical and DSM images across both domains
as a Markov chain. This is accomplished by minimizing a denoising score
matching objective while being conditioned on the source image to generate
realistic high resolution 3D surfaces. In this paper we experiment with
conditional denoising diffusion probabilistic models (DDPM) for height
estimation from a single remotely sensed image and show promising results on
the Vaihingen benchmark dataset.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 00:37:05 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Corley",
"Isaac",
""
],
[
"Najafirad",
"Peyman",
""
]
] | TITLE: Single-View Height Estimation with Conditional Diffusion Probabilistic
Models
ABSTRACT: Digital Surface Models (DSM) offer a wealth of height information for
understanding the Earth's surface as well as monitoring the existence or change
in natural and man-made structures. Classical height estimation requires
multi-view geospatial imagery or LiDAR point clouds which can be expensive to
acquire. Single-view height estimation using neural network based models shows
promise however it can struggle with reconstructing high resolution features.
The latest advancements in diffusion models for high resolution image synthesis
and editing have yet to be utilized for remote sensing imagery, particularly
height estimation. Our approach involves training a generative diffusion model
to learn the joint distribution of optical and DSM images across both domains
as a Markov chain. This is accomplished by minimizing a denoising score
matching objective while being conditioned on the source image to generate
realistic high resolution 3D surfaces. In this paper we experiment with
conditional denoising diffusion probabilistic models (DDPM) for height
estimation from a single remotely sensed image and show promising results on
the Vaihingen benchmark dataset.
| no_new_dataset | 0.710791 |
2304.13232 | Amin Sarihi | Amin Sarihi, Peter Jamieson, Ahmad Patooghy, Abdel-Hameed A. Badawy | Multi-criteria Hardware Trojan Detection: A Reinforcement Learning
Approach | null | null | null | null | cs.AR cs.CR cs.LG | http://creativecommons.org/licenses/by/4.0/ | Hardware Trojans (HTs) are undesired design or manufacturing modifications
that can severely alter the security and functionality of digital integrated
circuits. HTs can be inserted according to various design criteria, e.g., nets
switching activity, observability, controllability, etc. However, to our
knowledge, most HT detection methods are only based on a single criterion,
i.e., nets switching activity. This paper proposes a multi-criteria
reinforcement learning (RL) HT detection tool that features a tunable reward
function for different HT detection scenarios. The tool allows for exploring
existing detection strategies and can adapt new detection scenarios with
minimal effort. We also propose a generic methodology for comparing HT
detection methods fairly. Our preliminary results show an average of 84.2%
successful HT detection in ISCAS-85 benchmark
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 01:40:55 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Sarihi",
"Amin",
""
],
[
"Jamieson",
"Peter",
""
],
[
"Patooghy",
"Ahmad",
""
],
[
"Badawy",
"Abdel-Hameed A.",
""
]
] | TITLE: Multi-criteria Hardware Trojan Detection: A Reinforcement Learning
Approach
ABSTRACT: Hardware Trojans (HTs) are undesired design or manufacturing modifications
that can severely alter the security and functionality of digital integrated
circuits. HTs can be inserted according to various design criteria, e.g., nets
switching activity, observability, controllability, etc. However, to our
knowledge, most HT detection methods are only based on a single criterion,
i.e., nets switching activity. This paper proposes a multi-criteria
reinforcement learning (RL) HT detection tool that features a tunable reward
function for different HT detection scenarios. The tool allows for exploring
existing detection strategies and can adapt new detection scenarios with
minimal effort. We also propose a generic methodology for comparing HT
detection methods fairly. Our preliminary results show an average of 84.2%
successful HT detection in ISCAS-85 benchmark
| no_new_dataset | 0.709831 |
2304.13239 | Nathaniel Strawn | Mitchell Rimerman and Nate Strawn | Numerical Approximation of Andrews Plots with Optimal Spatial-Spectral
Smoothing | 25 pages, 12 figures | null | null | null | math.NA cs.HC cs.NA stat.ML | http://creativecommons.org/licenses/by/4.0/ | Andrews plots provide aesthetically pleasant visualizations of
high-dimensional datasets. This work proves that Andrews plots (when defined in
terms of the principal component scores of a dataset) are optimally ``smooth''
on average, and solve an infinite-dimensional quadratic minimization program
over the set of linear isometries from the Euclidean data space to
$L^2([0,1])$. By building technical machinery that characterizes the solutions
to general infinite-dimensional quadratic minimization programs over linear
isometries, we further show that the solution set is (in the generic case) a
manifold. To avoid the ambiguities presented by this manifold of solutions, we
add ``spectral smoothing'' terms to the infinite-dimensional optimization
program to induce Andrews plots with optimal spatial-spectral smoothing. We
characterize the (generic) set of solutions to this program and prove that the
resulting plots admit efficient numerical approximations. These
spatial-spectral smooth Andrews plots tend to avoid some ``visual clutter''
that arises due to the oscillation of trigonometric polynomials.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 02:01:02 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Rimerman",
"Mitchell",
""
],
[
"Strawn",
"Nate",
""
]
] | TITLE: Numerical Approximation of Andrews Plots with Optimal Spatial-Spectral
Smoothing
ABSTRACT: Andrews plots provide aesthetically pleasant visualizations of
high-dimensional datasets. This work proves that Andrews plots (when defined in
terms of the principal component scores of a dataset) are optimally ``smooth''
on average, and solve an infinite-dimensional quadratic minimization program
over the set of linear isometries from the Euclidean data space to
$L^2([0,1])$. By building technical machinery that characterizes the solutions
to general infinite-dimensional quadratic minimization programs over linear
isometries, we further show that the solution set is (in the generic case) a
manifold. To avoid the ambiguities presented by this manifold of solutions, we
add ``spectral smoothing'' terms to the infinite-dimensional optimization
program to induce Andrews plots with optimal spatial-spectral smoothing. We
characterize the (generic) set of solutions to this program and prove that the
resulting plots admit efficient numerical approximations. These
spatial-spectral smooth Andrews plots tend to avoid some ``visual clutter''
that arises due to the oscillation of trigonometric polynomials.
| no_new_dataset | 0.710427 |
2304.13249 | Kentaro Ohno | Kentaro Ohno, Misato Nakabayashi | A Security Verification Framework of Cryptographic Protocols Using
Machine Learning | 14 pages, 5 figures | null | null | null | cs.CR cs.LG | http://creativecommons.org/licenses/by/4.0/ | We propose a security verification framework for cryptographic protocols
using machine learning. In recent years, as cryptographic protocols have become
more complex, research on automatic verification techniques has been focused
on. The main technique is formal verification. However, the formal verification
has two problems: it requires a large amount of computational time and does not
guarantee decidability. We propose a method that allows security verification
with computational time on the order of linear with respect to the size of the
protocol using machine learning. In training machine learning models for
security verification of cryptographic protocols, a sufficient amount of data,
i.e., a set of protocol data with security labels, is difficult to collect from
academic papers and other sources. To overcome this issue, we propose a way to
create arbitrarily large datasets by automatically generating random protocols
and assigning security labels to them using formal verification tools.
Furthermore, to exploit structural features of protocols, we construct a neural
network that processes a protocol along its series and tree structures. We
evaluate the proposed method by applying it to verification of practical
cryptographic protocols.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 02:37:43 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Ohno",
"Kentaro",
""
],
[
"Nakabayashi",
"Misato",
""
]
] | TITLE: A Security Verification Framework of Cryptographic Protocols Using
Machine Learning
ABSTRACT: We propose a security verification framework for cryptographic protocols
using machine learning. In recent years, as cryptographic protocols have become
more complex, research on automatic verification techniques has been focused
on. The main technique is formal verification. However, the formal verification
has two problems: it requires a large amount of computational time and does not
guarantee decidability. We propose a method that allows security verification
with computational time on the order of linear with respect to the size of the
protocol using machine learning. In training machine learning models for
security verification of cryptographic protocols, a sufficient amount of data,
i.e., a set of protocol data with security labels, is difficult to collect from
academic papers and other sources. To overcome this issue, we propose a way to
create arbitrarily large datasets by automatically generating random protocols
and assigning security labels to them using formal verification tools.
Furthermore, to exploit structural features of protocols, we construct a neural
network that processes a protocol along its series and tree structures. We
evaluate the proposed method by applying it to verification of practical
cryptographic protocols.
| no_new_dataset | 0.706577 |
2304.13255 | Aziz Mohaisen | Mohammed Abuhamad and Changhun Jung and David Mohaisen and DaeHun
Nyang | SHIELD: Thwarting Code Authorship Attribution | 12 pages, 13 figures | null | null | null | cs.CR cs.CY cs.LG cs.SE | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Authorship attribution has become increasingly accurate, posing a serious
privacy risk for programmers who wish to remain anonymous. In this paper, we
introduce SHIELD to examine the robustness of different code authorship
attribution approaches against adversarial code examples. We define four
attacks on attribution techniques, which include targeted and non-targeted
attacks, and realize them using adversarial code perturbation. We experiment
with a dataset of 200 programmers from the Google Code Jam competition to
validate our methods targeting six state-of-the-art authorship attribution
methods that adopt a variety of techniques for extracting authorship traits
from source-code, including RNN, CNN, and code stylometry. Our experiments
demonstrate the vulnerability of current authorship attribution methods against
adversarial attacks. For the non-targeted attack, our experiments demonstrate
the vulnerability of current authorship attribution methods against the attack
with an attack success rate exceeds 98.5\% accompanied by a degradation of the
identification confidence that exceeds 13\%. For the targeted attacks, we show
the possibility of impersonating a programmer using targeted-adversarial
perturbations with a success rate ranging from 66\% to 88\% for different
authorship attribution techniques under several adversarial scenarios.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 02:55:28 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Abuhamad",
"Mohammed",
""
],
[
"Jung",
"Changhun",
""
],
[
"Mohaisen",
"David",
""
],
[
"Nyang",
"DaeHun",
""
]
] | TITLE: SHIELD: Thwarting Code Authorship Attribution
ABSTRACT: Authorship attribution has become increasingly accurate, posing a serious
privacy risk for programmers who wish to remain anonymous. In this paper, we
introduce SHIELD to examine the robustness of different code authorship
attribution approaches against adversarial code examples. We define four
attacks on attribution techniques, which include targeted and non-targeted
attacks, and realize them using adversarial code perturbation. We experiment
with a dataset of 200 programmers from the Google Code Jam competition to
validate our methods targeting six state-of-the-art authorship attribution
methods that adopt a variety of techniques for extracting authorship traits
from source-code, including RNN, CNN, and code stylometry. Our experiments
demonstrate the vulnerability of current authorship attribution methods against
adversarial attacks. For the non-targeted attack, our experiments demonstrate
the vulnerability of current authorship attribution methods against the attack
with an attack success rate exceeds 98.5\% accompanied by a degradation of the
identification confidence that exceeds 13\%. For the targeted attacks, we show
the possibility of impersonating a programmer using targeted-adversarial
perturbations with a success rate ranging from 66\% to 88\% for different
authorship attribution techniques under several adversarial scenarios.
| no_new_dataset | 0.709629 |
2304.13265 | Nikita Dvornik | Nikita Dvornik, Isma Hadji, Ran Zhang, Konstantinos G. Derpanis,
Animesh Garg, Richard P. Wildes, Allan D. Jepson | StepFormer: Self-supervised Step Discovery and Localization in
Instructional Videos | CVPR'23 | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Instructional videos are an important resource to learn procedural tasks from
human demonstrations. However, the instruction steps in such videos are
typically short and sparse, with most of the video being irrelevant to the
procedure. This motivates the need to temporally localize the instruction steps
in such videos, i.e. the task called key-step localization. Traditional methods
for key-step localization require video-level human annotations and thus do not
scale to large datasets. In this work, we tackle the problem with no human
supervision and introduce StepFormer, a self-supervised model that discovers
and localizes instruction steps in a video. StepFormer is a transformer decoder
that attends to the video with learnable queries, and produces a sequence of
slots capturing the key-steps in the video. We train our system on a large
dataset of instructional videos, using their automatically-generated subtitles
as the only source of supervision. In particular, we supervise our system with
a sequence of text narrations using an order-aware loss function that filters
out irrelevant phrases. We show that our model outperforms all previous
unsupervised and weakly-supervised approaches on step detection and
localization by a large margin on three challenging benchmarks. Moreover, our
model demonstrates an emergent property to solve zero-shot multi-step
localization and outperforms all relevant baselines at this task.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 03:37:28 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Dvornik",
"Nikita",
""
],
[
"Hadji",
"Isma",
""
],
[
"Zhang",
"Ran",
""
],
[
"Derpanis",
"Konstantinos G.",
""
],
[
"Garg",
"Animesh",
""
],
[
"Wildes",
"Richard P.",
""
],
[
"Jepson",
"Allan D.",
""
]
] | TITLE: StepFormer: Self-supervised Step Discovery and Localization in
Instructional Videos
ABSTRACT: Instructional videos are an important resource to learn procedural tasks from
human demonstrations. However, the instruction steps in such videos are
typically short and sparse, with most of the video being irrelevant to the
procedure. This motivates the need to temporally localize the instruction steps
in such videos, i.e. the task called key-step localization. Traditional methods
for key-step localization require video-level human annotations and thus do not
scale to large datasets. In this work, we tackle the problem with no human
supervision and introduce StepFormer, a self-supervised model that discovers
and localizes instruction steps in a video. StepFormer is a transformer decoder
that attends to the video with learnable queries, and produces a sequence of
slots capturing the key-steps in the video. We train our system on a large
dataset of instructional videos, using their automatically-generated subtitles
as the only source of supervision. In particular, we supervise our system with
a sequence of text narrations using an order-aware loss function that filters
out irrelevant phrases. We show that our model outperforms all previous
unsupervised and weakly-supervised approaches on step detection and
localization by a large margin on three challenging benchmarks. Moreover, our
model demonstrates an emergent property to solve zero-shot multi-step
localization and outperforms all relevant baselines at this task.
| no_new_dataset | 0.709466 |
2304.13275 | Hao Lu | Hao Lu, Adam Thelen, Olga Fink, Chao Hu, Simon Laflamme | Federated Learning with Uncertainty-Based Client Clustering for
Fleet-Wide Fault Diagnosis | null | null | null | null | cs.LG cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Operators from various industries have been pushing the adoption of wireless
sensing nodes for industrial monitoring, and such efforts have produced
sizeable condition monitoring datasets that can be used to build diagnosis
algorithms capable of warning maintenance engineers of impending failure or
identifying current system health conditions. However, single operators may not
have sufficiently large fleets of systems or component units to collect
sufficient data to develop data-driven algorithms. Collecting a satisfactory
quantity of fault patterns for safety-critical systems is particularly
difficult due to the rarity of faults. Federated learning (FL) has emerged as a
promising solution to leverage datasets from multiple operators to train a
decentralized asset fault diagnosis model while maintaining data
confidentiality. However, there are still considerable obstacles to overcome
when it comes to optimizing the federation strategy without leaking sensitive
data and addressing the issue of client dataset heterogeneity. This is
particularly prevalent in fault diagnosis applications due to the high
diversity of operating conditions and system configurations. To address these
two challenges, we propose a novel clustering-based FL algorithm where clients
are clustered for federating based on dataset similarity. To quantify dataset
similarity between clients without explicitly sharing data, each client sets
aside a local test dataset and evaluates the other clients' model prediction
accuracy and uncertainty on this test dataset. Clients are then clustered for
FL based on relative prediction accuracy and uncertainty.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 04:23:59 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Lu",
"Hao",
""
],
[
"Thelen",
"Adam",
""
],
[
"Fink",
"Olga",
""
],
[
"Hu",
"Chao",
""
],
[
"Laflamme",
"Simon",
""
]
] | TITLE: Federated Learning with Uncertainty-Based Client Clustering for
Fleet-Wide Fault Diagnosis
ABSTRACT: Operators from various industries have been pushing the adoption of wireless
sensing nodes for industrial monitoring, and such efforts have produced
sizeable condition monitoring datasets that can be used to build diagnosis
algorithms capable of warning maintenance engineers of impending failure or
identifying current system health conditions. However, single operators may not
have sufficiently large fleets of systems or component units to collect
sufficient data to develop data-driven algorithms. Collecting a satisfactory
quantity of fault patterns for safety-critical systems is particularly
difficult due to the rarity of faults. Federated learning (FL) has emerged as a
promising solution to leverage datasets from multiple operators to train a
decentralized asset fault diagnosis model while maintaining data
confidentiality. However, there are still considerable obstacles to overcome
when it comes to optimizing the federation strategy without leaking sensitive
data and addressing the issue of client dataset heterogeneity. This is
particularly prevalent in fault diagnosis applications due to the high
diversity of operating conditions and system configurations. To address these
two challenges, we propose a novel clustering-based FL algorithm where clients
are clustered for federating based on dataset similarity. To quantify dataset
similarity between clients without explicitly sharing data, each client sets
aside a local test dataset and evaluates the other clients' model prediction
accuracy and uncertainty on this test dataset. Clients are then clustered for
FL based on relative prediction accuracy and uncertainty.
| no_new_dataset | 0.709265 |
2304.13277 | Kunzhe Song | Kunzhe Song, Qingfeng Sun, Can Xu, Kai Zheng, Yaming Yang | Self-Supervised Multi-Modal Sequential Recommendation | null | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With the increasing development of e-commerce and online services,
personalized recommendation systems have become crucial for enhancing user
satisfaction and driving business revenue. Traditional sequential
recommendation methods that rely on explicit item IDs encounter challenges in
handling item cold start and domain transfer problems. Recent approaches have
attempted to use modal features associated with items as a replacement for item
IDs, enabling the transfer of learned knowledge across different datasets.
However, these methods typically calculate the correlation between the model's
output and item embeddings, which may suffer from inconsistencies between
high-level feature vectors and low-level feature embeddings, thereby hindering
further model learning. To address this issue, we propose a dual-tower
retrieval architecture for sequence recommendation. In this architecture, the
predicted embedding from the user encoder is used to retrieve the generated
embedding from the item encoder, thereby alleviating the issue of inconsistent
feature levels. Moreover, in order to further improve the retrieval performance
of the model, we also propose a self-supervised multi-modal pretraining method
inspired by the consistency property of contrastive learning. This pretraining
method enables the model to align various feature combinations of items,
thereby effectively generalizing to diverse datasets with different item
features. We evaluate the proposed method on five publicly available datasets
and conduct extensive experiments. The results demonstrate significant
performance improvement of our method.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 04:36:51 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Song",
"Kunzhe",
""
],
[
"Sun",
"Qingfeng",
""
],
[
"Xu",
"Can",
""
],
[
"Zheng",
"Kai",
""
],
[
"Yang",
"Yaming",
""
]
] | TITLE: Self-Supervised Multi-Modal Sequential Recommendation
ABSTRACT: With the increasing development of e-commerce and online services,
personalized recommendation systems have become crucial for enhancing user
satisfaction and driving business revenue. Traditional sequential
recommendation methods that rely on explicit item IDs encounter challenges in
handling item cold start and domain transfer problems. Recent approaches have
attempted to use modal features associated with items as a replacement for item
IDs, enabling the transfer of learned knowledge across different datasets.
However, these methods typically calculate the correlation between the model's
output and item embeddings, which may suffer from inconsistencies between
high-level feature vectors and low-level feature embeddings, thereby hindering
further model learning. To address this issue, we propose a dual-tower
retrieval architecture for sequence recommendation. In this architecture, the
predicted embedding from the user encoder is used to retrieve the generated
embedding from the item encoder, thereby alleviating the issue of inconsistent
feature levels. Moreover, in order to further improve the retrieval performance
of the model, we also propose a self-supervised multi-modal pretraining method
inspired by the consistency property of contrastive learning. This pretraining
method enables the model to align various feature combinations of items,
thereby effectively generalizing to diverse datasets with different item
features. We evaluate the proposed method on five publicly available datasets
and conduct extensive experiments. The results demonstrate significant
performance improvement of our method.
| no_new_dataset | 0.707847 |
2304.13287 | Yi Rong | Yi Rong, Xiongbo Lu, Zhaoyang Sun, Yaxiong Chen, Shengwu Xiong | ESPT: A Self-Supervised Episodic Spatial Pretext Task for Improving
Few-Shot Learning | Accepted by AAAI 2023 | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | Self-supervised learning (SSL) techniques have recently been integrated into
the few-shot learning (FSL) framework and have shown promising results in
improving the few-shot image classification performance. However, existing SSL
approaches used in FSL typically seek the supervision signals from the global
embedding of every single image. Therefore, during the episodic training of
FSL, these methods cannot capture and fully utilize the local visual
information in image samples and the data structure information of the whole
episode, which are beneficial to FSL. To this end, we propose to augment the
few-shot learning objective with a novel self-supervised Episodic Spatial
Pretext Task (ESPT). Specifically, for each few-shot episode, we generate its
corresponding transformed episode by applying a random geometric transformation
to all the images in it. Based on these, our ESPT objective is defined as
maximizing the local spatial relationship consistency between the original
episode and the transformed one. With this definition, the ESPT-augmented FSL
objective promotes learning more transferable feature representations that
capture the local spatial features of different images and their
inter-relational structural information in each input episode, thus enabling
the model to generalize better to new categories with only a few samples.
Extensive experiments indicate that our ESPT method achieves new
state-of-the-art performance for few-shot image classification on three
mainstay benchmark datasets. The source code will be available at:
https://github.com/Whut-YiRong/ESPT.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 04:52:08 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Rong",
"Yi",
""
],
[
"Lu",
"Xiongbo",
""
],
[
"Sun",
"Zhaoyang",
""
],
[
"Chen",
"Yaxiong",
""
],
[
"Xiong",
"Shengwu",
""
]
] | TITLE: ESPT: A Self-Supervised Episodic Spatial Pretext Task for Improving
Few-Shot Learning
ABSTRACT: Self-supervised learning (SSL) techniques have recently been integrated into
the few-shot learning (FSL) framework and have shown promising results in
improving the few-shot image classification performance. However, existing SSL
approaches used in FSL typically seek the supervision signals from the global
embedding of every single image. Therefore, during the episodic training of
FSL, these methods cannot capture and fully utilize the local visual
information in image samples and the data structure information of the whole
episode, which are beneficial to FSL. To this end, we propose to augment the
few-shot learning objective with a novel self-supervised Episodic Spatial
Pretext Task (ESPT). Specifically, for each few-shot episode, we generate its
corresponding transformed episode by applying a random geometric transformation
to all the images in it. Based on these, our ESPT objective is defined as
maximizing the local spatial relationship consistency between the original
episode and the transformed one. With this definition, the ESPT-augmented FSL
objective promotes learning more transferable feature representations that
capture the local spatial features of different images and their
inter-relational structural information in each input episode, thus enabling
the model to generalize better to new categories with only a few samples.
Extensive experiments indicate that our ESPT method achieves new
state-of-the-art performance for few-shot image classification on three
mainstay benchmark datasets. The source code will be available at:
https://github.com/Whut-YiRong/ESPT.
| no_new_dataset | 0.710189 |
2304.13289 | Siqi Wang | Siqi Wang, Tee Hiang Cheng, Meng-Hiot Lim | Membrane Potential Distribution Adjustment and Parametric Surrogate
Gradient in Spiking Neural Networks | 10 pages, 8 figures | null | null | null | cs.LG cs.NE | http://creativecommons.org/licenses/by-nc-nd/4.0/ | As an emerging network model, spiking neural networks (SNNs) have aroused
significant research attentions in recent years. However, the energy-efficient
binary spikes do not augur well with gradient descent-based training
approaches. Surrogate gradient (SG) strategy is investigated and applied to
circumvent this issue and train SNNs from scratch. Due to the lack of
well-recognized SG selection rule, most SGs are chosen intuitively. We propose
the parametric surrogate gradient (PSG) method to iteratively update SG and
eventually determine an optimal surrogate gradient parameter, which calibrates
the shape of candidate SGs. In SNNs, neural potential distribution tends to
deviate unpredictably due to quantization error. We evaluate such potential
shift and propose methodology for potential distribution adjustment (PDA) to
minimize the loss of undesired pre-activations. Experimental results
demonstrate that the proposed methods can be readily integrated with
backpropagation through time (BPTT) algorithm and help modulated SNNs to
achieve state-of-the-art performance on both static and dynamic dataset with
fewer timesteps.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 05:02:41 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Wang",
"Siqi",
""
],
[
"Cheng",
"Tee Hiang",
""
],
[
"Lim",
"Meng-Hiot",
""
]
] | TITLE: Membrane Potential Distribution Adjustment and Parametric Surrogate
Gradient in Spiking Neural Networks
ABSTRACT: As an emerging network model, spiking neural networks (SNNs) have aroused
significant research attentions in recent years. However, the energy-efficient
binary spikes do not augur well with gradient descent-based training
approaches. Surrogate gradient (SG) strategy is investigated and applied to
circumvent this issue and train SNNs from scratch. Due to the lack of
well-recognized SG selection rule, most SGs are chosen intuitively. We propose
the parametric surrogate gradient (PSG) method to iteratively update SG and
eventually determine an optimal surrogate gradient parameter, which calibrates
the shape of candidate SGs. In SNNs, neural potential distribution tends to
deviate unpredictably due to quantization error. We evaluate such potential
shift and propose methodology for potential distribution adjustment (PDA) to
minimize the loss of undesired pre-activations. Experimental results
demonstrate that the proposed methods can be readily integrated with
backpropagation through time (BPTT) algorithm and help modulated SNNs to
achieve state-of-the-art performance on both static and dynamic dataset with
fewer timesteps.
| no_new_dataset | 0.708421 |
2304.13290 | Jia-Huei Ju | Jia-Huei Ju, Sheng-Chieh Lin, Ming-Feng Tsai, Chuan-Ju Wang | Improving Conversational Passage Re-ranking with View Ensemble | SIGIR 2023 | null | 10.1145/3539618.3592002 | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents ConvRerank, a conversational passage re-ranker that
employs a newly developed pseudo-labeling approach. Our proposed view-ensemble
method enhances the quality of pseudo-labeled data, thus improving the
fine-tuning of ConvRerank. Our experimental evaluation on benchmark datasets
shows that combining ConvRerank with a conversational dense retriever in a
cascaded manner achieves a good balance between effectiveness and efficiency.
Compared to baseline methods, our cascaded pipeline demonstrates lower latency
and higher top-ranking effectiveness. Furthermore, the in-depth analysis
confirms the potential of our approach to improving the effectiveness of
conversational search.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 05:06:22 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Ju",
"Jia-Huei",
""
],
[
"Lin",
"Sheng-Chieh",
""
],
[
"Tsai",
"Ming-Feng",
""
],
[
"Wang",
"Chuan-Ju",
""
]
] | TITLE: Improving Conversational Passage Re-ranking with View Ensemble
ABSTRACT: This paper presents ConvRerank, a conversational passage re-ranker that
employs a newly developed pseudo-labeling approach. Our proposed view-ensemble
method enhances the quality of pseudo-labeled data, thus improving the
fine-tuning of ConvRerank. Our experimental evaluation on benchmark datasets
shows that combining ConvRerank with a conversational dense retriever in a
cascaded manner achieves a good balance between effectiveness and efficiency.
Compared to baseline methods, our cascaded pipeline demonstrates lower latency
and higher top-ranking effectiveness. Furthermore, the in-depth analysis
confirms the potential of our approach to improving the effectiveness of
conversational search.
| no_new_dataset | 0.709617 |
2304.13327 | KANN Bonpagna | Bonpagna Kann (UGA, M-PSI), Sandra Castellanos-Paez (UGA, M-PSI),
Philippe Lalanda (UGA, M-PSI) | Evaluation of Regularization-based Continual Learning Approaches:
Application to HAR | null | 2023 IEEE International Conference on Pervasive Computing and
Communications Workshops and other Affiliated Events (PerCom Workshops), Mar
2023, Atlanta, United States | null | null | cs.AI stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Pervasive computing allows the provision of services in many important areas,
including the relevant and dynamic field of health and well-being. In this
domain, Human Activity Recognition (HAR) has gained a lot of attention in
recent years. Current solutions rely on Machine Learning (ML) models and
achieve impressive results. However, the evolution of these models remains
difficult, as long as a complete retraining is not performed. To overcome this
problem, the concept of Continual Learning is very promising today and, more
particularly, the techniques based on regularization. These techniques are
particularly interesting for their simplicity and their low cost. Initial
studies have been conducted and have shown promising outcomes. However, they
remain very specific and difficult to compare. In this paper, we provide a
comprehensive comparison of three regularization-based methods that we adapted
to the HAR domain, highlighting their strengths and limitations. Our
experiments were conducted on the UCI HAR dataset and the results showed that
no single technique outperformed all others in all scenarios considered.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 06:56:49 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Kann",
"Bonpagna",
"",
"UGA, M-PSI"
],
[
"Castellanos-Paez",
"Sandra",
"",
"UGA, M-PSI"
],
[
"Lalanda",
"Philippe",
"",
"UGA, M-PSI"
]
] | TITLE: Evaluation of Regularization-based Continual Learning Approaches:
Application to HAR
ABSTRACT: Pervasive computing allows the provision of services in many important areas,
including the relevant and dynamic field of health and well-being. In this
domain, Human Activity Recognition (HAR) has gained a lot of attention in
recent years. Current solutions rely on Machine Learning (ML) models and
achieve impressive results. However, the evolution of these models remains
difficult, as long as a complete retraining is not performed. To overcome this
problem, the concept of Continual Learning is very promising today and, more
particularly, the techniques based on regularization. These techniques are
particularly interesting for their simplicity and their low cost. Initial
studies have been conducted and have shown promising outcomes. However, they
remain very specific and difficult to compare. In this paper, we provide a
comprehensive comparison of three regularization-based methods that we adapted
to the HAR domain, highlighting their strengths and limitations. Our
experiments were conducted on the UCI HAR dataset and the results showed that
no single technique outperformed all others in all scenarios considered.
| no_new_dataset | 0.708364 |
2304.13350 | Manasi Patwardhan | Krishnam Hasija, Shrishti Pradhan, Manasi Patwardhan, Raveendra Kumar
Medicherla, Lovekesh Vig, Ravindra Naik | Neuro-symbolic Zero-Shot Code Cloning with Cross-Language Intermediate
Representation | 10 pages, 4 tables, 2 figures | null | null | null | cs.AI cs.SE | http://creativecommons.org/publicdomain/zero/1.0/ | In this paper, we define a neuro-symbolic approach to address the task of
finding semantically similar clones for the codes of the legacy programming
language COBOL, without training data. We define a meta-model that is
instantiated to have an Intermediate Representation (IR) in the form of
Abstract Syntax Trees (ASTs) common across codes in C and COBOL. We linearize
the IRs using Structure Based Traversal (SBT) to create sequential inputs. We
further fine-tune UnixCoder, the best-performing model for zero-shot
cross-programming language code search, for the Code Cloning task with the SBT
IRs of C code-pairs, available in the CodeNet dataset. This allows us to learn
latent representations for the IRs of the C codes, which are transferable to
the IRs of the COBOL codes. With this fine-tuned UnixCoder, we get a
performance improvement of 12.85 MAP@2 over the pre-trained UniXCoder model, in
a zero-shot setting, on the COBOL test split synthesized from the CodeNet
dataset. This demonstrates the efficacy of our meta-model based approach to
facilitate cross-programming language transfer.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 07:41:26 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Hasija",
"Krishnam",
""
],
[
"Pradhan",
"Shrishti",
""
],
[
"Patwardhan",
"Manasi",
""
],
[
"Medicherla",
"Raveendra Kumar",
""
],
[
"Vig",
"Lovekesh",
""
],
[
"Naik",
"Ravindra",
""
]
] | TITLE: Neuro-symbolic Zero-Shot Code Cloning with Cross-Language Intermediate
Representation
ABSTRACT: In this paper, we define a neuro-symbolic approach to address the task of
finding semantically similar clones for the codes of the legacy programming
language COBOL, without training data. We define a meta-model that is
instantiated to have an Intermediate Representation (IR) in the form of
Abstract Syntax Trees (ASTs) common across codes in C and COBOL. We linearize
the IRs using Structure Based Traversal (SBT) to create sequential inputs. We
further fine-tune UnixCoder, the best-performing model for zero-shot
cross-programming language code search, for the Code Cloning task with the SBT
IRs of C code-pairs, available in the CodeNet dataset. This allows us to learn
latent representations for the IRs of the C codes, which are transferable to
the IRs of the COBOL codes. With this fine-tuned UnixCoder, we get a
performance improvement of 12.85 MAP@2 over the pre-trained UniXCoder model, in
a zero-shot setting, on the COBOL test split synthesized from the CodeNet
dataset. This demonstrates the efficacy of our meta-model based approach to
facilitate cross-programming language transfer.
| no_new_dataset | 0.712226 |
2304.13352 | Aditya Kalapaaking | Aditya Pribadi Kalapaaking, Veronika Stephanie, Ibrahim Khalil,
Mohammed Atiquzzaman, Xun Yi, Mahathir Almashor | SMPC-based Federated Learning for 6G enabled Internet of Medical Things | null | IEEE Network, vol. 36, no. 4, pp. 182-189, July/August 2022 | 10.1109/MNET.007.2100717 | null | cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Rapidly developing intelligent healthcare systems are underpinned by Sixth
Generation (6G) connectivity, ubiquitous Internet of Things (IoT), and Deep
Learning (DL) techniques. This portends a future where 6G powers the Internet
of Medical Things (IoMT) with seamless, large-scale, and real-time connectivity
amongst entities. This article proposes a Convolutional Neural Network (CNN)
based Federated Learning framework that combines Secure Multi-Party Computation
(SMPC) based aggregation and Encrypted Inference methods, all within the
context of 6G and IoMT. We consider multiple hospitals with clusters of mixed
IoMT and edge devices that encrypt locally trained models. Subsequently, each
hospital sends the encrypted local models for SMPC-based encrypted aggregation
in the cloud, which generates the encrypted global model. Ultimately, the
encrypted global model is returned to each edge server for more localized
training, further improving model accuracy. Moreover, hospitals can perform
encrypted inference on their edge servers or the cloud while maintaining data
and model privacy. Multiple experiments were conducted with varying CNN models
and datasets to evaluate the proposed framework's performance.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 07:46:07 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Kalapaaking",
"Aditya Pribadi",
""
],
[
"Stephanie",
"Veronika",
""
],
[
"Khalil",
"Ibrahim",
""
],
[
"Atiquzzaman",
"Mohammed",
""
],
[
"Yi",
"Xun",
""
],
[
"Almashor",
"Mahathir",
""
]
] | TITLE: SMPC-based Federated Learning for 6G enabled Internet of Medical Things
ABSTRACT: Rapidly developing intelligent healthcare systems are underpinned by Sixth
Generation (6G) connectivity, ubiquitous Internet of Things (IoT), and Deep
Learning (DL) techniques. This portends a future where 6G powers the Internet
of Medical Things (IoMT) with seamless, large-scale, and real-time connectivity
amongst entities. This article proposes a Convolutional Neural Network (CNN)
based Federated Learning framework that combines Secure Multi-Party Computation
(SMPC) based aggregation and Encrypted Inference methods, all within the
context of 6G and IoMT. We consider multiple hospitals with clusters of mixed
IoMT and edge devices that encrypt locally trained models. Subsequently, each
hospital sends the encrypted local models for SMPC-based encrypted aggregation
in the cloud, which generates the encrypted global model. Ultimately, the
encrypted global model is returned to each edge server for more localized
training, further improving model accuracy. Moreover, hospitals can perform
encrypted inference on their edge servers or the cloud while maintaining data
and model privacy. Multiple experiments were conducted with varying CNN models
and datasets to evaluate the proposed framework's performance.
| no_new_dataset | 0.711049 |
2304.13357 | Liming Xu | Liming Xu, Hanqi Li, Bochuan Zheng, Weisheng Li, Jiancheng Lv | Deep Lifelong Cross-modal Hashing | null | null | null | null | cs.CV cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Hashing methods have made significant progress in cross-modal retrieval tasks
with fast query speed and low storage cost. Among them, deep learning-based
hashing achieves better performance on large-scale data due to its excellent
extraction and representation ability for nonlinear heterogeneous features.
However, there are still two main challenges in catastrophic forgetting when
data with new categories arrive continuously, and time-consuming for
non-continuous hashing retrieval to retrain for updating. To this end, we, in
this paper, propose a novel deep lifelong cross-modal hashing to achieve
lifelong hashing retrieval instead of re-training hash function repeatedly when
new data arrive. Specifically, we design lifelong learning strategy to update
hash functions by directly training the incremental data instead of retraining
new hash functions using all the accumulated data, which significantly reduce
training time. Then, we propose lifelong hashing loss to enable original hash
codes participate in lifelong learning but remain invariant, and further
preserve the similarity and dis-similarity among original and incremental hash
codes to maintain performance. Additionally, considering distribution
heterogeneity when new data arriving continuously, we introduce multi-label
semantic similarity to supervise hash learning, and it has been proven that the
similarity improves performance with detailed analysis. Experimental results on
benchmark datasets show that the proposed methods achieves comparative
performance comparing with recent state-of-the-art cross-modal hashing methods,
and it yields substantial average increments over 20\% in retrieval accuracy
and almost reduces over 80\% training time when new data arrives continuously.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 07:56:22 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Xu",
"Liming",
""
],
[
"Li",
"Hanqi",
""
],
[
"Zheng",
"Bochuan",
""
],
[
"Li",
"Weisheng",
""
],
[
"Lv",
"Jiancheng",
""
]
] | TITLE: Deep Lifelong Cross-modal Hashing
ABSTRACT: Hashing methods have made significant progress in cross-modal retrieval tasks
with fast query speed and low storage cost. Among them, deep learning-based
hashing achieves better performance on large-scale data due to its excellent
extraction and representation ability for nonlinear heterogeneous features.
However, there are still two main challenges in catastrophic forgetting when
data with new categories arrive continuously, and time-consuming for
non-continuous hashing retrieval to retrain for updating. To this end, we, in
this paper, propose a novel deep lifelong cross-modal hashing to achieve
lifelong hashing retrieval instead of re-training hash function repeatedly when
new data arrive. Specifically, we design lifelong learning strategy to update
hash functions by directly training the incremental data instead of retraining
new hash functions using all the accumulated data, which significantly reduce
training time. Then, we propose lifelong hashing loss to enable original hash
codes participate in lifelong learning but remain invariant, and further
preserve the similarity and dis-similarity among original and incremental hash
codes to maintain performance. Additionally, considering distribution
heterogeneity when new data arriving continuously, we introduce multi-label
semantic similarity to supervise hash learning, and it has been proven that the
similarity improves performance with detailed analysis. Experimental results on
benchmark datasets show that the proposed methods achieves comparative
performance comparing with recent state-of-the-art cross-modal hashing methods,
and it yields substantial average increments over 20\% in retrieval accuracy
and almost reduces over 80\% training time when new data arrives continuously.
| no_new_dataset | 0.710584 |
2304.13360 | Aditya Kalapaaking | Aditya Pribadi Kalapaaking, Ibrahim Khalil and Xun Yi | Blockchain-based Federated Learning with SMPC Model Verification Against
Poisoning Attack for Healthcare Systems | null | null | 10.1109/TETC.2023.3268186 | null | cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Due to the rising awareness of privacy and security in machine learning
applications, federated learning (FL) has received widespread attention and
applied to several areas, e.g., intelligence healthcare systems, IoT-based
industries, and smart cities. FL enables clients to train a global model
collaboratively without accessing their local training data. However, the
current FL schemes are vulnerable to adversarial attacks. Its architecture
makes detecting and defending against malicious model updates difficult. In
addition, most recent studies to detect FL from malicious updates while
maintaining the model's privacy have not been sufficiently explored. This paper
proposed blockchain-based federated learning with SMPC model verification
against poisoning attacks for healthcare systems. First, we check the machine
learning model from the FL participants through an encrypted inference process
and remove the compromised model. Once the participants' local models have been
verified, the models are sent to the blockchain node to be securely aggregated.
We conducted several experiments with different medical datasets to evaluate
our proposed framework.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 08:04:46 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Kalapaaking",
"Aditya Pribadi",
""
],
[
"Khalil",
"Ibrahim",
""
],
[
"Yi",
"Xun",
""
]
] | TITLE: Blockchain-based Federated Learning with SMPC Model Verification Against
Poisoning Attack for Healthcare Systems
ABSTRACT: Due to the rising awareness of privacy and security in machine learning
applications, federated learning (FL) has received widespread attention and
applied to several areas, e.g., intelligence healthcare systems, IoT-based
industries, and smart cities. FL enables clients to train a global model
collaboratively without accessing their local training data. However, the
current FL schemes are vulnerable to adversarial attacks. Its architecture
makes detecting and defending against malicious model updates difficult. In
addition, most recent studies to detect FL from malicious updates while
maintaining the model's privacy have not been sufficiently explored. This paper
proposed blockchain-based federated learning with SMPC model verification
against poisoning attacks for healthcare systems. First, we check the machine
learning model from the FL participants through an encrypted inference process
and remove the compromised model. Once the participants' local models have been
verified, the models are sent to the blockchain node to be securely aggregated.
We conducted several experiments with different medical datasets to evaluate
our proposed framework.
| no_new_dataset | 0.709447 |
2304.13366 | Eslam Eldeeb | Eslam Eldeeb and Hirley Alves | LoRaWAN-enabled Smart Campus: The Dataset and a People Counter Use Case | null | null | null | null | cs.LG cs.AI cs.NI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | IoT has a significant role in the smart campus. This paper presents a
detailed description of the Smart Campus dataset based on LoRaWAN. LoRaWAN is
an emerging technology that enables serving hundreds of IoT devices. First, we
describe the LoRa network that connects the devices to the server. Afterward,
we analyze the missing transmissions and propose a k-nearest neighbor solution
to handle the missing values. Then, we predict future readings using a long
short-term memory (LSTM). Finally, as one example application, we build a deep
neural network to predict the number of people inside a room based on the
selected sensor's readings. Our results show that our model achieves an
accuracy of $95 \: \%$ in predicting the number of people. Moreover, the
dataset is openly available and described in detail, which is opportunity for
exploration of other features and applications.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 08:14:56 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Eldeeb",
"Eslam",
""
],
[
"Alves",
"Hirley",
""
]
] | TITLE: LoRaWAN-enabled Smart Campus: The Dataset and a People Counter Use Case
ABSTRACT: IoT has a significant role in the smart campus. This paper presents a
detailed description of the Smart Campus dataset based on LoRaWAN. LoRaWAN is
an emerging technology that enables serving hundreds of IoT devices. First, we
describe the LoRa network that connects the devices to the server. Afterward,
we analyze the missing transmissions and propose a k-nearest neighbor solution
to handle the missing values. Then, we predict future readings using a long
short-term memory (LSTM). Finally, as one example application, we build a deep
neural network to predict the number of people inside a room based on the
selected sensor's readings. Our results show that our model achieves an
accuracy of $95 \: \%$ in predicting the number of people. Moreover, the
dataset is openly available and described in detail, which is opportunity for
exploration of other features and applications.
| no_new_dataset | 0.708042 |
2304.13374 | Tan Zhiquan | Zhiquan Tan, Zihao Wang, Yifan Zhang | SEAL: Simultaneous Label Hierarchy Exploration And Learning | null | null | null | null | cs.LG cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Label hierarchy is an important source of external knowledge that can enhance
classification performance. However, most existing methods rely on predefined
label hierarchies that may not match the data distribution. To address this
issue, we propose Simultaneous label hierarchy Exploration And Learning (SEAL),
a new framework that explores the label hierarchy by augmenting the observed
labels with latent labels that follow a prior hierarchical structure. Our
approach uses a 1-Wasserstein metric over the tree metric space as an objective
function, which enables us to simultaneously learn a data-driven label
hierarchy and perform (semi-)supervised learning. We evaluate our method on
several datasets and show that it achieves superior results in both supervised
and semi-supervised scenarios and reveals insightful label structures. Our
implementation is available at https://github.com/tzq1999/SEAL.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 08:31:59 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Tan",
"Zhiquan",
""
],
[
"Wang",
"Zihao",
""
],
[
"Zhang",
"Yifan",
""
]
] | TITLE: SEAL: Simultaneous Label Hierarchy Exploration And Learning
ABSTRACT: Label hierarchy is an important source of external knowledge that can enhance
classification performance. However, most existing methods rely on predefined
label hierarchies that may not match the data distribution. To address this
issue, we propose Simultaneous label hierarchy Exploration And Learning (SEAL),
a new framework that explores the label hierarchy by augmenting the observed
labels with latent labels that follow a prior hierarchical structure. Our
approach uses a 1-Wasserstein metric over the tree metric space as an objective
function, which enables us to simultaneously learn a data-driven label
hierarchy and perform (semi-)supervised learning. We evaluate our method on
several datasets and show that it achieves superior results in both supervised
and semi-supervised scenarios and reveals insightful label structures. Our
implementation is available at https://github.com/tzq1999/SEAL.
| no_new_dataset | 0.712013 |
2304.13397 | Jun Wang | Xiaorui Wang, Jun Wang, Xin Tang, Peng Gao, Rui Fang, Guotong Xie | Filter Pruning via Filters Similarity in Consecutive Layers | Accepted by ICASSP 2023 (oral) | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | Filter pruning is widely adopted to compress and accelerate the Convolutional
Neural Networks (CNNs), but most previous works ignore the relationship between
filters and channels in different layers. Processing each layer independently
fails to utilize the collaborative relationship across layers. In this paper,
we intuitively propose a novel pruning method by explicitly leveraging the
Filters Similarity in Consecutive Layers (FSCL). FSCL compresses models by
pruning filters whose corresponding features are more worthless in the model.
The extensive experiments demonstrate the effectiveness of FSCL, and it yields
remarkable improvement over state-of-the-art on accuracy, FLOPs and parameter
reduction on several benchmark models and datasets.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 09:18:38 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Wang",
"Xiaorui",
""
],
[
"Wang",
"Jun",
""
],
[
"Tang",
"Xin",
""
],
[
"Gao",
"Peng",
""
],
[
"Fang",
"Rui",
""
],
[
"Xie",
"Guotong",
""
]
] | TITLE: Filter Pruning via Filters Similarity in Consecutive Layers
ABSTRACT: Filter pruning is widely adopted to compress and accelerate the Convolutional
Neural Networks (CNNs), but most previous works ignore the relationship between
filters and channels in different layers. Processing each layer independently
fails to utilize the collaborative relationship across layers. In this paper,
we intuitively propose a novel pruning method by explicitly leveraging the
Filters Similarity in Consecutive Layers (FSCL). FSCL compresses models by
pruning filters whose corresponding features are more worthless in the model.
The extensive experiments demonstrate the effectiveness of FSCL, and it yields
remarkable improvement over state-of-the-art on accuracy, FLOPs and parameter
reduction on several benchmark models and datasets.
| no_new_dataset | 0.713619 |
2304.13398 | Longkun Guo | Weijie Fang, Yanggeng Fu, Jiaquan Gao, Longkun Guo, Gregory Gutin,
Xiaoyan Zhang | Acceleration for Timing-Aware Gate-Level Logic Simulation with One-Pass
GPU Parallelism | null | null | null | null | cs.DS cs.AR cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Witnessing the advancing scale and complexity of chip design and benefiting
from high-performance computation technologies, the simulation of Very Large
Scale Integration (VLSI) Circuits imposes an increasing requirement for
acceleration through parallel computing with GPU devices. However, the
conventional parallel strategies do not fully align with modern GPU abilities,
leading to new challenges in the parallelism of VLSI simulation when using GPU,
despite some previous successful demonstrations of significant acceleration. In
this paper, we propose a novel approach to accelerate 4-value logic
timing-aware gate-level logic simulation using waveform-based GPU parallelism.
Our approach utilizes a new strategy that can effectively handle the dependency
between tasks during the parallelism, reducing the synchronization requirement
between CPU and GPU when parallelizing the simulation on combinational
circuits. This approach requires only one round of data transfer and hence
achieves one-pass parallelism. Moreover, to overcome the difficulty within the
adoption of our strategy in GPU devices, we design a series of data structures
and tune them to dynamically allocate and store new-generated output with
uncertain scale. Finally, experiments are carried out on industrial-scale
open-source benchmarks to demonstrate the performance gain of our approach
compared to several state-of-the-art baselines.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 09:19:21 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Fang",
"Weijie",
""
],
[
"Fu",
"Yanggeng",
""
],
[
"Gao",
"Jiaquan",
""
],
[
"Guo",
"Longkun",
""
],
[
"Gutin",
"Gregory",
""
],
[
"Zhang",
"Xiaoyan",
""
]
] | TITLE: Acceleration for Timing-Aware Gate-Level Logic Simulation with One-Pass
GPU Parallelism
ABSTRACT: Witnessing the advancing scale and complexity of chip design and benefiting
from high-performance computation technologies, the simulation of Very Large
Scale Integration (VLSI) Circuits imposes an increasing requirement for
acceleration through parallel computing with GPU devices. However, the
conventional parallel strategies do not fully align with modern GPU abilities,
leading to new challenges in the parallelism of VLSI simulation when using GPU,
despite some previous successful demonstrations of significant acceleration. In
this paper, we propose a novel approach to accelerate 4-value logic
timing-aware gate-level logic simulation using waveform-based GPU parallelism.
Our approach utilizes a new strategy that can effectively handle the dependency
between tasks during the parallelism, reducing the synchronization requirement
between CPU and GPU when parallelizing the simulation on combinational
circuits. This approach requires only one round of data transfer and hence
achieves one-pass parallelism. Moreover, to overcome the difficulty within the
adoption of our strategy in GPU devices, we design a series of data structures
and tune them to dynamically allocate and store new-generated output with
uncertain scale. Finally, experiments are carried out on industrial-scale
open-source benchmarks to demonstrate the performance gain of our approach
compared to several state-of-the-art baselines.
| no_new_dataset | 0.709416 |
2304.13403 | Luca Ciampi | Pawe{\l} Foszner, Agnieszka Szcz\k{e}sna, Luca Ciampi, Nicola Messina,
Adam Cygan, Bartosz Bizo\'n, Micha{\l} Cogiel, Dominik Golba, El\.zbieta
Macioszek, Micha{\l} Staniszewski | Development of a Realistic Crowd Simulation Environment for Fine-grained
Validation of People Tracking Methods | null | Proceedings of the 18th International Joint Conference on Computer
Vision, Imaging and Computer Graphics Theory and Applications, 2023 | 10.5220/0011691500003417 | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Generally, crowd datasets can be collected or generated from real or
synthetic sources. Real data is generated by using infrastructure-based sensors
(such as static cameras or other sensors). The use of simulation tools can
significantly reduce the time required to generate scenario-specific crowd
datasets, facilitate data-driven research, and next build functional machine
learning models. The main goal of this work was to develop an extension of
crowd simulation (named CrowdSim2) and prove its usability in the application
of people-tracking algorithms. The simulator is developed using the very
popular Unity 3D engine with particular emphasis on the aspects of realism in
the environment, weather conditions, traffic, and the movement and models of
individual agents. Finally, three methods of tracking were used to validate
generated dataset: IOU-Tracker, Deep-Sort, and Deep-TAMA.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 09:29:58 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Foszner",
"Paweł",
""
],
[
"Szczęsna",
"Agnieszka",
""
],
[
"Ciampi",
"Luca",
""
],
[
"Messina",
"Nicola",
""
],
[
"Cygan",
"Adam",
""
],
[
"Bizoń",
"Bartosz",
""
],
[
"Cogiel",
"Michał",
""
],
[
"Golba",
"Dominik",
""
],
[
"Macioszek",
"Elżbieta",
""
],
[
"Staniszewski",
"Michał",
""
]
] | TITLE: Development of a Realistic Crowd Simulation Environment for Fine-grained
Validation of People Tracking Methods
ABSTRACT: Generally, crowd datasets can be collected or generated from real or
synthetic sources. Real data is generated by using infrastructure-based sensors
(such as static cameras or other sensors). The use of simulation tools can
significantly reduce the time required to generate scenario-specific crowd
datasets, facilitate data-driven research, and next build functional machine
learning models. The main goal of this work was to develop an extension of
crowd simulation (named CrowdSim2) and prove its usability in the application
of people-tracking algorithms. The simulator is developed using the very
popular Unity 3D engine with particular emphasis on the aspects of realism in
the environment, weather conditions, traffic, and the movement and models of
individual agents. Finally, three methods of tracking were used to validate
generated dataset: IOU-Tracker, Deep-Sort, and Deep-TAMA.
| no_new_dataset | 0.709038 |
2304.13425 | Zhongxi Qiu | Zhongxi Qiu and Yan Hu and Heng Li and Jiang Liu | Learnable Ophthalmology SAM | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Segmentation is vital for ophthalmology image analysis. But its various modal
images hinder most of the existing segmentation algorithms applications, as
they rely on training based on a large number of labels or hold weak
generalization ability. Based on Segment Anything (SAM), we propose a simple
but effective learnable prompt layer suitable for multiple target segmentation
in ophthalmology multi-modal images, named Learnable Ophthalmology Segment
Anything (SAM). The learnable prompt layer learns medical prior knowledge from
each transformer layer. During training, we only train the prompt layer and
task head based on a one-shot mechanism. We demonstrate the effectiveness of
our thought based on four medical segmentation tasks based on nine publicly
available datasets. Moreover, we only provide a new improvement thought for
applying the existing fundamental CV models in the medical field. Our codes are
available at \href{https://github.com/Qsingle/LearnablePromptSAM}{website}.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 10:14:03 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Qiu",
"Zhongxi",
""
],
[
"Hu",
"Yan",
""
],
[
"Li",
"Heng",
""
],
[
"Liu",
"Jiang",
""
]
] | TITLE: Learnable Ophthalmology SAM
ABSTRACT: Segmentation is vital for ophthalmology image analysis. But its various modal
images hinder most of the existing segmentation algorithms applications, as
they rely on training based on a large number of labels or hold weak
generalization ability. Based on Segment Anything (SAM), we propose a simple
but effective learnable prompt layer suitable for multiple target segmentation
in ophthalmology multi-modal images, named Learnable Ophthalmology Segment
Anything (SAM). The learnable prompt layer learns medical prior knowledge from
each transformer layer. During training, we only train the prompt layer and
task head based on a one-shot mechanism. We demonstrate the effectiveness of
our thought based on four medical segmentation tasks based on nine publicly
available datasets. Moreover, we only provide a new improvement thought for
applying the existing fundamental CV models in the medical field. Our codes are
available at \href{https://github.com/Qsingle/LearnablePromptSAM}{website}.
| no_new_dataset | 0.709806 |
2304.13428 | Timo Kaiser | Timo Kaiser, Christoph Reinders, Bodo Rosenhahn | Compensation Learning in Semantic Segmentation | 8 pages, 6 figures, 4 tables, Vision Datasets Understanding Workshop
on CVPR23 | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Label noise and ambiguities between similar classes are challenging problems
in developing new models and annotating new data for semantic segmentation. In
this paper, we propose Compensation Learning in Semantic Segmentation, a
framework to identify and compensate ambiguities as well as label noise. More
specifically, we add a ground truth depending and globally learned bias to the
classification logits and introduce a novel uncertainty branch for neural
networks to induce the compensation bias only to relevant regions. Our method
is employed into state-of-the-art segmentation frameworks and several
experiments demonstrate that our proposed compensation learns inter-class
relations that allow global identification of challenging ambiguities as well
as the exact localization of subsequent label noise. Additionally, it enlarges
robustness against label noise during training and allows target-oriented
manipulation during inference. We evaluate the proposed method on %the widely
used datasets Cityscapes, KITTI-STEP, ADE20k, and COCO-stuff10k.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 10:26:11 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Kaiser",
"Timo",
""
],
[
"Reinders",
"Christoph",
""
],
[
"Rosenhahn",
"Bodo",
""
]
] | TITLE: Compensation Learning in Semantic Segmentation
ABSTRACT: Label noise and ambiguities between similar classes are challenging problems
in developing new models and annotating new data for semantic segmentation. In
this paper, we propose Compensation Learning in Semantic Segmentation, a
framework to identify and compensate ambiguities as well as label noise. More
specifically, we add a ground truth depending and globally learned bias to the
classification logits and introduce a novel uncertainty branch for neural
networks to induce the compensation bias only to relevant regions. Our method
is employed into state-of-the-art segmentation frameworks and several
experiments demonstrate that our proposed compensation learns inter-class
relations that allow global identification of challenging ambiguities as well
as the exact localization of subsequent label noise. Additionally, it enlarges
robustness against label noise during training and allows target-oriented
manipulation during inference. We evaluate the proposed method on %the widely
used datasets Cityscapes, KITTI-STEP, ADE20k, and COCO-stuff10k.
| no_new_dataset | 0.711243 |
2304.13429 | Michael Bidollahkhani | Michael Bidollahkhani, Ferhat Atasoy, Elnaz Abedini, Ali Davar, Omid
Hamza, F{\i}rat Sefao\u{g}lu, Amin Jafari, Muhammed Nadir Yal\c{c}{\i}n,
Hamdan Abdellatef | GENIE-NF-AI: Identifying Neurofibromatosis Tumors using Liquid Neural
Network (LTC) trained on AACR GENIE Datasets | The authors would like to acknowledge the American Association for
Cancer Research and its financial and material support in the development of
the AACR Project GENIE registry, as well as members of the consortium for
their commitment to data sharing. Interpretations are the responsibility of
study authors | null | null | null | cs.LG cs.NE | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In recent years, the field of medicine has been increasingly adopting
artificial intelligence (AI) technologies to provide faster and more accurate
disease detection, prediction, and assessment. In this study, we propose an
interpretable AI approach to diagnose patients with neurofibromatosis using
blood tests and pathogenic variables. We evaluated the proposed method using a
dataset from the AACR GENIE project and compared its performance with modern
approaches. Our proposed approach outperformed existing models with 99.86%
accuracy. We also conducted NF1 and interpretable AI tests to validate our
approach. Our work provides an explainable approach model using logistic
regression and explanatory stimulus as well as a black-box model. The
explainable models help to explain the predictions of black-box models while
the glass-box models provide information about the best-fit features. Overall,
our study presents an interpretable AI approach for diagnosing patients with
neurofibromatosis and demonstrates the potential of AI in the medical field.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 10:28:59 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Bidollahkhani",
"Michael",
""
],
[
"Atasoy",
"Ferhat",
""
],
[
"Abedini",
"Elnaz",
""
],
[
"Davar",
"Ali",
""
],
[
"Hamza",
"Omid",
""
],
[
"Sefaoğlu",
"Fırat",
""
],
[
"Jafari",
"Amin",
""
],
[
"Yalçın",
"Muhammed Nadir",
""
],
[
"Abdellatef",
"Hamdan",
""
]
] | TITLE: GENIE-NF-AI: Identifying Neurofibromatosis Tumors using Liquid Neural
Network (LTC) trained on AACR GENIE Datasets
ABSTRACT: In recent years, the field of medicine has been increasingly adopting
artificial intelligence (AI) technologies to provide faster and more accurate
disease detection, prediction, and assessment. In this study, we propose an
interpretable AI approach to diagnose patients with neurofibromatosis using
blood tests and pathogenic variables. We evaluated the proposed method using a
dataset from the AACR GENIE project and compared its performance with modern
approaches. Our proposed approach outperformed existing models with 99.86%
accuracy. We also conducted NF1 and interpretable AI tests to validate our
approach. Our work provides an explainable approach model using logistic
regression and explanatory stimulus as well as a black-box model. The
explainable models help to explain the predictions of black-box models while
the glass-box models provide information about the best-fit features. Overall,
our study presents an interpretable AI approach for diagnosing patients with
neurofibromatosis and demonstrates the potential of AI in the medical field.
| no_new_dataset | 0.712063 |
2304.13490 | Xiaoqing Liu | Xiaoqing Liu, Kenji Ono, Ryoma Bise | Mixing Data Augmentation with Preserving Foreground Regions in Medical
Image Segmentation | Accepted by IEEE ISBI'23 | null | null | null | eess.IV cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The development of medical image segmentation using deep learning can
significantly support doctors' diagnoses. Deep learning needs large amounts of
data for training, which also requires data augmentation to extend diversity
for preventing overfitting. However, the existing methods for data augmentation
of medical image segmentation are mainly based on models which need to update
parameters and cost extra computing resources. We proposed data augmentation
methods designed to train a high accuracy deep learning network for medical
image segmentation. The proposed data augmentation approaches are called
KeepMask and KeepMix, which can create medical images by better identifying the
boundary of the organ with no more parameters. Our methods achieved better
performance and obtained more precise boundaries for medical image segmentation
on datasets. The dice coefficient of our methods achieved 94.15% (3.04% higher
than baseline) on CHAOS and 74.70% (5.25% higher than baseline) on MSD spleen
with Unet.
| [
{
"version": "v1",
"created": "Wed, 26 Apr 2023 12:29:54 GMT"
}
] | 2023-04-27T00:00:00 | [
[
"Liu",
"Xiaoqing",
""
],
[
"Ono",
"Kenji",
""
],
[
"Bise",
"Ryoma",
""
]
] | TITLE: Mixing Data Augmentation with Preserving Foreground Regions in Medical
Image Segmentation
ABSTRACT: The development of medical image segmentation using deep learning can
significantly support doctors' diagnoses. Deep learning needs large amounts of
data for training, which also requires data augmentation to extend diversity
for preventing overfitting. However, the existing methods for data augmentation
of medical image segmentation are mainly based on models which need to update
parameters and cost extra computing resources. We proposed data augmentation
methods designed to train a high accuracy deep learning network for medical
image segmentation. The proposed data augmentation approaches are called
KeepMask and KeepMix, which can create medical images by better identifying the
boundary of the organ with no more parameters. Our methods achieved better
performance and obtained more precise boundaries for medical image segmentation
on datasets. The dice coefficient of our methods achieved 94.15% (3.04% higher
than baseline) on CHAOS and 74.70% (5.25% higher than baseline) on MSD spleen
with Unet.
| no_new_dataset | 0.712657 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.