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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2304.12743 | Islem Bouzenia | Islem Bouzenia, Yangruibo Ding, Kexin Pei, Baishakhi Ray, Michael
Pradel | TraceFixer: Execution Trace-Driven Program Repair | null | null | null | null | cs.SE | http://creativecommons.org/licenses/by/4.0/ | When debugging unintended program behavior, developers can often identify the
point in the execution where the actual behavior diverges from the desired
behavior. For example, a variable may get assigned a wrong value, which then
negatively influences the remaining computation. Once a developer identifies
such a divergence, how to fix the code so that it provides the desired
behavior? This paper presents TraceFixer, a technique for predicting how to
edit source code so that it does not diverge from the expected behavior
anymore. The key idea is to train a neural program repair model that not only
learns from source code edits but also exploits excerpts of runtime traces. The
input to the model is a partial execution trace of the incorrect code, which
can be obtained automatically through code instrumentation, and the correct
state that the program should reach at the divergence point, which the user
provides, e.g., in an interactive debugger. Our approach fundamentally differs
from current program repair techniques, which share a similar goal but exploit
neither execution traces nor information about the desired program state. We
evaluate TraceFixer on single-line mistakes in Python code. After training the
model on hundreds of thousands of code edits created by a neural model that
mimics real-world bugs, we find that exploiting execution traces improves the
bug-fixing ability by 13% to 20% (depending on the dataset, within the top-10
predictions) compared to a baseline that learns from source code edits only.
Applying TraceFixer to 20 real-world Python bugs shows that the approach
successfully fixes 10 of them.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 11:41:30 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Bouzenia",
"Islem",
""
],
[
"Ding",
"Yangruibo",
""
],
[
"Pei",
"Kexin",
""
],
[
"Ray",
"Baishakhi",
""
],
[
"Pradel",
"Michael",
""
]
] | TITLE: TraceFixer: Execution Trace-Driven Program Repair
ABSTRACT: When debugging unintended program behavior, developers can often identify the
point in the execution where the actual behavior diverges from the desired
behavior. For example, a variable may get assigned a wrong value, which then
negatively influences the remaining computation. Once a developer identifies
such a divergence, how to fix the code so that it provides the desired
behavior? This paper presents TraceFixer, a technique for predicting how to
edit source code so that it does not diverge from the expected behavior
anymore. The key idea is to train a neural program repair model that not only
learns from source code edits but also exploits excerpts of runtime traces. The
input to the model is a partial execution trace of the incorrect code, which
can be obtained automatically through code instrumentation, and the correct
state that the program should reach at the divergence point, which the user
provides, e.g., in an interactive debugger. Our approach fundamentally differs
from current program repair techniques, which share a similar goal but exploit
neither execution traces nor information about the desired program state. We
evaluate TraceFixer on single-line mistakes in Python code. After training the
model on hundreds of thousands of code edits created by a neural model that
mimics real-world bugs, we find that exploiting execution traces improves the
bug-fixing ability by 13% to 20% (depending on the dataset, within the top-10
predictions) compared to a baseline that learns from source code edits only.
Applying TraceFixer to 20 real-world Python bugs shows that the approach
successfully fixes 10 of them.
| no_new_dataset | 0.707834 |
2304.12762 | Sandeep Chandran | Rajshekar Kalayappan, Sandeep Chandran | Dynamic Ineffectuality-based Clustered Architectures | null | null | null | null | cs.AR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The direction of conditional branches is predicted correctly in modern
processors with great accuracy. We find several instructions in the dynamic
instruction stream that contribute only towards computing the condition of
these instructions. Hence, when the predicted direction of conditional branches
is indeed correct, these instructions become Ineffectual - the functional state
of the program would not be different had these instructions been dropped.
However, the execution of ineffectual instructions cannot be avoided altogether
because it is possible that the prediction of the branch direction is wrong. In
this work, we determine all sources of ineffectuality in an instruction stream
such as conditional branches, predicated instructions, indirect jumps and
dynamically dead instructions. Then, we propose a technique to steer the
ineffectual instructions away from the primary execution cluster so that
effectual instructions can execute uncontended. We find that such
ineffectuality-based clustering of instructions naturally simplifies the design
and avoids several caveats of a clustered architecture. Finally, we propose a
technique to detect instances when instructions were incorrectly marked as
ineffectual, say due to a branch misprediction, and recover the pipeline. The
empirical evaluation of the proposed changes on the SPEC CPU2017 and GAPBS
benchmarks show performance uplifts of up to 4.9% and 10.3% on average
respectively.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 12:26:09 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Kalayappan",
"Rajshekar",
""
],
[
"Chandran",
"Sandeep",
""
]
] | TITLE: Dynamic Ineffectuality-based Clustered Architectures
ABSTRACT: The direction of conditional branches is predicted correctly in modern
processors with great accuracy. We find several instructions in the dynamic
instruction stream that contribute only towards computing the condition of
these instructions. Hence, when the predicted direction of conditional branches
is indeed correct, these instructions become Ineffectual - the functional state
of the program would not be different had these instructions been dropped.
However, the execution of ineffectual instructions cannot be avoided altogether
because it is possible that the prediction of the branch direction is wrong. In
this work, we determine all sources of ineffectuality in an instruction stream
such as conditional branches, predicated instructions, indirect jumps and
dynamically dead instructions. Then, we propose a technique to steer the
ineffectual instructions away from the primary execution cluster so that
effectual instructions can execute uncontended. We find that such
ineffectuality-based clustering of instructions naturally simplifies the design
and avoids several caveats of a clustered architecture. Finally, we propose a
technique to detect instances when instructions were incorrectly marked as
ineffectual, say due to a branch misprediction, and recover the pipeline. The
empirical evaluation of the proposed changes on the SPEC CPU2017 and GAPBS
benchmarks show performance uplifts of up to 4.9% and 10.3% on average
respectively.
| no_new_dataset | 0.708055 |
2304.12769 | Simon Schneider | Simon Schneider and Riccardo Scandariato | Automatic Extraction of Security-Rich Dataflow Diagrams for Microservice
Applications written in Java | null | null | null | null | cs.SE | http://creativecommons.org/licenses/by/4.0/ | Dataflow diagrams (DFDs) are a valuable asset for securing applications, as
they are the starting point for many security assessment techniques. Their
creation, however, is often done manually, which is time-consuming and
introduces problems concerning their correctness. Furthermore, as applications
are continuously extended and modified in CI/CD pipelines, the DFDs need to be
kept in sync, which is also challenging. In this paper, we present a novel,
tool-supported technique to automatically extract DFDs from the implementation
code of microservices. The technique parses source code and configuration files
in search for keywords that are used as evidence for the model extraction. Our
approach uses a novel technique that iteratively detects new keywords, thereby
snowballing through an application's codebase. Coupled with other detection
techniques, it produces a fully-fledged DFD enriched with security-relevant
annotations. The extracted DFDs further provide full traceability between model
items and code snippets. We evaluate our approach and the accompanying
prototype for applications written in Java on a manually curated dataset of 17
open-source applications. In our testing set of applications, we observe an
overall precision of 93% and recall of 85%.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 12:43:22 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Schneider",
"Simon",
""
],
[
"Scandariato",
"Riccardo",
""
]
] | TITLE: Automatic Extraction of Security-Rich Dataflow Diagrams for Microservice
Applications written in Java
ABSTRACT: Dataflow diagrams (DFDs) are a valuable asset for securing applications, as
they are the starting point for many security assessment techniques. Their
creation, however, is often done manually, which is time-consuming and
introduces problems concerning their correctness. Furthermore, as applications
are continuously extended and modified in CI/CD pipelines, the DFDs need to be
kept in sync, which is also challenging. In this paper, we present a novel,
tool-supported technique to automatically extract DFDs from the implementation
code of microservices. The technique parses source code and configuration files
in search for keywords that are used as evidence for the model extraction. Our
approach uses a novel technique that iteratively detects new keywords, thereby
snowballing through an application's codebase. Coupled with other detection
techniques, it produces a fully-fledged DFD enriched with security-relevant
annotations. The extracted DFDs further provide full traceability between model
items and code snippets. We evaluate our approach and the accompanying
prototype for applications written in Java on a manually curated dataset of 17
open-source applications. In our testing set of applications, we observe an
overall precision of 93% and recall of 85%.
| new_dataset | 0.712607 |
2304.12777 | Ziyao Guo | Ziyao Guo, Haonan Yan, Hui Li, Xiaodong Lin | Class Attention Transfer Based Knowledge Distillation | Accepted by CVPR 2023 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Previous knowledge distillation methods have shown their impressive
performance on model compression tasks, however, it is hard to explain how the
knowledge they transferred helps to improve the performance of the student
network. In this work, we focus on proposing a knowledge distillation method
that has both high interpretability and competitive performance. We first
revisit the structure of mainstream CNN models and reveal that possessing the
capacity of identifying class discriminative regions of input is critical for
CNN to perform classification. Furthermore, we demonstrate that this capacity
can be obtained and enhanced by transferring class activation maps. Based on
our findings, we propose class attention transfer based knowledge distillation
(CAT-KD). Different from previous KD methods, we explore and present several
properties of the knowledge transferred by our method, which not only improve
the interpretability of CAT-KD but also contribute to a better understanding of
CNN. While having high interpretability, CAT-KD achieves state-of-the-art
performance on multiple benchmarks. Code is available at:
https://github.com/GzyAftermath/CAT-KD.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 12:56:06 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Guo",
"Ziyao",
""
],
[
"Yan",
"Haonan",
""
],
[
"Li",
"Hui",
""
],
[
"Lin",
"Xiaodong",
""
]
] | TITLE: Class Attention Transfer Based Knowledge Distillation
ABSTRACT: Previous knowledge distillation methods have shown their impressive
performance on model compression tasks, however, it is hard to explain how the
knowledge they transferred helps to improve the performance of the student
network. In this work, we focus on proposing a knowledge distillation method
that has both high interpretability and competitive performance. We first
revisit the structure of mainstream CNN models and reveal that possessing the
capacity of identifying class discriminative regions of input is critical for
CNN to perform classification. Furthermore, we demonstrate that this capacity
can be obtained and enhanced by transferring class activation maps. Based on
our findings, we propose class attention transfer based knowledge distillation
(CAT-KD). Different from previous KD methods, we explore and present several
properties of the knowledge transferred by our method, which not only improve
the interpretability of CAT-KD but also contribute to a better understanding of
CNN. While having high interpretability, CAT-KD achieves state-of-the-art
performance on multiple benchmarks. Code is available at:
https://github.com/GzyAftermath/CAT-KD.
| no_new_dataset | 0.71086 |
2304.12814 | Arthur Jun Zhang | Arthur Jun Zhang | A Novel Dual of Shannon Information and Weighting Scheme | 8 pages, 3 figures, submitted to ACL 2023 | null | null | null | cs.CL cs.IR cs.IT cs.LG math.IT | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Shannon Information theory has achieved great success in not only
communication technology where it was originally developed for but also many
other science and engineering fields such as machine learning and artificial
intelligence. Inspired by the famous weighting scheme TF-IDF, we discovered
that information entropy has a natural dual. We complement the classical
Shannon information theory by proposing a novel quantity, namely troenpy.
Troenpy measures the certainty, commonness and similarity of the underlying
distribution. To demonstrate its usefulness, we propose a troenpy based
weighting scheme for document with class labels, namely positive class
frequency (PCF). On a collection of public datasets we show the PCF based
weighting scheme outperforms the classical TF-IDF and a popular Optimal
Transportation based word moving distance algorithm in a kNN setting. We
further developed a new odds-ratio type feature, namely Expected Class
Information Bias(ECIB), which can be regarded as the expected odds ratio of the
information quantity entropy and troenpy. In the experiments we observe that
including the new ECIB features and simple binary term features in a simple
logistic regression model can further significantly improve the performance.
The simple new weighting scheme and ECIB features are very effective and can be
computed with linear order complexity.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 13:40:50 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Zhang",
"Arthur Jun",
""
]
] | TITLE: A Novel Dual of Shannon Information and Weighting Scheme
ABSTRACT: Shannon Information theory has achieved great success in not only
communication technology where it was originally developed for but also many
other science and engineering fields such as machine learning and artificial
intelligence. Inspired by the famous weighting scheme TF-IDF, we discovered
that information entropy has a natural dual. We complement the classical
Shannon information theory by proposing a novel quantity, namely troenpy.
Troenpy measures the certainty, commonness and similarity of the underlying
distribution. To demonstrate its usefulness, we propose a troenpy based
weighting scheme for document with class labels, namely positive class
frequency (PCF). On a collection of public datasets we show the PCF based
weighting scheme outperforms the classical TF-IDF and a popular Optimal
Transportation based word moving distance algorithm in a kNN setting. We
further developed a new odds-ratio type feature, namely Expected Class
Information Bias(ECIB), which can be regarded as the expected odds ratio of the
information quantity entropy and troenpy. In the experiments we observe that
including the new ECIB features and simple binary term features in a simple
logistic regression model can further significantly improve the performance.
The simple new weighting scheme and ECIB features are very effective and can be
computed with linear order complexity.
| no_new_dataset | 0.710396 |
2304.12828 | Xin Su | Kuo Yang, Zecong Yu, Xin Su, Xiong He, Ning Wang, Qiguang Zheng,
Feidie Yu, Zhuang Liu, Tiancai Wen and Xuezhong Zhou | A optimization framework for herbal prescription planning based on deep
reinforcement learning | 13 pages, 4 figures | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Treatment planning for chronic diseases is a critical task in medical
artificial intelligence, particularly in traditional Chinese medicine (TCM).
However, generating optimized sequential treatment strategies for patients with
chronic diseases in different clinical encounters remains a challenging issue
that requires further exploration. In this study, we proposed a TCM herbal
prescription planning framework based on deep reinforcement learning for
chronic disease treatment (PrescDRL). PrescDRL is a sequential herbal
prescription optimization model that focuses on long-term effectiveness rather
than achieving maximum reward at every step, thereby ensuring better patient
outcomes. We constructed a high-quality benchmark dataset for sequential
diagnosis and treatment of diabetes and evaluated PrescDRL against this
benchmark. Our results showed that PrescDRL achieved a higher curative effect,
with the single-step reward improving by 117% and 153% compared to doctors.
Furthermore, PrescDRL outperformed the benchmark in prescription prediction,
with precision improving by 40.5% and recall improving by 63%. Overall, our
study demonstrates the potential of using artificial intelligence to improve
clinical intelligent diagnosis and treatment in TCM.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 13:55:02 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Yang",
"Kuo",
""
],
[
"Yu",
"Zecong",
""
],
[
"Su",
"Xin",
""
],
[
"He",
"Xiong",
""
],
[
"Wang",
"Ning",
""
],
[
"Zheng",
"Qiguang",
""
],
[
"Yu",
"Feidie",
""
],
[
"Liu",
"Zhuang",
""
],
[
"Wen",
"Tiancai",
""
],
[
"Zhou",
"Xuezhong",
""
]
] | TITLE: A optimization framework for herbal prescription planning based on deep
reinforcement learning
ABSTRACT: Treatment planning for chronic diseases is a critical task in medical
artificial intelligence, particularly in traditional Chinese medicine (TCM).
However, generating optimized sequential treatment strategies for patients with
chronic diseases in different clinical encounters remains a challenging issue
that requires further exploration. In this study, we proposed a TCM herbal
prescription planning framework based on deep reinforcement learning for
chronic disease treatment (PrescDRL). PrescDRL is a sequential herbal
prescription optimization model that focuses on long-term effectiveness rather
than achieving maximum reward at every step, thereby ensuring better patient
outcomes. We constructed a high-quality benchmark dataset for sequential
diagnosis and treatment of diabetes and evaluated PrescDRL against this
benchmark. Our results showed that PrescDRL achieved a higher curative effect,
with the single-step reward improving by 117% and 153% compared to doctors.
Furthermore, PrescDRL outperformed the benchmark in prescription prediction,
with precision improving by 40.5% and recall improving by 63%. Overall, our
study demonstrates the potential of using artificial intelligence to improve
clinical intelligent diagnosis and treatment in TCM.
| new_dataset | 0.714392 |
2304.12829 | Jos\'e Cano | Ferheen Ayaz, Idris Zakariyya, Jos\'e Cano, Sye Loong Keoh, Jeremy
Singer, Danilo Pau, Mounia Kharbouche-Harrari | Improving Robustness Against Adversarial Attacks with Deeply Quantized
Neural Networks | Accepted at IJCNN 2023. 8 pages, 5 figures | null | null | null | cs.LG cs.CR cs.PF | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Reducing the memory footprint of Machine Learning (ML) models, particularly
Deep Neural Networks (DNNs), is essential to enable their deployment into
resource-constrained tiny devices. However, a disadvantage of DNN models is
their vulnerability to adversarial attacks, as they can be fooled by adding
slight perturbations to the inputs. Therefore, the challenge is how to create
accurate, robust, and tiny DNN models deployable on resource-constrained
embedded devices. This paper reports the results of devising a tiny DNN model,
robust to adversarial black and white box attacks, trained with an automatic
quantizationaware training framework, i.e. QKeras, with deep quantization loss
accounted in the learning loop, thereby making the designed DNNs more accurate
for deployment on tiny devices. We investigated how QKeras and an adversarial
robustness technique, Jacobian Regularization (JR), can provide a
co-optimization strategy by exploiting the DNN topology and the per layer JR
approach to produce robust yet tiny deeply quantized DNN models. As a result, a
new DNN model implementing this cooptimization strategy was conceived,
developed and tested on three datasets containing both images and audio inputs,
as well as compared its performance with existing benchmarks against various
white-box and black-box attacks. Experimental results demonstrated that on
average our proposed DNN model resulted in 8.3% and 79.5% higher accuracy than
MLCommons/Tiny benchmarks in the presence of white-box and black-box attacks on
the CIFAR-10 image dataset and a subset of the Google Speech Commands audio
dataset respectively. It was also 6.5% more accurate for black-box attacks on
the SVHN image dataset.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 13:56:35 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Ayaz",
"Ferheen",
""
],
[
"Zakariyya",
"Idris",
""
],
[
"Cano",
"José",
""
],
[
"Keoh",
"Sye Loong",
""
],
[
"Singer",
"Jeremy",
""
],
[
"Pau",
"Danilo",
""
],
[
"Kharbouche-Harrari",
"Mounia",
""
]
] | TITLE: Improving Robustness Against Adversarial Attacks with Deeply Quantized
Neural Networks
ABSTRACT: Reducing the memory footprint of Machine Learning (ML) models, particularly
Deep Neural Networks (DNNs), is essential to enable their deployment into
resource-constrained tiny devices. However, a disadvantage of DNN models is
their vulnerability to adversarial attacks, as they can be fooled by adding
slight perturbations to the inputs. Therefore, the challenge is how to create
accurate, robust, and tiny DNN models deployable on resource-constrained
embedded devices. This paper reports the results of devising a tiny DNN model,
robust to adversarial black and white box attacks, trained with an automatic
quantizationaware training framework, i.e. QKeras, with deep quantization loss
accounted in the learning loop, thereby making the designed DNNs more accurate
for deployment on tiny devices. We investigated how QKeras and an adversarial
robustness technique, Jacobian Regularization (JR), can provide a
co-optimization strategy by exploiting the DNN topology and the per layer JR
approach to produce robust yet tiny deeply quantized DNN models. As a result, a
new DNN model implementing this cooptimization strategy was conceived,
developed and tested on three datasets containing both images and audio inputs,
as well as compared its performance with existing benchmarks against various
white-box and black-box attacks. Experimental results demonstrated that on
average our proposed DNN model resulted in 8.3% and 79.5% higher accuracy than
MLCommons/Tiny benchmarks in the presence of white-box and black-box attacks on
the CIFAR-10 image dataset and a subset of the Google Speech Commands audio
dataset respectively. It was also 6.5% more accurate for black-box attacks on
the SVHN image dataset.
| no_new_dataset | 0.710434 |
2304.12836 | Ji-Ung Lee | Jan-Christoph Klie, Ji-Ung Lee, Kevin Stowe, G\"ozde G\"ul \c{S}ahin,
Nafise Sadat Moosavi, Luke Bates, Dominic Petrak, Richard Eckart de Castilho,
Iryna Gurevych | Lessons Learned from a Citizen Science Project for Natural Language
Processing | Accepted to EACL 2023. Code will be published on github:
https://github.com/UKPLab/eacl2023-citizen-science-lessons-learned | null | null | null | cs.CL | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Many Natural Language Processing (NLP) systems use annotated corpora for
training and evaluation. However, labeled data is often costly to obtain and
scaling annotation projects is difficult, which is why annotation tasks are
often outsourced to paid crowdworkers. Citizen Science is an alternative to
crowdsourcing that is relatively unexplored in the context of NLP. To
investigate whether and how well Citizen Science can be applied in this
setting, we conduct an exploratory study into engaging different groups of
volunteers in Citizen Science for NLP by re-annotating parts of a pre-existing
crowdsourced dataset. Our results show that this can yield high-quality
annotations and attract motivated volunteers, but also requires considering
factors such as scalability, participation over time, and legal and ethical
issues. We summarize lessons learned in the form of guidelines and provide our
code and data to aid future work on Citizen Science.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 14:08:53 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Klie",
"Jan-Christoph",
""
],
[
"Lee",
"Ji-Ung",
""
],
[
"Stowe",
"Kevin",
""
],
[
"Şahin",
"Gözde Gül",
""
],
[
"Moosavi",
"Nafise Sadat",
""
],
[
"Bates",
"Luke",
""
],
[
"Petrak",
"Dominic",
""
],
[
"de Castilho",
"Richard Eckart",
""
],
[
"Gurevych",
"Iryna",
""
]
] | TITLE: Lessons Learned from a Citizen Science Project for Natural Language
Processing
ABSTRACT: Many Natural Language Processing (NLP) systems use annotated corpora for
training and evaluation. However, labeled data is often costly to obtain and
scaling annotation projects is difficult, which is why annotation tasks are
often outsourced to paid crowdworkers. Citizen Science is an alternative to
crowdsourcing that is relatively unexplored in the context of NLP. To
investigate whether and how well Citizen Science can be applied in this
setting, we conduct an exploratory study into engaging different groups of
volunteers in Citizen Science for NLP by re-annotating parts of a pre-existing
crowdsourced dataset. Our results show that this can yield high-quality
annotations and attract motivated volunteers, but also requires considering
factors such as scalability, participation over time, and legal and ethical
issues. We summarize lessons learned in the form of guidelines and provide our
code and data to aid future work on Citizen Science.
| no_new_dataset | 0.712282 |
2304.12849 | Kyuhong Shim | Kyuhong Shim, Jiyoung Kim, Gusang Lee, Byonghyo Shim | Depth-Relative Self Attention for Monocular Depth Estimation | Accepted for IJCAI 2023 | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Monocular depth estimation is very challenging because clues to the exact
depth are incomplete in a single RGB image. To overcome the limitation, deep
neural networks rely on various visual hints such as size, shade, and texture
extracted from RGB information. However, we observe that if such hints are
overly exploited, the network can be biased on RGB information without
considering the comprehensive view. We propose a novel depth estimation model
named RElative Depth Transformer (RED-T) that uses relative depth as guidance
in self-attention. Specifically, the model assigns high attention weights to
pixels of close depth and low attention weights to pixels of distant depth. As
a result, the features of similar depth can become more likely to each other
and thus less prone to misused visual hints. We show that the proposed model
achieves competitive results in monocular depth estimation benchmarks and is
less biased to RGB information. In addition, we propose a novel monocular depth
estimation benchmark that limits the observable depth range during training in
order to evaluate the robustness of the model for unseen depths.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 14:20:31 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Shim",
"Kyuhong",
""
],
[
"Kim",
"Jiyoung",
""
],
[
"Lee",
"Gusang",
""
],
[
"Shim",
"Byonghyo",
""
]
] | TITLE: Depth-Relative Self Attention for Monocular Depth Estimation
ABSTRACT: Monocular depth estimation is very challenging because clues to the exact
depth are incomplete in a single RGB image. To overcome the limitation, deep
neural networks rely on various visual hints such as size, shade, and texture
extracted from RGB information. However, we observe that if such hints are
overly exploited, the network can be biased on RGB information without
considering the comprehensive view. We propose a novel depth estimation model
named RElative Depth Transformer (RED-T) that uses relative depth as guidance
in self-attention. Specifically, the model assigns high attention weights to
pixels of close depth and low attention weights to pixels of distant depth. As
a result, the features of similar depth can become more likely to each other
and thus less prone to misused visual hints. We show that the proposed model
achieves competitive results in monocular depth estimation benchmarks and is
less biased to RGB information. In addition, we propose a novel monocular depth
estimation benchmark that limits the observable depth range during training in
order to evaluate the robustness of the model for unseen depths.
| no_new_dataset | 0.709239 |
2304.12856 | Syed Saud Naqvi | Tariq M. Khan, Syed S. Naqvi, Antonio Robles-Kelly, Imran Razzak | Retinal Vessel Segmentation via a Multi-resolution Contextual Network
and Adversarial Learning | null | null | null | null | eess.IV cs.CV cs.LG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Timely and affordable computer-aided diagnosis of retinal diseases is pivotal
in precluding blindness. Accurate retinal vessel segmentation plays an
important role in disease progression and diagnosis of such vision-threatening
diseases. To this end, we propose a Multi-resolution Contextual Network
(MRC-Net) that addresses these issues by extracting multi-scale features to
learn contextual dependencies between semantically different features and using
bi-directional recurrent learning to model former-latter and latter-former
dependencies. Another key idea is training in adversarial settings for
foreground segmentation improvement through optimization of the region-based
scores. This novel strategy boosts the performance of the segmentation network
in terms of the Dice score (and correspondingly Jaccard index) while keeping
the number of trainable parameters comparatively low. We have evaluated our
method on three benchmark datasets, including DRIVE, STARE, and CHASE,
demonstrating its superior performance as compared with competitive approaches
elsewhere in the literature.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 14:27:34 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Khan",
"Tariq M.",
""
],
[
"Naqvi",
"Syed S.",
""
],
[
"Robles-Kelly",
"Antonio",
""
],
[
"Razzak",
"Imran",
""
]
] | TITLE: Retinal Vessel Segmentation via a Multi-resolution Contextual Network
and Adversarial Learning
ABSTRACT: Timely and affordable computer-aided diagnosis of retinal diseases is pivotal
in precluding blindness. Accurate retinal vessel segmentation plays an
important role in disease progression and diagnosis of such vision-threatening
diseases. To this end, we propose a Multi-resolution Contextual Network
(MRC-Net) that addresses these issues by extracting multi-scale features to
learn contextual dependencies between semantically different features and using
bi-directional recurrent learning to model former-latter and latter-former
dependencies. Another key idea is training in adversarial settings for
foreground segmentation improvement through optimization of the region-based
scores. This novel strategy boosts the performance of the segmentation network
in terms of the Dice score (and correspondingly Jaccard index) while keeping
the number of trainable parameters comparatively low. We have evaluated our
method on three benchmark datasets, including DRIVE, STARE, and CHASE,
demonstrating its superior performance as compared with competitive approaches
elsewhere in the literature.
| no_new_dataset | 0.710986 |
2304.12889 | Aditya Kalapaaking | Aditya Pribadi Kalapaaking, Ibrahim Khalil, Mohammad Saidur Rahman,
Mohammed Atiquzzaman, Xun Yi, and Mahathir Almashor | Blockchain-based Federated Learning with Secure Aggregation in Trusted
Execution Environment for Internet-of-Things | null | IEEE Transactions on Industrial Informatics, vol. 19, no. 2, pp.
1703-1714, Feb. 2023 | 10.1109/TII.2022.3170348 | null | cs.CR cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper proposes a blockchain-based Federated Learning (FL) framework with
Intel Software Guard Extension (SGX)-based Trusted Execution Environment (TEE)
to securely aggregate local models in Industrial Internet-of-Things (IIoTs). In
FL, local models can be tampered with by attackers. Hence, a global model
generated from the tampered local models can be erroneous. Therefore, the
proposed framework leverages a blockchain network for secure model aggregation.
Each blockchain node hosts an SGX-enabled processor that securely performs the
FL-based aggregation tasks to generate a global model. Blockchain nodes can
verify the authenticity of the aggregated model, run a blockchain consensus
mechanism to ensure the integrity of the model, and add it to the distributed
ledger for tamper-proof storage. Each cluster can obtain the aggregated model
from the blockchain and verify its integrity before using it. We conducted
several experiments with different CNN models and datasets to evaluate the
performance of the proposed framework.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 15:00:39 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Kalapaaking",
"Aditya Pribadi",
""
],
[
"Khalil",
"Ibrahim",
""
],
[
"Rahman",
"Mohammad Saidur",
""
],
[
"Atiquzzaman",
"Mohammed",
""
],
[
"Yi",
"Xun",
""
],
[
"Almashor",
"Mahathir",
""
]
] | TITLE: Blockchain-based Federated Learning with Secure Aggregation in Trusted
Execution Environment for Internet-of-Things
ABSTRACT: This paper proposes a blockchain-based Federated Learning (FL) framework with
Intel Software Guard Extension (SGX)-based Trusted Execution Environment (TEE)
to securely aggregate local models in Industrial Internet-of-Things (IIoTs). In
FL, local models can be tampered with by attackers. Hence, a global model
generated from the tampered local models can be erroneous. Therefore, the
proposed framework leverages a blockchain network for secure model aggregation.
Each blockchain node hosts an SGX-enabled processor that securely performs the
FL-based aggregation tasks to generate a global model. Blockchain nodes can
verify the authenticity of the aggregated model, run a blockchain consensus
mechanism to ensure the integrity of the model, and add it to the distributed
ledger for tamper-proof storage. Each cluster can obtain the aggregated model
from the blockchain and verify its integrity before using it. We conducted
several experiments with different CNN models and datasets to evaluate the
performance of the proposed framework.
| no_new_dataset | 0.709453 |
2304.12891 | Jussi Leinonen | Jussi Leinonen, Ulrich Hamann, Daniele Nerini, Urs Germann, Gabriele
Franch | Latent diffusion models for generative precipitation nowcasting with
accurate uncertainty quantification | 18 pages, 6 figures. Submitted for publication | null | null | null | physics.ao-ph cs.LG eess.IV | http://creativecommons.org/licenses/by-sa/4.0/ | Diffusion models have been widely adopted in image generation, producing
higher-quality and more diverse samples than generative adversarial networks
(GANs). We introduce a latent diffusion model (LDM) for precipitation
nowcasting - short-term forecasting based on the latest observational data. The
LDM is more stable and requires less computation to train than GANs, albeit
with more computationally expensive generation. We benchmark it against the
GAN-based Deep Generative Models of Rainfall (DGMR) and a statistical model,
PySTEPS. The LDM produces more accurate precipitation predictions, while the
comparisons are more mixed when predicting whether the precipitation exceeds
predefined thresholds. The clearest advantage of the LDM is that it generates
more diverse predictions than DGMR or PySTEPS. Rank distribution tests indicate
that the distribution of samples from the LDM accurately reflects the
uncertainty of the predictions. Thus, LDMs are promising for any applications
where uncertainty quantification is important, such as weather and climate.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 15:03:15 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Leinonen",
"Jussi",
""
],
[
"Hamann",
"Ulrich",
""
],
[
"Nerini",
"Daniele",
""
],
[
"Germann",
"Urs",
""
],
[
"Franch",
"Gabriele",
""
]
] | TITLE: Latent diffusion models for generative precipitation nowcasting with
accurate uncertainty quantification
ABSTRACT: Diffusion models have been widely adopted in image generation, producing
higher-quality and more diverse samples than generative adversarial networks
(GANs). We introduce a latent diffusion model (LDM) for precipitation
nowcasting - short-term forecasting based on the latest observational data. The
LDM is more stable and requires less computation to train than GANs, albeit
with more computationally expensive generation. We benchmark it against the
GAN-based Deep Generative Models of Rainfall (DGMR) and a statistical model,
PySTEPS. The LDM produces more accurate precipitation predictions, while the
comparisons are more mixed when predicting whether the precipitation exceeds
predefined thresholds. The clearest advantage of the LDM is that it generates
more diverse predictions than DGMR or PySTEPS. Rank distribution tests indicate
that the distribution of samples from the LDM accurately reflects the
uncertainty of the predictions. Thus, LDMs are promising for any applications
where uncertainty quantification is important, such as weather and climate.
| no_new_dataset | 0.710038 |
2304.12904 | Carlos Lassance | Carlos Lassance, St\'ephane Clinchant | The tale of two MS MARCO -- and their unfair comparisons | Short paper accepted at SIGIR 2023 | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The MS MARCO-passage dataset has been the main large-scale dataset open to
the IR community and it has fostered successfully the development of novel
neural retrieval models over the years. But, it turns out that two different
corpora of MS MARCO are used in the literature, the official one and a second
one where passages were augmented with titles, mostly due to the introduction
of the Tevatron code base. However, the addition of titles actually leaks
relevance information, while breaking the original guidelines of the MS
MARCO-passage dataset. In this work, we investigate the differences between the
two corpora and demonstrate empirically that they make a significant difference
when evaluating a new method. In other words, we show that if a paper does not
properly report which version is used, reproducing fairly its results is
basically impossible. Furthermore, given the current status of reviewing, where
monitoring state-of-the-art results is of great importance, having two
different versions of a dataset is a large problem. This is why this paper aims
to report the importance of this issue so that researchers can be made aware of
this problem and appropriately report their results.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 15:15:49 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Lassance",
"Carlos",
""
],
[
"Clinchant",
"Stéphane",
""
]
] | TITLE: The tale of two MS MARCO -- and their unfair comparisons
ABSTRACT: The MS MARCO-passage dataset has been the main large-scale dataset open to
the IR community and it has fostered successfully the development of novel
neural retrieval models over the years. But, it turns out that two different
corpora of MS MARCO are used in the literature, the official one and a second
one where passages were augmented with titles, mostly due to the introduction
of the Tevatron code base. However, the addition of titles actually leaks
relevance information, while breaking the original guidelines of the MS
MARCO-passage dataset. In this work, we investigate the differences between the
two corpora and demonstrate empirically that they make a significant difference
when evaluating a new method. In other words, we show that if a paper does not
properly report which version is used, reproducing fairly its results is
basically impossible. Furthermore, given the current status of reviewing, where
monitoring state-of-the-art results is of great importance, having two
different versions of a dataset is a large problem. This is why this paper aims
to report the importance of this issue so that researchers can be made aware of
this problem and appropriately report their results.
| no_new_dataset | 0.703511 |
2304.12918 | Fazl Barez | Alex Foote, Neel Nanda, Esben Kran, Ionnis Konstas, Fazl Barez | N2G: A Scalable Approach for Quantifying Interpretable Neuron
Representations in Large Language Models | To be published at ICLR 2023 Workshop on Trustworthy and Reliable
Large-Scale Machine Learning Models | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Understanding the function of individual neurons within language models is
essential for mechanistic interpretability research. We propose $\textbf{Neuron
to Graph (N2G)}$, a tool which takes a neuron and its dataset examples, and
automatically distills the neuron's behaviour on those examples to an
interpretable graph. This presents a less labour intensive approach to
interpreting neurons than current manual methods, that will better scale these
methods to Large Language Models (LLMs). We use truncation and saliency methods
to only present the important tokens, and augment the dataset examples with
more diverse samples to better capture the extent of neuron behaviour. These
graphs can be visualised to aid manual interpretation by researchers, but can
also output token activations on text to compare to the neuron's ground truth
activations for automatic validation. N2G represents a step towards scalable
interpretability methods by allowing us to convert neurons in an LLM to
interpretable representations of measurable quality.
| [
{
"version": "v1",
"created": "Sat, 22 Apr 2023 19:06:13 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Foote",
"Alex",
""
],
[
"Nanda",
"Neel",
""
],
[
"Kran",
"Esben",
""
],
[
"Konstas",
"Ionnis",
""
],
[
"Barez",
"Fazl",
""
]
] | TITLE: N2G: A Scalable Approach for Quantifying Interpretable Neuron
Representations in Large Language Models
ABSTRACT: Understanding the function of individual neurons within language models is
essential for mechanistic interpretability research. We propose $\textbf{Neuron
to Graph (N2G)}$, a tool which takes a neuron and its dataset examples, and
automatically distills the neuron's behaviour on those examples to an
interpretable graph. This presents a less labour intensive approach to
interpreting neurons than current manual methods, that will better scale these
methods to Large Language Models (LLMs). We use truncation and saliency methods
to only present the important tokens, and augment the dataset examples with
more diverse samples to better capture the extent of neuron behaviour. These
graphs can be visualised to aid manual interpretation by researchers, but can
also output token activations on text to compare to the neuron's ground truth
activations for automatic validation. N2G represents a step towards scalable
interpretability methods by allowing us to convert neurons in an LLM to
interpretable representations of measurable quality.
| no_new_dataset | 0.710666 |
2304.12943 | Victor Guyomard | Victor Guyomard, Fran\c{c}oise Fessant, Thomas Guyet, Tassadit Bouadi,
Alexandre Termier | Generating robust counterfactual explanations | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Counterfactual explanations have become a mainstay of the XAI field. This
particularly intuitive statement allows the user to understand what small but
necessary changes would have to be made to a given situation in order to change
a model prediction. The quality of a counterfactual depends on several
criteria: realism, actionability, validity, robustness, etc. In this paper, we
are interested in the notion of robustness of a counterfactual. More precisely,
we focus on robustness to counterfactual input changes. This form of robustness
is particularly challenging as it involves a trade-off between the robustness
of the counterfactual and the proximity with the example to explain. We propose
a new framework, CROCO, that generates robust counterfactuals while managing
effectively this trade-off, and guarantees the user a minimal robustness. An
empirical evaluation on tabular datasets confirms the relevance and
effectiveness of our approach.
| [
{
"version": "v1",
"created": "Mon, 24 Apr 2023 09:00:31 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Guyomard",
"Victor",
""
],
[
"Fessant",
"Françoise",
""
],
[
"Guyet",
"Thomas",
""
],
[
"Bouadi",
"Tassadit",
""
],
[
"Termier",
"Alexandre",
""
]
] | TITLE: Generating robust counterfactual explanations
ABSTRACT: Counterfactual explanations have become a mainstay of the XAI field. This
particularly intuitive statement allows the user to understand what small but
necessary changes would have to be made to a given situation in order to change
a model prediction. The quality of a counterfactual depends on several
criteria: realism, actionability, validity, robustness, etc. In this paper, we
are interested in the notion of robustness of a counterfactual. More precisely,
we focus on robustness to counterfactual input changes. This form of robustness
is particularly challenging as it involves a trade-off between the robustness
of the counterfactual and the proximity with the example to explain. We propose
a new framework, CROCO, that generates robust counterfactuals while managing
effectively this trade-off, and guarantees the user a minimal robustness. An
empirical evaluation on tabular datasets confirms the relevance and
effectiveness of our approach.
| no_new_dataset | 0.711425 |
2304.12982 | James Gung | James Gung, Raphael Shu, Emily Moeng, Wesley Rose, Salvatore Romeo,
Yassine Benajiba, Arshit Gupta, Saab Mansour and Yi Zhang | Intent Induction from Conversations for Task-Oriented Dialogue Track at
DSTC 11 | 18 pages, 1 figure. Accepted at the DSTC 11 Workshop to be located at
SIGDIAL 2023 | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With increasing demand for and adoption of virtual assistants, recent work
has investigated ways to accelerate bot schema design through the automatic
induction of intents or the induction of slots and dialogue states. However, a
lack of dedicated benchmarks and standardized evaluation has made progress
difficult to track and comparisons between systems difficult to make. This
challenge track, held as part of the Eleventh Dialog Systems Technology
Challenge, introduces a benchmark that aims to evaluate methods for the
automatic induction of customer intents in a realistic setting of customer
service interactions between human agents and customers. We propose two
subtasks for progressively tackling the automatic induction of intents and
corresponding evaluation methodologies. We then present three datasets suitable
for evaluating the tasks and propose simple baselines. Finally, we summarize
the submissions and results of the challenge track, for which we received
submissions from 34 teams.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 16:45:50 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Gung",
"James",
""
],
[
"Shu",
"Raphael",
""
],
[
"Moeng",
"Emily",
""
],
[
"Rose",
"Wesley",
""
],
[
"Romeo",
"Salvatore",
""
],
[
"Benajiba",
"Yassine",
""
],
[
"Gupta",
"Arshit",
""
],
[
"Mansour",
"Saab",
""
],
[
"Zhang",
"Yi",
""
]
] | TITLE: Intent Induction from Conversations for Task-Oriented Dialogue Track at
DSTC 11
ABSTRACT: With increasing demand for and adoption of virtual assistants, recent work
has investigated ways to accelerate bot schema design through the automatic
induction of intents or the induction of slots and dialogue states. However, a
lack of dedicated benchmarks and standardized evaluation has made progress
difficult to track and comparisons between systems difficult to make. This
challenge track, held as part of the Eleventh Dialog Systems Technology
Challenge, introduces a benchmark that aims to evaluate methods for the
automatic induction of customer intents in a realistic setting of customer
service interactions between human agents and customers. We propose two
subtasks for progressively tackling the automatic induction of intents and
corresponding evaluation methodologies. We then present three datasets suitable
for evaluating the tasks and propose simple baselines. Finally, we summarize
the submissions and results of the challenge track, for which we received
submissions from 34 teams.
| new_dataset | 0.711067 |
2304.12987 | Bill Cassidy | Imran Chowdhury Dipto, Bill Cassidy, Connah Kendrick, Neil D. Reeves,
Joseph M. Pappachan, Vishnu Chandrabalan, Moi Hoon Yap | Quantifying the Effect of Image Similarity on Diabetic Foot Ulcer
Classification | null | null | 10.1007/978-3-031-26354-5_1 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This research conducts an investigation on the effect of visually similar
images within a publicly available diabetic foot ulcer dataset when training
deep learning classification networks. The presence of binary-identical
duplicate images in datasets used to train deep learning algorithms is a well
known issue that can introduce unwanted bias which can degrade network
performance. However, the effect of visually similar non-identical images is an
under-researched topic, and has so far not been investigated in any diabetic
foot ulcer studies. We use an open-source fuzzy algorithm to identify groups of
increasingly similar images in the Diabetic Foot Ulcers Challenge 2021
(DFUC2021) training dataset. Based on each similarity threshold, we create new
training sets that we use to train a range of deep learning multi-class
classifiers. We then evaluate the performance of the best performing model on
the DFUC2021 test set. Our findings show that the model trained on the training
set with the 80\% similarity threshold images removed achieved the best
performance using the InceptionResNetV2 network. This model showed improvements
in F1-score, precision, and recall of 0.023, 0.029, and 0.013, respectively.
These results indicate that highly similar images can contribute towards the
presence of performance degrading bias within the Diabetic Foot Ulcers
Challenge 2021 dataset, and that the removal of images that are 80\% similar
from the training set can help to boost classification performance.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 16:54:27 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Dipto",
"Imran Chowdhury",
""
],
[
"Cassidy",
"Bill",
""
],
[
"Kendrick",
"Connah",
""
],
[
"Reeves",
"Neil D.",
""
],
[
"Pappachan",
"Joseph M.",
""
],
[
"Chandrabalan",
"Vishnu",
""
],
[
"Yap",
"Moi Hoon",
""
]
] | TITLE: Quantifying the Effect of Image Similarity on Diabetic Foot Ulcer
Classification
ABSTRACT: This research conducts an investigation on the effect of visually similar
images within a publicly available diabetic foot ulcer dataset when training
deep learning classification networks. The presence of binary-identical
duplicate images in datasets used to train deep learning algorithms is a well
known issue that can introduce unwanted bias which can degrade network
performance. However, the effect of visually similar non-identical images is an
under-researched topic, and has so far not been investigated in any diabetic
foot ulcer studies. We use an open-source fuzzy algorithm to identify groups of
increasingly similar images in the Diabetic Foot Ulcers Challenge 2021
(DFUC2021) training dataset. Based on each similarity threshold, we create new
training sets that we use to train a range of deep learning multi-class
classifiers. We then evaluate the performance of the best performing model on
the DFUC2021 test set. Our findings show that the model trained on the training
set with the 80\% similarity threshold images removed achieved the best
performance using the InceptionResNetV2 network. This model showed improvements
in F1-score, precision, and recall of 0.023, 0.029, and 0.013, respectively.
These results indicate that highly similar images can contribute towards the
presence of performance degrading bias within the Diabetic Foot Ulcers
Challenge 2021 dataset, and that the removal of images that are 80\% similar
from the training set can help to boost classification performance.
| no_new_dataset | 0.706368 |
2304.12998 | Rui Hao | Rui Hao, Linmei Hu, Weijian Qi, Qingliu Wu, Yirui Zhang, Liqiang Nie | ChatLLM Network: More brains, More intelligence | null | null | null | null | cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Dialogue-based language models mark a huge milestone in the field of
artificial intelligence, by their impressive ability to interact with users, as
well as a series of challenging tasks prompted by customized instructions.
However, the prevalent large-scale dialogue-based language models like ChatGPT
still have room for improvement, such as unstable responses to questions and
the inability to think cooperatively like humans. Considering the ability of
dialogue-based language models in conversation and their inherent randomness in
thinking, we propose ChatLLM network that allows multiple dialogue-based
language models to interact, provide feedback, and think together. We design
the network of ChatLLMs based on ChatGPT. Specifically, individual instances of
ChatGPT may possess distinct perspectives towards the same problem, and by
consolidating these diverse viewpoints via a separate ChatGPT, the ChatLLM
network system can conduct decision-making more objectively and
comprehensively. In addition, a language-based feedback mechanism comparable to
backpropagation is devised to update the ChatGPTs within the network.
Experiments on two datasets demonstrate that our network attains significant
improvements in problem-solving, leading to observable progress amongst each
member.
| [
{
"version": "v1",
"created": "Mon, 24 Apr 2023 08:29:14 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Hao",
"Rui",
""
],
[
"Hu",
"Linmei",
""
],
[
"Qi",
"Weijian",
""
],
[
"Wu",
"Qingliu",
""
],
[
"Zhang",
"Yirui",
""
],
[
"Nie",
"Liqiang",
""
]
] | TITLE: ChatLLM Network: More brains, More intelligence
ABSTRACT: Dialogue-based language models mark a huge milestone in the field of
artificial intelligence, by their impressive ability to interact with users, as
well as a series of challenging tasks prompted by customized instructions.
However, the prevalent large-scale dialogue-based language models like ChatGPT
still have room for improvement, such as unstable responses to questions and
the inability to think cooperatively like humans. Considering the ability of
dialogue-based language models in conversation and their inherent randomness in
thinking, we propose ChatLLM network that allows multiple dialogue-based
language models to interact, provide feedback, and think together. We design
the network of ChatLLMs based on ChatGPT. Specifically, individual instances of
ChatGPT may possess distinct perspectives towards the same problem, and by
consolidating these diverse viewpoints via a separate ChatGPT, the ChatLLM
network system can conduct decision-making more objectively and
comprehensively. In addition, a language-based feedback mechanism comparable to
backpropagation is devised to update the ChatGPTs within the network.
Experiments on two datasets demonstrate that our network attains significant
improvements in problem-solving, leading to observable progress amongst each
member.
| no_new_dataset | 0.707209 |
2304.13001 | Andrea Dittadi | Andrea Dittadi | On the Generalization of Learned Structured Representations | Minor revision of my PhD thesis: fixed hyperlinks, fixed minor typos,
and compressed images. The original can be found at
https://orbit.dtu.dk/en/publications/on-the-generalization-of-learned-structured-representations | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Despite tremendous progress over the past decade, deep learning methods
generally fall short of human-level systematic generalization. It has been
argued that explicitly capturing the underlying structure of data should allow
connectionist systems to generalize in a more predictable and systematic
manner. Indeed, evidence in humans suggests that interpreting the world in
terms of symbol-like compositional entities may be crucial for intelligent
behavior and high-level reasoning. Another common limitation of deep learning
systems is that they require large amounts of training data, which can be
expensive to obtain. In representation learning, large datasets are leveraged
to learn generic data representations that may be useful for efficient learning
of arbitrary downstream tasks.
This thesis is about structured representation learning. We study methods
that learn, with little or no supervision, representations of unstructured data
that capture its hidden structure. In the first part of the thesis, we focus on
representations that disentangle the explanatory factors of variation of the
data. We scale up disentangled representation learning to a novel robotic
dataset, and perform a systematic large-scale study on the role of pretrained
representations for out-of-distribution generalization in downstream robotic
tasks. The second part of this thesis focuses on object-centric
representations, which capture the compositional structure of the input in
terms of symbol-like entities, such as objects in visual scenes. Object-centric
learning methods learn to form meaningful entities from unstructured input,
enabling symbolic information processing on a connectionist substrate. In this
study, we train a selection of methods on several common datasets, and
investigate their usefulness for downstream tasks and their ability to
generalize out of distribution.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 17:14:36 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Dittadi",
"Andrea",
""
]
] | TITLE: On the Generalization of Learned Structured Representations
ABSTRACT: Despite tremendous progress over the past decade, deep learning methods
generally fall short of human-level systematic generalization. It has been
argued that explicitly capturing the underlying structure of data should allow
connectionist systems to generalize in a more predictable and systematic
manner. Indeed, evidence in humans suggests that interpreting the world in
terms of symbol-like compositional entities may be crucial for intelligent
behavior and high-level reasoning. Another common limitation of deep learning
systems is that they require large amounts of training data, which can be
expensive to obtain. In representation learning, large datasets are leveraged
to learn generic data representations that may be useful for efficient learning
of arbitrary downstream tasks.
This thesis is about structured representation learning. We study methods
that learn, with little or no supervision, representations of unstructured data
that capture its hidden structure. In the first part of the thesis, we focus on
representations that disentangle the explanatory factors of variation of the
data. We scale up disentangled representation learning to a novel robotic
dataset, and perform a systematic large-scale study on the role of pretrained
representations for out-of-distribution generalization in downstream robotic
tasks. The second part of this thesis focuses on object-centric
representations, which capture the compositional structure of the input in
terms of symbol-like entities, such as objects in visual scenes. Object-centric
learning methods learn to form meaningful entities from unstructured input,
enabling symbolic information processing on a connectionist substrate. In this
study, we train a selection of methods on several common datasets, and
investigate their usefulness for downstream tasks and their ability to
generalize out of distribution.
| no_new_dataset | 0.707209 |
2304.13027 | Shilong Liu | Tianhe Ren, Jianwei Yang, Shilong Liu, Ailing Zeng, Feng Li, Hao
Zhang, Hongyang Li, Zhaoyang Zeng, Lei Zhang | A Strong and Reproducible Object Detector with Only Public Datasets | 64.8 AP on COCO test-dev | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work presents Focal-Stable-DINO, a strong and reproducible object
detection model which achieves 64.6 AP on COCO val2017 and 64.8 AP on COCO
test-dev using only 700M parameters without any test time augmentation. It
explores the combination of the powerful FocalNet-Huge backbone with the
effective Stable-DINO detector. Different from existing SOTA models that
utilize an extensive number of parameters and complex training techniques on
large-scale private data or merged data, our model is exclusively trained on
the publicly available dataset Objects365, which ensures the reproducibility of
our approach.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 17:57:44 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Ren",
"Tianhe",
""
],
[
"Yang",
"Jianwei",
""
],
[
"Liu",
"Shilong",
""
],
[
"Zeng",
"Ailing",
""
],
[
"Li",
"Feng",
""
],
[
"Zhang",
"Hao",
""
],
[
"Li",
"Hongyang",
""
],
[
"Zeng",
"Zhaoyang",
""
],
[
"Zhang",
"Lei",
""
]
] | TITLE: A Strong and Reproducible Object Detector with Only Public Datasets
ABSTRACT: This work presents Focal-Stable-DINO, a strong and reproducible object
detection model which achieves 64.6 AP on COCO val2017 and 64.8 AP on COCO
test-dev using only 700M parameters without any test time augmentation. It
explores the combination of the powerful FocalNet-Huge backbone with the
effective Stable-DINO detector. Different from existing SOTA models that
utilize an extensive number of parameters and complex training techniques on
large-scale private data or merged data, our model is exclusively trained on
the publicly available dataset Objects365, which ensures the reproducibility of
our approach.
| no_new_dataset | 0.7116 |
2304.13030 | Youmin Zhang | Zhang Youmin, Guo Xianda, Poggi Matteo, Zhu Zheng, Huang Guan,
Mattoccia Stefano | CompletionFormer: Depth Completion with Convolutions and Vision
Transformers | Accepted by CVPR 2023. Code:
https://github.com/youmi-zym/CompletionFormer. Project:
https://youmi-zym.github.io/projects/CompletionFormer/ | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Given sparse depths and the corresponding RGB images, depth completion aims
at spatially propagating the sparse measurements throughout the whole image to
get a dense depth prediction. Despite the tremendous progress of
deep-learning-based depth completion methods, the locality of the convolutional
layer or graph model makes it hard for the network to model the long-range
relationship between pixels. While recent fully Transformer-based architecture
has reported encouraging results with the global receptive field, the
performance and efficiency gaps to the well-developed CNN models still exist
because of its deteriorative local feature details. This paper proposes a Joint
Convolutional Attention and Transformer block (JCAT), which deeply couples the
convolutional attention layer and Vision Transformer into one block, as the
basic unit to construct our depth completion model in a pyramidal structure.
This hybrid architecture naturally benefits both the local connectivity of
convolutions and the global context of the Transformer in one single model. As
a result, our CompletionFormer outperforms state-of-the-art CNNs-based methods
on the outdoor KITTI Depth Completion benchmark and indoor NYUv2 dataset,
achieving significantly higher efficiency (nearly 1/3 FLOPs) compared to pure
Transformer-based methods. Code is available at
\url{https://github.com/youmi-zym/CompletionFormer}.
| [
{
"version": "v1",
"created": "Tue, 25 Apr 2023 17:59:47 GMT"
}
] | 2023-04-26T00:00:00 | [
[
"Youmin",
"Zhang",
""
],
[
"Xianda",
"Guo",
""
],
[
"Matteo",
"Poggi",
""
],
[
"Zheng",
"Zhu",
""
],
[
"Guan",
"Huang",
""
],
[
"Stefano",
"Mattoccia",
""
]
] | TITLE: CompletionFormer: Depth Completion with Convolutions and Vision
Transformers
ABSTRACT: Given sparse depths and the corresponding RGB images, depth completion aims
at spatially propagating the sparse measurements throughout the whole image to
get a dense depth prediction. Despite the tremendous progress of
deep-learning-based depth completion methods, the locality of the convolutional
layer or graph model makes it hard for the network to model the long-range
relationship between pixels. While recent fully Transformer-based architecture
has reported encouraging results with the global receptive field, the
performance and efficiency gaps to the well-developed CNN models still exist
because of its deteriorative local feature details. This paper proposes a Joint
Convolutional Attention and Transformer block (JCAT), which deeply couples the
convolutional attention layer and Vision Transformer into one block, as the
basic unit to construct our depth completion model in a pyramidal structure.
This hybrid architecture naturally benefits both the local connectivity of
convolutions and the global context of the Transformer in one single model. As
a result, our CompletionFormer outperforms state-of-the-art CNNs-based methods
on the outdoor KITTI Depth Completion benchmark and indoor NYUv2 dataset,
achieving significantly higher efficiency (nearly 1/3 FLOPs) compared to pure
Transformer-based methods. Code is available at
\url{https://github.com/youmi-zym/CompletionFormer}.
| no_new_dataset | 0.709403 |
1901.09193 | Fangneng Zhan | Changgong Zhang, Fangneng Zhan, Hongyuan Zhu, Shijian Lu | Scene Text Synthesis for Efficient and Effective Deep Network Training | This work has been merged into another project | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A large amount of annotated training images is critical for training accurate
and robust deep network models but the collection of a large amount of
annotated training images is often time-consuming and costly. Image synthesis
alleviates this constraint by generating annotated training images
automatically by machines which has attracted increasing interest in the recent
deep learning research. We develop an innovative image synthesis technique that
composes annotated training images by realistically embedding foreground
objects of interest (OOI) into background images. The proposed technique
consists of two key components that in principle boost the usefulness of the
synthesized images in deep network training. The first is context-aware
semantic coherence which ensures that the OOI are placed around semantically
coherent regions within the background image. The second is harmonious
appearance adaptation which ensures that the embedded OOI are agreeable to the
surrounding background from both geometry alignment and appearance realism. The
proposed technique has been evaluated over two related but very different
computer vision challenges, namely, scene text detection and scene text
recognition. Experiments over a number of public datasets demonstrate the
effectiveness of our proposed image synthesis technique - the use of our
synthesized images in deep network training is capable of achieving similar or
even better scene text detection and scene text recognition performance as
compared with using real images.
| [
{
"version": "v1",
"created": "Sat, 26 Jan 2019 10:15:24 GMT"
},
{
"version": "v2",
"created": "Mon, 3 Oct 2022 17:47:28 GMT"
},
{
"version": "v3",
"created": "Mon, 24 Apr 2023 12:35:49 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Zhang",
"Changgong",
""
],
[
"Zhan",
"Fangneng",
""
],
[
"Zhu",
"Hongyuan",
""
],
[
"Lu",
"Shijian",
""
]
] | TITLE: Scene Text Synthesis for Efficient and Effective Deep Network Training
ABSTRACT: A large amount of annotated training images is critical for training accurate
and robust deep network models but the collection of a large amount of
annotated training images is often time-consuming and costly. Image synthesis
alleviates this constraint by generating annotated training images
automatically by machines which has attracted increasing interest in the recent
deep learning research. We develop an innovative image synthesis technique that
composes annotated training images by realistically embedding foreground
objects of interest (OOI) into background images. The proposed technique
consists of two key components that in principle boost the usefulness of the
synthesized images in deep network training. The first is context-aware
semantic coherence which ensures that the OOI are placed around semantically
coherent regions within the background image. The second is harmonious
appearance adaptation which ensures that the embedded OOI are agreeable to the
surrounding background from both geometry alignment and appearance realism. The
proposed technique has been evaluated over two related but very different
computer vision challenges, namely, scene text detection and scene text
recognition. Experiments over a number of public datasets demonstrate the
effectiveness of our proposed image synthesis technique - the use of our
synthesized images in deep network training is capable of achieving similar or
even better scene text detection and scene text recognition performance as
compared with using real images.
| no_new_dataset | 0.712013 |
2107.08574 | Mohamed Abdelhack | Mohamed Abdelhack, Jiaming Zhang, Sandhya Tripathi, Bradley A Fritz,
Daniel Felsky, Michael S Avidan, Yixin Chen, Christopher R King | A Modulation Layer to Increase Neural Network Robustness Against Data
Quality Issues | null | Transactions on Machine Learning Research 2023 | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Data missingness and quality are common problems in machine learning,
especially for high-stakes applications such as healthcare. Developers often
train machine learning models on carefully curated datasets using only high
quality data; however, this reduces the utility of such models in production
environments. We propose a novel neural network modification to mitigate the
impacts of low quality and missing data which involves replacing the fixed
weights of a fully-connected layer with a function of an additional input. This
is inspired from neuromodulation in biological neural networks where the cortex
can up- and down-regulate inputs based on their reliability and the presence of
other data. In testing, with reliability scores as a modulating signal, models
with modulating layers were found to be more robust against degradation of data
quality, including additional missingness. These models are superior to
imputation as they save on training time by completely skipping the imputation
process and further allow the introduction of other data quality measures that
imputation cannot handle. Our results suggest that explicitly accounting for
reduced information quality with a modulating fully connected layer can enable
the deployment of artificial intelligence systems in real-time applications.
| [
{
"version": "v1",
"created": "Mon, 19 Jul 2021 01:29:16 GMT"
},
{
"version": "v2",
"created": "Mon, 10 Oct 2022 17:32:42 GMT"
},
{
"version": "v3",
"created": "Tue, 18 Apr 2023 00:48:08 GMT"
},
{
"version": "v4",
"created": "Sat, 22 Apr 2023 19:48:08 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Abdelhack",
"Mohamed",
""
],
[
"Zhang",
"Jiaming",
""
],
[
"Tripathi",
"Sandhya",
""
],
[
"Fritz",
"Bradley A",
""
],
[
"Felsky",
"Daniel",
""
],
[
"Avidan",
"Michael S",
""
],
[
"Chen",
"Yixin",
""
],
[
"King",
"Christopher R",
""
]
] | TITLE: A Modulation Layer to Increase Neural Network Robustness Against Data
Quality Issues
ABSTRACT: Data missingness and quality are common problems in machine learning,
especially for high-stakes applications such as healthcare. Developers often
train machine learning models on carefully curated datasets using only high
quality data; however, this reduces the utility of such models in production
environments. We propose a novel neural network modification to mitigate the
impacts of low quality and missing data which involves replacing the fixed
weights of a fully-connected layer with a function of an additional input. This
is inspired from neuromodulation in biological neural networks where the cortex
can up- and down-regulate inputs based on their reliability and the presence of
other data. In testing, with reliability scores as a modulating signal, models
with modulating layers were found to be more robust against degradation of data
quality, including additional missingness. These models are superior to
imputation as they save on training time by completely skipping the imputation
process and further allow the introduction of other data quality measures that
imputation cannot handle. Our results suggest that explicitly accounting for
reduced information quality with a modulating fully connected layer can enable
the deployment of artificial intelligence systems in real-time applications.
| no_new_dataset | 0.710019 |
2108.06027 | Ruiyang Ren | Ruiyang Ren, Shangwen Lv, Yingqi Qu, Jing Liu, Wayne Xin Zhao,
QiaoQiao She, Hua Wu, Haifeng Wang, Ji-Rong Wen | PAIR: Leveraging Passage-Centric Similarity Relation for Improving Dense
Passage Retrieval | ACL 2021 | null | 10.18653/v1/2021.findings-acl.191 | null | cs.IR cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recently, dense passage retrieval has become a mainstream approach to finding
relevant information in various natural language processing tasks. A number of
studies have been devoted to improving the widely adopted dual-encoder
architecture. However, most of the previous studies only consider query-centric
similarity relation when learning the dual-encoder retriever. In order to
capture more comprehensive similarity relations, we propose a novel approach
that leverages both query-centric and PAssage-centric sImilarity Relations
(called PAIR) for dense passage retrieval. To implement our approach, we make
three major technical contributions by introducing formal formulations of the
two kinds of similarity relations, generating high-quality pseudo labeled data
via knowledge distillation, and designing an effective two-stage training
procedure that incorporates passage-centric similarity relation constraint.
Extensive experiments show that our approach significantly outperforms previous
state-of-the-art models on both MSMARCO and Natural Questions datasets.
| [
{
"version": "v1",
"created": "Fri, 13 Aug 2021 02:07:43 GMT"
},
{
"version": "v2",
"created": "Sun, 23 Apr 2023 17:06:18 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Ren",
"Ruiyang",
""
],
[
"Lv",
"Shangwen",
""
],
[
"Qu",
"Yingqi",
""
],
[
"Liu",
"Jing",
""
],
[
"Zhao",
"Wayne Xin",
""
],
[
"She",
"QiaoQiao",
""
],
[
"Wu",
"Hua",
""
],
[
"Wang",
"Haifeng",
""
],
[
"Wen",
"Ji-Rong",
""
]
] | TITLE: PAIR: Leveraging Passage-Centric Similarity Relation for Improving Dense
Passage Retrieval
ABSTRACT: Recently, dense passage retrieval has become a mainstream approach to finding
relevant information in various natural language processing tasks. A number of
studies have been devoted to improving the widely adopted dual-encoder
architecture. However, most of the previous studies only consider query-centric
similarity relation when learning the dual-encoder retriever. In order to
capture more comprehensive similarity relations, we propose a novel approach
that leverages both query-centric and PAssage-centric sImilarity Relations
(called PAIR) for dense passage retrieval. To implement our approach, we make
three major technical contributions by introducing formal formulations of the
two kinds of similarity relations, generating high-quality pseudo labeled data
via knowledge distillation, and designing an effective two-stage training
procedure that incorporates passage-centric similarity relation constraint.
Extensive experiments show that our approach significantly outperforms previous
state-of-the-art models on both MSMARCO and Natural Questions datasets.
| no_new_dataset | 0.710226 |
2110.07367 | Ruiyang Ren | Ruiyang Ren, Yingqi Qu, Jing Liu, Wayne Xin Zhao, Qiaoqiao She, Hua
Wu, Haifeng Wang and Ji-Rong Wen | RocketQAv2: A Joint Training Method for Dense Passage Retrieval and
Passage Re-ranking | EMNLP 2021 | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In various natural language processing tasks, passage retrieval and passage
re-ranking are two key procedures in finding and ranking relevant information.
Since both the two procedures contribute to the final performance, it is
important to jointly optimize them in order to achieve mutual improvement. In
this paper, we propose a novel joint training approach for dense passage
retrieval and passage re-ranking. A major contribution is that we introduce the
dynamic listwise distillation, where we design a unified listwise training
approach for both the retriever and the re-ranker. During the dynamic
distillation, the retriever and the re-ranker can be adaptively improved
according to each other's relevance information. We also propose a hybrid data
augmentation strategy to construct diverse training instances for listwise
training approach. Extensive experiments show the effectiveness of our approach
on both MSMARCO and Natural Questions datasets. Our code is available at
https://github.com/PaddlePaddle/RocketQA.
| [
{
"version": "v1",
"created": "Thu, 14 Oct 2021 13:52:55 GMT"
},
{
"version": "v2",
"created": "Sun, 23 Apr 2023 16:56:52 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Ren",
"Ruiyang",
""
],
[
"Qu",
"Yingqi",
""
],
[
"Liu",
"Jing",
""
],
[
"Zhao",
"Wayne Xin",
""
],
[
"She",
"Qiaoqiao",
""
],
[
"Wu",
"Hua",
""
],
[
"Wang",
"Haifeng",
""
],
[
"Wen",
"Ji-Rong",
""
]
] | TITLE: RocketQAv2: A Joint Training Method for Dense Passage Retrieval and
Passage Re-ranking
ABSTRACT: In various natural language processing tasks, passage retrieval and passage
re-ranking are two key procedures in finding and ranking relevant information.
Since both the two procedures contribute to the final performance, it is
important to jointly optimize them in order to achieve mutual improvement. In
this paper, we propose a novel joint training approach for dense passage
retrieval and passage re-ranking. A major contribution is that we introduce the
dynamic listwise distillation, where we design a unified listwise training
approach for both the retriever and the re-ranker. During the dynamic
distillation, the retriever and the re-ranker can be adaptively improved
according to each other's relevance information. We also propose a hybrid data
augmentation strategy to construct diverse training instances for listwise
training approach. Extensive experiments show the effectiveness of our approach
on both MSMARCO and Natural Questions datasets. Our code is available at
https://github.com/PaddlePaddle/RocketQA.
| no_new_dataset | 0.71022 |
2112.01801 | Huan Lei | Huan Lei, Naveed Akhtar, Mubarak Shah, and Ajmal Mian | Mesh Convolution with Continuous Filters for 3D Surface Parsing | Accepted to TNNLS | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Geometric feature learning for 3D surfaces is critical for many applications
in computer graphics and 3D vision. However, deep learning currently lags in
hierarchical modeling of 3D surfaces due to the lack of required operations
and/or their efficient implementations. In this paper, we propose a series of
modular operations for effective geometric feature learning from 3D triangle
meshes. These operations include novel mesh convolutions, efficient mesh
decimation and associated mesh (un)poolings. Our mesh convolutions exploit
spherical harmonics as orthonormal bases to create continuous convolutional
filters. The mesh decimation module is GPU-accelerated and able to process
batched meshes on-the-fly, while the (un)pooling operations compute features
for up/down-sampled meshes. We provide open-source implementation of these
operations, collectively termed Picasso. Picasso supports heterogeneous mesh
batching and processing. Leveraging its modular operations, we further
contribute a novel hierarchical neural network for perceptual parsing of 3D
surfaces, named PicassoNet++. It achieves highly competitive performance for
shape analysis and scene segmentation on prominent 3D benchmarks. The code,
data and trained models are available at
https://github.com/EnyaHermite/Picasso.
| [
{
"version": "v1",
"created": "Fri, 3 Dec 2021 09:16:49 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Apr 2023 06:35:08 GMT"
},
{
"version": "v3",
"created": "Sat, 22 Apr 2023 02:14:33 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Lei",
"Huan",
""
],
[
"Akhtar",
"Naveed",
""
],
[
"Shah",
"Mubarak",
""
],
[
"Mian",
"Ajmal",
""
]
] | TITLE: Mesh Convolution with Continuous Filters for 3D Surface Parsing
ABSTRACT: Geometric feature learning for 3D surfaces is critical for many applications
in computer graphics and 3D vision. However, deep learning currently lags in
hierarchical modeling of 3D surfaces due to the lack of required operations
and/or their efficient implementations. In this paper, we propose a series of
modular operations for effective geometric feature learning from 3D triangle
meshes. These operations include novel mesh convolutions, efficient mesh
decimation and associated mesh (un)poolings. Our mesh convolutions exploit
spherical harmonics as orthonormal bases to create continuous convolutional
filters. The mesh decimation module is GPU-accelerated and able to process
batched meshes on-the-fly, while the (un)pooling operations compute features
for up/down-sampled meshes. We provide open-source implementation of these
operations, collectively termed Picasso. Picasso supports heterogeneous mesh
batching and processing. Leveraging its modular operations, we further
contribute a novel hierarchical neural network for perceptual parsing of 3D
surfaces, named PicassoNet++. It achieves highly competitive performance for
shape analysis and scene segmentation on prominent 3D benchmarks. The code,
data and trained models are available at
https://github.com/EnyaHermite/Picasso.
| no_new_dataset | 0.711017 |
2112.05871 | Jiacen Xu | Jiacen Xu, Zhe Zhou, Boyuan Feng, Yufei Ding, Zhou Li | On Adversarial Robustness of Point Cloud Semantic Segmentation | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Recent research efforts on 3D point cloud semantic segmentation (PCSS) have
achieved outstanding performance by adopting neural networks. However, the
robustness of these complex models have not been systematically analyzed. Given
that PCSS has been applied in many safety-critical applications like autonomous
driving, it is important to fill this knowledge gap, especially, how these
models are affected under adversarial samples. As such, we present a
comparative study of PCSS robustness. First, we formally define the attacker's
objective under performance degradation and object hiding. Then, we develop new
attack by whether to bound the norm. We evaluate different attack options on
two datasets and three PCSS models. We found all the models are vulnerable and
attacking point color is more effective. With this study, we call the attention
of the research community to develop new approaches to harden PCSS models.
| [
{
"version": "v1",
"created": "Sat, 11 Dec 2021 00:10:00 GMT"
},
{
"version": "v2",
"created": "Sat, 18 Dec 2021 21:23:47 GMT"
},
{
"version": "v3",
"created": "Sun, 30 Oct 2022 20:52:37 GMT"
},
{
"version": "v4",
"created": "Sun, 23 Apr 2023 23:45:13 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Xu",
"Jiacen",
""
],
[
"Zhou",
"Zhe",
""
],
[
"Feng",
"Boyuan",
""
],
[
"Ding",
"Yufei",
""
],
[
"Li",
"Zhou",
""
]
] | TITLE: On Adversarial Robustness of Point Cloud Semantic Segmentation
ABSTRACT: Recent research efforts on 3D point cloud semantic segmentation (PCSS) have
achieved outstanding performance by adopting neural networks. However, the
robustness of these complex models have not been systematically analyzed. Given
that PCSS has been applied in many safety-critical applications like autonomous
driving, it is important to fill this knowledge gap, especially, how these
models are affected under adversarial samples. As such, we present a
comparative study of PCSS robustness. First, we formally define the attacker's
objective under performance degradation and object hiding. Then, we develop new
attack by whether to bound the norm. We evaluate different attack options on
two datasets and three PCSS models. We found all the models are vulnerable and
attacking point color is more effective. With this study, we call the attention
of the research community to develop new approaches to harden PCSS models.
| no_new_dataset | 0.700332 |
2112.13143 | Rishabh Ranjan | Rishabh Ranjan, Siddharth Grover, Sourav Medya, Venkatesan
Chakaravarthy, Yogish Sabharwal, Sayan Ranu | GREED: A Neural Framework for Learning Graph Distance Functions | Published as a conference paper at NeurIPS 2022 | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Among various distance functions for graphs, graph and subgraph edit
distances (GED and SED respectively) are two of the most popular and expressive
measures. Unfortunately, exact computations for both are NP-hard. To overcome
this computational bottleneck, neural approaches to learn and predict edit
distance in polynomial time have received much interest. While considerable
progress has been made, there exist limitations that need to be addressed.
First, the efficacy of an approximate distance function lies not only in its
approximation accuracy, but also in the preservation of its properties. To
elaborate, although GED is a metric, its neural approximations do not provide
such a guarantee. This prohibits their usage in higher order tasks that rely on
metric distance functions, such as clustering or indexing. Second, several
existing frameworks for GED do not extend to SED due to SED being asymmetric.
In this work, we design a novel siamese graph neural network called GREED,
which through a carefully crafted inductive bias, learns GED and SED in a
property-preserving manner. Through extensive experiments across 10 real graph
datasets containing up to 7 million edges, we establish that GREED is not only
more accurate than the state of the art, but also up to 3 orders of magnitude
faster. Even more significantly, due to preserving the triangle inequality, the
generated embeddings are indexable and consequently, even in a CPU-only
environment, GREED is up to 50 times faster than GPU-powered baselines for
graph / subgraph retrieval.
| [
{
"version": "v1",
"created": "Fri, 24 Dec 2021 21:46:40 GMT"
},
{
"version": "v2",
"created": "Tue, 28 Dec 2021 16:52:26 GMT"
},
{
"version": "v3",
"created": "Fri, 21 Apr 2023 22:52:25 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Ranjan",
"Rishabh",
""
],
[
"Grover",
"Siddharth",
""
],
[
"Medya",
"Sourav",
""
],
[
"Chakaravarthy",
"Venkatesan",
""
],
[
"Sabharwal",
"Yogish",
""
],
[
"Ranu",
"Sayan",
""
]
] | TITLE: GREED: A Neural Framework for Learning Graph Distance Functions
ABSTRACT: Among various distance functions for graphs, graph and subgraph edit
distances (GED and SED respectively) are two of the most popular and expressive
measures. Unfortunately, exact computations for both are NP-hard. To overcome
this computational bottleneck, neural approaches to learn and predict edit
distance in polynomial time have received much interest. While considerable
progress has been made, there exist limitations that need to be addressed.
First, the efficacy of an approximate distance function lies not only in its
approximation accuracy, but also in the preservation of its properties. To
elaborate, although GED is a metric, its neural approximations do not provide
such a guarantee. This prohibits their usage in higher order tasks that rely on
metric distance functions, such as clustering or indexing. Second, several
existing frameworks for GED do not extend to SED due to SED being asymmetric.
In this work, we design a novel siamese graph neural network called GREED,
which through a carefully crafted inductive bias, learns GED and SED in a
property-preserving manner. Through extensive experiments across 10 real graph
datasets containing up to 7 million edges, we establish that GREED is not only
more accurate than the state of the art, but also up to 3 orders of magnitude
faster. Even more significantly, due to preserving the triangle inequality, the
generated embeddings are indexable and consequently, even in a CPU-only
environment, GREED is up to 50 times faster than GPU-powered baselines for
graph / subgraph retrieval.
| no_new_dataset | 0.709177 |
2201.08953 | Guoyang Xie | Jinbao Wang, Guoyang Xie, Yawen Huang, Jiayi Lyu, Yefeng Zheng, Feng
Zheng, Yaochu Jin | FedMed-GAN: Federated Domain Translation on Unsupervised Cross-Modality
Brain Image Synthesis | null | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Utilizing multi-modal neuroimaging data has been proved to be effective to
investigate human cognitive activities and certain pathologies. However, it is
not practical to obtain the full set of paired neuroimaging data centrally
since the collection faces several constraints, e.g., high examination cost,
long acquisition time, and image corruption. In addition, these data are
dispersed into different medical institutions and thus cannot be aggregated for
centralized training considering the privacy issues. There is a clear need to
launch a federated learning and facilitate the integration of the dispersed
data from different institutions. In this paper, we propose a new benchmark for
federated domain translation on unsupervised brain image synthesis (termed as
FedMed-GAN) to bridge the gap between federated learning and medical GAN.
FedMed-GAN mitigates the mode collapse without sacrificing the performance of
generators, and is widely applied to different proportions of unpaired and
paired data with variation adaptation property. We treat the gradient penalties
by federally averaging algorithm and then leveraging differential privacy
gradient descent to regularize the training dynamics. A comprehensive
evaluation is provided for comparing FedMed-GAN and other centralized methods,
which shows the new state-of-the-art performance by our FedMed-GAN. Our code
has been released on the website: https://github.com/M-3LAB/FedMed-GAN
| [
{
"version": "v1",
"created": "Sat, 22 Jan 2022 02:50:29 GMT"
},
{
"version": "v2",
"created": "Tue, 1 Mar 2022 13:56:12 GMT"
},
{
"version": "v3",
"created": "Sun, 23 Apr 2023 03:43:23 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Wang",
"Jinbao",
""
],
[
"Xie",
"Guoyang",
""
],
[
"Huang",
"Yawen",
""
],
[
"Lyu",
"Jiayi",
""
],
[
"Zheng",
"Yefeng",
""
],
[
"Zheng",
"Feng",
""
],
[
"Jin",
"Yaochu",
""
]
] | TITLE: FedMed-GAN: Federated Domain Translation on Unsupervised Cross-Modality
Brain Image Synthesis
ABSTRACT: Utilizing multi-modal neuroimaging data has been proved to be effective to
investigate human cognitive activities and certain pathologies. However, it is
not practical to obtain the full set of paired neuroimaging data centrally
since the collection faces several constraints, e.g., high examination cost,
long acquisition time, and image corruption. In addition, these data are
dispersed into different medical institutions and thus cannot be aggregated for
centralized training considering the privacy issues. There is a clear need to
launch a federated learning and facilitate the integration of the dispersed
data from different institutions. In this paper, we propose a new benchmark for
federated domain translation on unsupervised brain image synthesis (termed as
FedMed-GAN) to bridge the gap between federated learning and medical GAN.
FedMed-GAN mitigates the mode collapse without sacrificing the performance of
generators, and is widely applied to different proportions of unpaired and
paired data with variation adaptation property. We treat the gradient penalties
by federally averaging algorithm and then leveraging differential privacy
gradient descent to regularize the training dynamics. A comprehensive
evaluation is provided for comparing FedMed-GAN and other centralized methods,
which shows the new state-of-the-art performance by our FedMed-GAN. Our code
has been released on the website: https://github.com/M-3LAB/FedMed-GAN
| no_new_dataset | 0.707815 |
2203.03549 | Roshni Iyer | Roshni G. Iyer, Thuy Vu, Alessandro Moschitti, Yizhou Sun | Question-Answer Sentence Graph for Joint Modeling Answer Selection | null | null | 10.48550/arXiv.2203.03549 | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This research studies graph-based approaches for Answer Sentence Selection
(AS2), an essential component for retrieval-based Question Answering (QA)
systems. During offline learning, our model constructs a small-scale relevant
training graph per question in an unsupervised manner, and integrates with
Graph Neural Networks. Graph nodes are question sentence to answer sentence
pairs. We train and integrate state-of-the-art (SOTA) models for computing
scores between question-question, question-answer, and answer-answer pairs, and
use thresholding on relevance scores for creating graph edges. Online inference
is then performed to solve the AS2 task on unseen queries. Experiments on two
well-known academic benchmarks and a real-world dataset show that our approach
consistently outperforms SOTA QA baseline models.
| [
{
"version": "v1",
"created": "Wed, 16 Feb 2022 05:59:53 GMT"
},
{
"version": "v2",
"created": "Sun, 23 Apr 2023 04:01:56 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Iyer",
"Roshni G.",
""
],
[
"Vu",
"Thuy",
""
],
[
"Moschitti",
"Alessandro",
""
],
[
"Sun",
"Yizhou",
""
]
] | TITLE: Question-Answer Sentence Graph for Joint Modeling Answer Selection
ABSTRACT: This research studies graph-based approaches for Answer Sentence Selection
(AS2), an essential component for retrieval-based Question Answering (QA)
systems. During offline learning, our model constructs a small-scale relevant
training graph per question in an unsupervised manner, and integrates with
Graph Neural Networks. Graph nodes are question sentence to answer sentence
pairs. We train and integrate state-of-the-art (SOTA) models for computing
scores between question-question, question-answer, and answer-answer pairs, and
use thresholding on relevance scores for creating graph edges. Online inference
is then performed to solve the AS2 task on unseen queries. Experiments on two
well-known academic benchmarks and a real-world dataset show that our approach
consistently outperforms SOTA QA baseline models.
| no_new_dataset | 0.710427 |
2203.14113 | Filipa Valdeira | Filipa Valdeira and Ricardo Ferreira and Alessandra Micheletti and
Cl\'audia Soares | Probabilistic Registration for Gaussian Process 3D shape modelling in
the presence of extensive missing data | 18 pages, 6 figures. Accepted for publication in SIAM Journal on
Mathematics of Data Science (SIMODS) | null | null | null | cs.CV stat.AP | http://creativecommons.org/licenses/by/4.0/ | We propose a shape fitting/registration method based on a Gaussian Processes
formulation, suitable for shapes with extensive regions of missing data.
Gaussian Processes are a proven powerful tool, as they provide a unified
setting for shape modelling and fitting. While the existing methods in this
area prove to work well for the general case of the human head, when looking at
more detailed and deformed data, with a high prevalence of missing data, such
as the ears, the results are not satisfactory. In order to overcome this, we
formulate the shape fitting problem as a multi-annotator Gaussian Process
Regression and establish a parallel with the standard probabilistic
registration. The achieved method SFGP shows better performance when dealing
with extensive areas of missing data when compared to a state-of-the-art
registration method and current approaches for registration with pre-existing
shape models. Experiments are conducted both for a 2D small dataset with
diverse transformations and a 3D dataset of ears.
| [
{
"version": "v1",
"created": "Sat, 26 Mar 2022 16:48:27 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Apr 2023 09:30:43 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Valdeira",
"Filipa",
""
],
[
"Ferreira",
"Ricardo",
""
],
[
"Micheletti",
"Alessandra",
""
],
[
"Soares",
"Cláudia",
""
]
] | TITLE: Probabilistic Registration for Gaussian Process 3D shape modelling in
the presence of extensive missing data
ABSTRACT: We propose a shape fitting/registration method based on a Gaussian Processes
formulation, suitable for shapes with extensive regions of missing data.
Gaussian Processes are a proven powerful tool, as they provide a unified
setting for shape modelling and fitting. While the existing methods in this
area prove to work well for the general case of the human head, when looking at
more detailed and deformed data, with a high prevalence of missing data, such
as the ears, the results are not satisfactory. In order to overcome this, we
formulate the shape fitting problem as a multi-annotator Gaussian Process
Regression and establish a parallel with the standard probabilistic
registration. The achieved method SFGP shows better performance when dealing
with extensive areas of missing data when compared to a state-of-the-art
registration method and current approaches for registration with pre-existing
shape models. Experiments are conducted both for a 2D small dataset with
diverse transformations and a 3D dataset of ears.
| no_new_dataset | 0.71103 |
2204.01266 | Chongming Gao | Chongming Gao, Shiqi Wang, Shijun Li, Jiawei Chen, Xiangnan He,
Wenqiang Lei, Biao Li, Yuan Zhang, Peng Jiang | CIRS: Bursting Filter Bubbles by Counterfactual Interactive Recommender
System | Accepted by ACM Transactions on Information Systems (TOIS 2023) | null | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While personalization increases the utility of recommender systems, it also
brings the issue of filter bubbles. E.g., if the system keeps exposing and
recommending the items that the user is interested in, it may also make the
user feel bored and less satisfied. Existing work studies filter bubbles in
static recommendation, where the effect of overexposure is hard to capture. In
contrast, we believe it is more meaningful to study the issue in interactive
recommendation and optimize long-term user satisfaction. Nevertheless, it is
unrealistic to train the model online due to the high cost. As such, we have to
leverage offline training data and disentangle the causal effect on user
satisfaction.
To achieve this goal, we propose a counterfactual interactive recommender
system (CIRS) that augments offline reinforcement learning (offline RL) with
causal inference. The basic idea is to first learn a causal user model on
historical data to capture the overexposure effect of items on user
satisfaction. It then uses the learned causal user model to help the planning
of the RL policy. To conduct evaluation offline, we innovatively create an
authentic RL environment (KuaiEnv) based on a real-world fully observed user
rating dataset. The experiments show the effectiveness of CIRS in bursting
filter bubbles and achieving long-term success in interactive recommendation.
The implementation of CIRS is available via
https://github.com/chongminggao/CIRS-codes.
| [
{
"version": "v1",
"created": "Mon, 4 Apr 2022 06:20:48 GMT"
},
{
"version": "v2",
"created": "Sat, 22 Apr 2023 12:21:48 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Gao",
"Chongming",
""
],
[
"Wang",
"Shiqi",
""
],
[
"Li",
"Shijun",
""
],
[
"Chen",
"Jiawei",
""
],
[
"He",
"Xiangnan",
""
],
[
"Lei",
"Wenqiang",
""
],
[
"Li",
"Biao",
""
],
[
"Zhang",
"Yuan",
""
],
[
"Jiang",
"Peng",
""
]
] | TITLE: CIRS: Bursting Filter Bubbles by Counterfactual Interactive Recommender
System
ABSTRACT: While personalization increases the utility of recommender systems, it also
brings the issue of filter bubbles. E.g., if the system keeps exposing and
recommending the items that the user is interested in, it may also make the
user feel bored and less satisfied. Existing work studies filter bubbles in
static recommendation, where the effect of overexposure is hard to capture. In
contrast, we believe it is more meaningful to study the issue in interactive
recommendation and optimize long-term user satisfaction. Nevertheless, it is
unrealistic to train the model online due to the high cost. As such, we have to
leverage offline training data and disentangle the causal effect on user
satisfaction.
To achieve this goal, we propose a counterfactual interactive recommender
system (CIRS) that augments offline reinforcement learning (offline RL) with
causal inference. The basic idea is to first learn a causal user model on
historical data to capture the overexposure effect of items on user
satisfaction. It then uses the learned causal user model to help the planning
of the RL policy. To conduct evaluation offline, we innovatively create an
authentic RL environment (KuaiEnv) based on a real-world fully observed user
rating dataset. The experiments show the effectiveness of CIRS in bursting
filter bubbles and achieving long-term success in interactive recommendation.
The implementation of CIRS is available via
https://github.com/chongminggao/CIRS-codes.
| no_new_dataset | 0.708629 |
2204.03574 | Nihal V. Nayak | Nihal V. Nayak, Peilin Yu, Stephen H. Bach | Learning to Compose Soft Prompts for Compositional Zero-Shot Learning | ICLR 2023 | null | null | null | cs.LG cs.CL cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce compositional soft prompting (CSP), a parameter-efficient
learning technique to improve the zero-shot compositionality of large-scale
pretrained vision-language models (VLMs) like CLIP. We develop CSP for
compositional zero-shot learning, the task of predicting unseen
attribute-object compositions (e.g., old cat and young tiger). VLMs have a
flexible text encoder that can represent arbitrary classes as natural language
prompts but they often underperform task-specific architectures on the
compositional zero-shot benchmark datasets. CSP treats the attributes and
objects that define classes as learnable tokens of vocabulary. During training,
the vocabulary is tuned to recognize classes that compose tokens in multiple
ways (e.g., old cat and white cat). At test time, we recompose the learned
attribute-object vocabulary in new combinations to recognize novel classes. We
show that CSP outperforms the CLIP on benchmark datasets by an average of 10.9
percentage points on AUC. CSP also outperforms CoOp, a soft prompting method
that fine-tunes the prefix context tokens, by an average of 5.8 percentage
points on AUC. We perform additional experiments to show that CSP improves
generalization to higher-order attribute-attribute-object compositions (e.g.,
old white cat) and combinations of pretrained attributes and fine-tuned
objects. The code is available at https://github.com/BatsResearch/csp.
| [
{
"version": "v1",
"created": "Thu, 7 Apr 2022 16:51:12 GMT"
},
{
"version": "v2",
"created": "Sat, 10 Sep 2022 23:34:24 GMT"
},
{
"version": "v3",
"created": "Mon, 24 Apr 2023 15:46:28 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Nayak",
"Nihal V.",
""
],
[
"Yu",
"Peilin",
""
],
[
"Bach",
"Stephen H.",
""
]
] | TITLE: Learning to Compose Soft Prompts for Compositional Zero-Shot Learning
ABSTRACT: We introduce compositional soft prompting (CSP), a parameter-efficient
learning technique to improve the zero-shot compositionality of large-scale
pretrained vision-language models (VLMs) like CLIP. We develop CSP for
compositional zero-shot learning, the task of predicting unseen
attribute-object compositions (e.g., old cat and young tiger). VLMs have a
flexible text encoder that can represent arbitrary classes as natural language
prompts but they often underperform task-specific architectures on the
compositional zero-shot benchmark datasets. CSP treats the attributes and
objects that define classes as learnable tokens of vocabulary. During training,
the vocabulary is tuned to recognize classes that compose tokens in multiple
ways (e.g., old cat and white cat). At test time, we recompose the learned
attribute-object vocabulary in new combinations to recognize novel classes. We
show that CSP outperforms the CLIP on benchmark datasets by an average of 10.9
percentage points on AUC. CSP also outperforms CoOp, a soft prompting method
that fine-tunes the prefix context tokens, by an average of 5.8 percentage
points on AUC. We perform additional experiments to show that CSP improves
generalization to higher-order attribute-attribute-object compositions (e.g.,
old white cat) and combinations of pretrained attributes and fine-tuned
objects. The code is available at https://github.com/BatsResearch/csp.
| no_new_dataset | 0.710823 |
2204.04792 | Yuzhou Jiang | Yuzhou Jiang, Emre Yilmaz, and Erman Ayday | Robust Fingerprint of Location Trajectories Under Differential Privacy | null | null | null | null | cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Directly releasing those data raises privacy and liability (e.g., due to
unauthorized distribution of such datasets) concerns since location data
contain users' sensitive information, e.g., regular moving patterns and
favorite spots. To address this, we propose a novel fingerprinting scheme that
simultaneously identifies unauthorized redistribution of location datasets and
provides differential privacy guarantees for the shared data. Observing data
utility degradation due to differentially-private mechanisms, we introduce a
utility-focused post-processing scheme to regain spatio-temporal correlations
between points in a location trajectory. We further integrate this
post-processing scheme into our fingerprinting scheme as a sampling method. The
proposed fingerprinting scheme alleviates the degradation in the utility of the
shared dataset due to the noise introduced by differentially-private mechanisms
(i.e., adds the fingerprint by preserving the publicly known statistics of the
data). Meanwhile, it does not violate differential privacy throughout the
entire process due to immunity to post-processing, a fundamental property of
differential privacy. Our proposed fingerprinting scheme is robust against
known and well-studied attacks against a fingerprinting scheme including random
flipping attacks, correlation-based flipping attacks, and collusions among
multiple parties, which makes it hard for the attackers to infer the
fingerprint codes and avoid accusation. Via experiments on two real-life
location datasets and two synthetic ones, we show that our scheme achieves high
fingerprinting robustness and outperforms existing approaches. Besides, the
proposed fingerprinting scheme increases data utility for
differentially-private datasets, which is beneficial for data analyzers.
| [
{
"version": "v1",
"created": "Sun, 10 Apr 2022 23:06:39 GMT"
},
{
"version": "v2",
"created": "Tue, 13 Sep 2022 22:20:35 GMT"
},
{
"version": "v3",
"created": "Fri, 21 Apr 2023 19:27:09 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Jiang",
"Yuzhou",
""
],
[
"Yilmaz",
"Emre",
""
],
[
"Ayday",
"Erman",
""
]
] | TITLE: Robust Fingerprint of Location Trajectories Under Differential Privacy
ABSTRACT: Directly releasing those data raises privacy and liability (e.g., due to
unauthorized distribution of such datasets) concerns since location data
contain users' sensitive information, e.g., regular moving patterns and
favorite spots. To address this, we propose a novel fingerprinting scheme that
simultaneously identifies unauthorized redistribution of location datasets and
provides differential privacy guarantees for the shared data. Observing data
utility degradation due to differentially-private mechanisms, we introduce a
utility-focused post-processing scheme to regain spatio-temporal correlations
between points in a location trajectory. We further integrate this
post-processing scheme into our fingerprinting scheme as a sampling method. The
proposed fingerprinting scheme alleviates the degradation in the utility of the
shared dataset due to the noise introduced by differentially-private mechanisms
(i.e., adds the fingerprint by preserving the publicly known statistics of the
data). Meanwhile, it does not violate differential privacy throughout the
entire process due to immunity to post-processing, a fundamental property of
differential privacy. Our proposed fingerprinting scheme is robust against
known and well-studied attacks against a fingerprinting scheme including random
flipping attacks, correlation-based flipping attacks, and collusions among
multiple parties, which makes it hard for the attackers to infer the
fingerprint codes and avoid accusation. Via experiments on two real-life
location datasets and two synthetic ones, we show that our scheme achieves high
fingerprinting robustness and outperforms existing approaches. Besides, the
proposed fingerprinting scheme increases data utility for
differentially-private datasets, which is beneficial for data analyzers.
| no_new_dataset | 0.710635 |
2204.08516 | Pedro Miguel Sanchez Sanchez | Pedro Miguel S\'anchez S\'anchez, Jos\'e Mar\'ia Jorquera Valero,
Alberto Huertas Celdr\'an, G\'er\^ome Bovet, Manuel Gil P\'erez, Gregorio
Mart\'inez P\'erez | LwHBench: A low-level hardware component benchmark and dataset for
Single Board Computers | null | null | 10.1016/j.iot.2023.100764 | null | cs.PF | http://creativecommons.org/licenses/by/4.0/ | In today's computing environment, where Artificial Intelligence (AI) and data
processing are moving toward the Internet of Things (IoT) and Edge computing
paradigms, benchmarking resource-constrained devices is a critical task to
evaluate their suitability and performance. Between the employed devices,
Single-Board Computers arise as multi-purpose and affordable systems. The
literature has explored Single-Board Computers performance when running
high-level benchmarks specialized in particular application scenarios, such as
AI or medical applications. However, lower-level benchmarking applications and
datasets are needed to enable new Edge-based AI solutions for network, system
and service management based on device and component performance, such as
individual device identification. Thus, this paper presents LwHBench, a
low-level hardware benchmarking application for Single-Board Computers that
measures the performance of CPU, GPU, Memory and Storage taking into account
the component constraints in these types of devices. LwHBench has been
implemented for Raspberry Pi devices and run for 100 days on a set of 45
devices to generate an extensive dataset that allows the usage of AI techniques
in scenarios where performance data can help in the device management process.
Besides, to demonstrate the inter-scenario capability of the dataset, a series
of AI-enabled use cases about device identification and context impact on
performance are presented as exploration of the published data. Finally, the
benchmark application has been adapted and applied to an agriculture-focused
scenario where three RockPro64 devices are present.
| [
{
"version": "v1",
"created": "Mon, 18 Apr 2022 18:58:38 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Oct 2022 12:02:35 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Sánchez",
"Pedro Miguel Sánchez",
""
],
[
"Valero",
"José María Jorquera",
""
],
[
"Celdrán",
"Alberto Huertas",
""
],
[
"Bovet",
"Gérôme",
""
],
[
"Pérez",
"Manuel Gil",
""
],
[
"Pérez",
"Gregorio Martínez",
""
]
] | TITLE: LwHBench: A low-level hardware component benchmark and dataset for
Single Board Computers
ABSTRACT: In today's computing environment, where Artificial Intelligence (AI) and data
processing are moving toward the Internet of Things (IoT) and Edge computing
paradigms, benchmarking resource-constrained devices is a critical task to
evaluate their suitability and performance. Between the employed devices,
Single-Board Computers arise as multi-purpose and affordable systems. The
literature has explored Single-Board Computers performance when running
high-level benchmarks specialized in particular application scenarios, such as
AI or medical applications. However, lower-level benchmarking applications and
datasets are needed to enable new Edge-based AI solutions for network, system
and service management based on device and component performance, such as
individual device identification. Thus, this paper presents LwHBench, a
low-level hardware benchmarking application for Single-Board Computers that
measures the performance of CPU, GPU, Memory and Storage taking into account
the component constraints in these types of devices. LwHBench has been
implemented for Raspberry Pi devices and run for 100 days on a set of 45
devices to generate an extensive dataset that allows the usage of AI techniques
in scenarios where performance data can help in the device management process.
Besides, to demonstrate the inter-scenario capability of the dataset, a series
of AI-enabled use cases about device identification and context impact on
performance are presented as exploration of the published data. Finally, the
benchmark application has been adapted and applied to an agriculture-focused
scenario where three RockPro64 devices are present.
| new_dataset | 0.716051 |
2204.08952 | Jianfeng Chi | Md Rizwan Parvez, Jianfeng Chi, Wasi Uddin Ahmad, Yuan Tian, Kai-Wei
Chang | Retrieval Enhanced Data Augmentation for Question Answering on Privacy
Policies | EACL 2023 | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Prior studies in privacy policies frame the question answering (QA) task as
identifying the most relevant text segment or a list of sentences from a policy
document given a user query. Existing labeled datasets are heavily imbalanced
(only a few relevant segments), limiting the QA performance in this domain. In
this paper, we develop a data augmentation framework based on ensembling
retriever models that captures the relevant text segments from unlabeled policy
documents and expand the positive examples in the training set. In addition, to
improve the diversity and quality of the augmented data, we leverage multiple
pre-trained language models (LMs) and cascade them with noise reduction filter
models. Using our augmented data on the PrivacyQA benchmark, we elevate the
existing baseline by a large margin (10\% F1) and achieve a new
state-of-the-art F1 score of 50\%. Our ablation studies provide further
insights into the effectiveness of our approach.
| [
{
"version": "v1",
"created": "Tue, 19 Apr 2022 15:45:23 GMT"
},
{
"version": "v2",
"created": "Thu, 28 Apr 2022 19:33:50 GMT"
},
{
"version": "v3",
"created": "Sat, 22 Apr 2023 05:21:45 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Parvez",
"Md Rizwan",
""
],
[
"Chi",
"Jianfeng",
""
],
[
"Ahmad",
"Wasi Uddin",
""
],
[
"Tian",
"Yuan",
""
],
[
"Chang",
"Kai-Wei",
""
]
] | TITLE: Retrieval Enhanced Data Augmentation for Question Answering on Privacy
Policies
ABSTRACT: Prior studies in privacy policies frame the question answering (QA) task as
identifying the most relevant text segment or a list of sentences from a policy
document given a user query. Existing labeled datasets are heavily imbalanced
(only a few relevant segments), limiting the QA performance in this domain. In
this paper, we develop a data augmentation framework based on ensembling
retriever models that captures the relevant text segments from unlabeled policy
documents and expand the positive examples in the training set. In addition, to
improve the diversity and quality of the augmented data, we leverage multiple
pre-trained language models (LMs) and cascade them with noise reduction filter
models. Using our augmented data on the PrivacyQA benchmark, we elevate the
existing baseline by a large margin (10\% F1) and achieve a new
state-of-the-art F1 score of 50\%. Our ablation studies provide further
insights into the effectiveness of our approach.
| no_new_dataset | 0.711606 |
2204.12755 | Ruiyang Ren | Ruiyang Ren, Yingqi Qu, Jing Liu, Wayne Xin Zhao, Qifei Wu, Yuchen
Ding, Hua Wu, Haifeng Wang, Ji-Rong Wen | A Thorough Examination on Zero-shot Dense Retrieval | null | null | null | null | cs.CL cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent years have witnessed the significant advance in dense retrieval (DR)
based on powerful pre-trained language models (PLM). DR models have achieved
excellent performance in several benchmark datasets, while they are shown to be
not as competitive as traditional sparse retrieval models (e.g., BM25) in a
zero-shot retrieval setting. However, in the related literature, there still
lacks a detailed and comprehensive study on zero-shot retrieval. In this paper,
we present the first thorough examination of the zero-shot capability of DR
models. We aim to identify the key factors and analyze how they affect
zero-shot retrieval performance. In particular, we discuss the effect of
several key factors related to source training set, analyze the potential bias
from the target dataset, and review and compare existing zero-shot DR models.
Our findings provide important evidence to better understand and develop
zero-shot DR models.
| [
{
"version": "v1",
"created": "Wed, 27 Apr 2022 07:59:07 GMT"
},
{
"version": "v2",
"created": "Sun, 23 Apr 2023 17:11:13 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Ren",
"Ruiyang",
""
],
[
"Qu",
"Yingqi",
""
],
[
"Liu",
"Jing",
""
],
[
"Zhao",
"Wayne Xin",
""
],
[
"Wu",
"Qifei",
""
],
[
"Ding",
"Yuchen",
""
],
[
"Wu",
"Hua",
""
],
[
"Wang",
"Haifeng",
""
],
[
"Wen",
"Ji-Rong",
""
]
] | TITLE: A Thorough Examination on Zero-shot Dense Retrieval
ABSTRACT: Recent years have witnessed the significant advance in dense retrieval (DR)
based on powerful pre-trained language models (PLM). DR models have achieved
excellent performance in several benchmark datasets, while they are shown to be
not as competitive as traditional sparse retrieval models (e.g., BM25) in a
zero-shot retrieval setting. However, in the related literature, there still
lacks a detailed and comprehensive study on zero-shot retrieval. In this paper,
we present the first thorough examination of the zero-shot capability of DR
models. We aim to identify the key factors and analyze how they affect
zero-shot retrieval performance. In particular, we discuss the effect of
several key factors related to source training set, analyze the potential bias
from the target dataset, and review and compare existing zero-shot DR models.
Our findings provide important evidence to better understand and develop
zero-shot DR models.
| no_new_dataset | 0.709265 |
2204.13662 | Zicong Fan | Zicong Fan, Omid Taheri, Dimitrios Tzionas, Muhammed Kocabas, Manuel
Kaufmann, Michael J. Black, and Otmar Hilliges | ARCTIC: A Dataset for Dexterous Bimanual Hand-Object Manipulation | Project page: https://arctic.is.tue.mpg.de | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Humans intuitively understand that inanimate objects do not move by
themselves, but that state changes are typically caused by human manipulation
(e.g., the opening of a book). This is not yet the case for machines. In part
this is because there exist no datasets with ground-truth 3D annotations for
the study of physically consistent and synchronised motion of hands and
articulated objects. To this end, we introduce ARCTIC -- a dataset of two hands
that dexterously manipulate objects, containing 2.1M video frames paired with
accurate 3D hand and object meshes and detailed, dynamic contact information.
It contains bi-manual articulation of objects such as scissors or laptops,
where hand poses and object states evolve jointly in time. We propose two novel
articulated hand-object interaction tasks: (1) Consistent motion
reconstruction: Given a monocular video, the goal is to reconstruct two hands
and articulated objects in 3D, so that their motions are spatio-temporally
consistent. (2) Interaction field estimation: Dense relative hand-object
distances must be estimated from images. We introduce two baselines ArcticNet
and InterField, respectively and evaluate them qualitatively and quantitatively
on ARCTIC. Our code and data are available at https://arctic.is.tue.mpg.de.
| [
{
"version": "v1",
"created": "Thu, 28 Apr 2022 17:23:59 GMT"
},
{
"version": "v2",
"created": "Thu, 23 Feb 2023 08:48:07 GMT"
},
{
"version": "v3",
"created": "Sun, 23 Apr 2023 13:11:57 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Fan",
"Zicong",
""
],
[
"Taheri",
"Omid",
""
],
[
"Tzionas",
"Dimitrios",
""
],
[
"Kocabas",
"Muhammed",
""
],
[
"Kaufmann",
"Manuel",
""
],
[
"Black",
"Michael J.",
""
],
[
"Hilliges",
"Otmar",
""
]
] | TITLE: ARCTIC: A Dataset for Dexterous Bimanual Hand-Object Manipulation
ABSTRACT: Humans intuitively understand that inanimate objects do not move by
themselves, but that state changes are typically caused by human manipulation
(e.g., the opening of a book). This is not yet the case for machines. In part
this is because there exist no datasets with ground-truth 3D annotations for
the study of physically consistent and synchronised motion of hands and
articulated objects. To this end, we introduce ARCTIC -- a dataset of two hands
that dexterously manipulate objects, containing 2.1M video frames paired with
accurate 3D hand and object meshes and detailed, dynamic contact information.
It contains bi-manual articulation of objects such as scissors or laptops,
where hand poses and object states evolve jointly in time. We propose two novel
articulated hand-object interaction tasks: (1) Consistent motion
reconstruction: Given a monocular video, the goal is to reconstruct two hands
and articulated objects in 3D, so that their motions are spatio-temporally
consistent. (2) Interaction field estimation: Dense relative hand-object
distances must be estimated from images. We introduce two baselines ArcticNet
and InterField, respectively and evaluate them qualitatively and quantitatively
on ARCTIC. Our code and data are available at https://arctic.is.tue.mpg.de.
| new_dataset | 0.716615 |
2205.08834 | Dominik B\"ar | Dominik B\"ar, Nicolas Pr\"ollochs, Stefan Feuerriegel | Finding Qs: Profiling QAnon Supporters on Parler | Accepted at the International AAAI Conference on Web and Social Media
(ICWSM, 2023) | null | null | null | cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The social media platform "Parler" has emerged into a prominent fringe
community where a significant part of the user base are self-reported
supporters of QAnon, a far-right conspiracy theory alleging that a cabal of
elites controls global politics. QAnon is considered to have had an influential
role in the public discourse during the 2020 U.S. presidential election.
However, little is known about QAnon supporters on Parler and what sets them
aside from other users. Building up on social identity theory, we aim at
profiling the characteristics of QAnon supporters on Parler. We analyze a
large-scale dataset with more than 600,000 profiles of English-speaking users
on Parler. Based on users' profiles, posts, and comments, we then extract a
comprehensive set of user features, linguistic features, network features, and
content features. This allows us to perform user profiling and understand to
what extent these features discriminate between QAnon and non-QAnon supporters
on Parler. Our analysis is three-fold: (1) We quantify the number of QAnon
supporters on Parler, finding that 34,913 users (5.5% of all users) openly
report to support the conspiracy. (2) We examine differences between QAnon vs.
non-QAnon supporters. We find that QAnon supporters differ statistically
significantly from non-QAnon supporters across multiple dimensions. For
example, they have, on average, a larger number of followers, followees, and
posts, and thus have a large impact on the Parler network. (3) We use machine
learning to identify which user characteristics discriminate QAnon from
non-QAnon supporters. We find that user features, linguistic features, network
features, and content features, can - to a large extent - discriminate QAnon
vs. non-QAnon supporters on Parler. In particular, we find that user features
are highly discriminatory, followed by content features and linguistic
features.
| [
{
"version": "v1",
"created": "Wed, 18 May 2022 10:07:12 GMT"
},
{
"version": "v2",
"created": "Wed, 20 Jul 2022 12:51:39 GMT"
},
{
"version": "v3",
"created": "Mon, 24 Apr 2023 12:13:32 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Bär",
"Dominik",
""
],
[
"Pröllochs",
"Nicolas",
""
],
[
"Feuerriegel",
"Stefan",
""
]
] | TITLE: Finding Qs: Profiling QAnon Supporters on Parler
ABSTRACT: The social media platform "Parler" has emerged into a prominent fringe
community where a significant part of the user base are self-reported
supporters of QAnon, a far-right conspiracy theory alleging that a cabal of
elites controls global politics. QAnon is considered to have had an influential
role in the public discourse during the 2020 U.S. presidential election.
However, little is known about QAnon supporters on Parler and what sets them
aside from other users. Building up on social identity theory, we aim at
profiling the characteristics of QAnon supporters on Parler. We analyze a
large-scale dataset with more than 600,000 profiles of English-speaking users
on Parler. Based on users' profiles, posts, and comments, we then extract a
comprehensive set of user features, linguistic features, network features, and
content features. This allows us to perform user profiling and understand to
what extent these features discriminate between QAnon and non-QAnon supporters
on Parler. Our analysis is three-fold: (1) We quantify the number of QAnon
supporters on Parler, finding that 34,913 users (5.5% of all users) openly
report to support the conspiracy. (2) We examine differences between QAnon vs.
non-QAnon supporters. We find that QAnon supporters differ statistically
significantly from non-QAnon supporters across multiple dimensions. For
example, they have, on average, a larger number of followers, followees, and
posts, and thus have a large impact on the Parler network. (3) We use machine
learning to identify which user characteristics discriminate QAnon from
non-QAnon supporters. We find that user features, linguistic features, network
features, and content features, can - to a large extent - discriminate QAnon
vs. non-QAnon supporters on Parler. In particular, we find that user features
are highly discriminatory, followed by content features and linguistic
features.
| no_new_dataset | 0.699704 |
2206.05437 | Yuelin Wang | Yuelin Wang, Kai Yi, Xinliang Liu, Yu Guang Wang, Shi Jin | ACMP: Allen-Cahn Message Passing for Graph Neural Networks with Particle
Phase Transition | 26 pages, 5 figures. NeurIPS 2022 Workshop on GLFrontiers (Oral).
ICLR 2023 (Spotlight) | null | null | null | cs.LG cs.AI math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Neural message passing is a basic feature extraction unit for
graph-structured data considering neighboring node features in network
propagation from one layer to the next. We model such process by an interacting
particle system with attractive and repulsive forces and the Allen-Cahn force
arising in the modeling of phase transition. The dynamics of the system is a
reaction-diffusion process which can separate particles without blowing up.
This induces an Allen-Cahn message passing (ACMP) for graph neural networks
where the numerical iteration for the particle system solution constitutes the
message passing propagation. ACMP which has a simple implementation with a
neural ODE solver can propel the network depth up to one hundred of layers with
theoretically proven strictly positive lower bound of the Dirichlet energy. It
thus provides a deep model of GNNs circumventing the common GNN problem of
oversmoothing. GNNs with ACMP achieve state of the art performance for
real-world node classification tasks on both homophilic and heterophilic
datasets.
| [
{
"version": "v1",
"created": "Sat, 11 Jun 2022 06:26:12 GMT"
},
{
"version": "v2",
"created": "Sun, 30 Oct 2022 08:11:58 GMT"
},
{
"version": "v3",
"created": "Mon, 24 Apr 2023 01:52:14 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Wang",
"Yuelin",
""
],
[
"Yi",
"Kai",
""
],
[
"Liu",
"Xinliang",
""
],
[
"Wang",
"Yu Guang",
""
],
[
"Jin",
"Shi",
""
]
] | TITLE: ACMP: Allen-Cahn Message Passing for Graph Neural Networks with Particle
Phase Transition
ABSTRACT: Neural message passing is a basic feature extraction unit for
graph-structured data considering neighboring node features in network
propagation from one layer to the next. We model such process by an interacting
particle system with attractive and repulsive forces and the Allen-Cahn force
arising in the modeling of phase transition. The dynamics of the system is a
reaction-diffusion process which can separate particles without blowing up.
This induces an Allen-Cahn message passing (ACMP) for graph neural networks
where the numerical iteration for the particle system solution constitutes the
message passing propagation. ACMP which has a simple implementation with a
neural ODE solver can propel the network depth up to one hundred of layers with
theoretically proven strictly positive lower bound of the Dirichlet energy. It
thus provides a deep model of GNNs circumventing the common GNN problem of
oversmoothing. GNNs with ACMP achieve state of the art performance for
real-world node classification tasks on both homophilic and heterophilic
datasets.
| no_new_dataset | 0.71103 |
2206.14486 | Robert Geirhos | Ben Sorscher, Robert Geirhos, Shashank Shekhar, Surya Ganguli, Ari S.
Morcos | Beyond neural scaling laws: beating power law scaling via data pruning | Outstanding Paper Award @ NeurIPS 2022. Added github link to metric
scores | null | null | null | cs.LG cs.AI cs.CV stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Widely observed neural scaling laws, in which error falls off as a power of
the training set size, model size, or both, have driven substantial performance
improvements in deep learning. However, these improvements through scaling
alone require considerable costs in compute and energy. Here we focus on the
scaling of error with dataset size and show how in theory we can break beyond
power law scaling and potentially even reduce it to exponential scaling instead
if we have access to a high-quality data pruning metric that ranks the order in
which training examples should be discarded to achieve any pruned dataset size.
We then test this improved scaling prediction with pruned dataset size
empirically, and indeed observe better than power law scaling in practice on
ResNets trained on CIFAR-10, SVHN, and ImageNet. Next, given the importance of
finding high-quality pruning metrics, we perform the first large-scale
benchmarking study of ten different data pruning metrics on ImageNet. We find
most existing high performing metrics scale poorly to ImageNet, while the best
are computationally intensive and require labels for every image. We therefore
developed a new simple, cheap and scalable self-supervised pruning metric that
demonstrates comparable performance to the best supervised metrics. Overall,
our work suggests that the discovery of good data-pruning metrics may provide a
viable path forward to substantially improved neural scaling laws, thereby
reducing the resource costs of modern deep learning.
| [
{
"version": "v1",
"created": "Wed, 29 Jun 2022 09:20:47 GMT"
},
{
"version": "v2",
"created": "Wed, 3 Aug 2022 15:43:22 GMT"
},
{
"version": "v3",
"created": "Thu, 4 Aug 2022 17:52:18 GMT"
},
{
"version": "v4",
"created": "Tue, 25 Oct 2022 17:42:32 GMT"
},
{
"version": "v5",
"created": "Tue, 15 Nov 2022 19:14:20 GMT"
},
{
"version": "v6",
"created": "Fri, 21 Apr 2023 20:14:07 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Sorscher",
"Ben",
""
],
[
"Geirhos",
"Robert",
""
],
[
"Shekhar",
"Shashank",
""
],
[
"Ganguli",
"Surya",
""
],
[
"Morcos",
"Ari S.",
""
]
] | TITLE: Beyond neural scaling laws: beating power law scaling via data pruning
ABSTRACT: Widely observed neural scaling laws, in which error falls off as a power of
the training set size, model size, or both, have driven substantial performance
improvements in deep learning. However, these improvements through scaling
alone require considerable costs in compute and energy. Here we focus on the
scaling of error with dataset size and show how in theory we can break beyond
power law scaling and potentially even reduce it to exponential scaling instead
if we have access to a high-quality data pruning metric that ranks the order in
which training examples should be discarded to achieve any pruned dataset size.
We then test this improved scaling prediction with pruned dataset size
empirically, and indeed observe better than power law scaling in practice on
ResNets trained on CIFAR-10, SVHN, and ImageNet. Next, given the importance of
finding high-quality pruning metrics, we perform the first large-scale
benchmarking study of ten different data pruning metrics on ImageNet. We find
most existing high performing metrics scale poorly to ImageNet, while the best
are computationally intensive and require labels for every image. We therefore
developed a new simple, cheap and scalable self-supervised pruning metric that
demonstrates comparable performance to the best supervised metrics. Overall,
our work suggests that the discovery of good data-pruning metrics may provide a
viable path forward to substantially improved neural scaling laws, thereby
reducing the resource costs of modern deep learning.
| no_new_dataset | 0.709453 |
2207.02727 | Yiting Dong | Yiting Dong, Dongcheng Zhao, Yang Li, Yi Zeng | An Unsupervised STDP-based Spiking Neural Network Inspired By
Biologically Plausible Learning Rules and Connections | null | null | null | null | cs.NE | http://creativecommons.org/licenses/by/4.0/ | The backpropagation algorithm has promoted the rapid development of deep
learning, but it relies on a large amount of labeled data and still has a large
gap with how humans learn. The human brain can quickly learn various conceptual
knowledge in a self-organized and unsupervised manner, accomplished through
coordinating various learning rules and structures in the human brain.
Spike-timing-dependent plasticity (STDP) is a general learning rule in the
brain, but spiking neural networks (SNNs) trained with STDP alone is
inefficient and perform poorly. In this paper, taking inspiration from
short-term synaptic plasticity, we design an adaptive synaptic filter and
introduce the adaptive spiking threshold as the neuron plasticity to enrich the
representation ability of SNNs. We also introduce an adaptive lateral
inhibitory connection to adjust the spikes balance dynamically to help the
network learn richer features. To speed up and stabilize the training of
unsupervised spiking neural networks, we design a samples temporal batch STDP
(STB-STDP), which updates weights based on multiple samples and moments. By
integrating the above three adaptive mechanisms and STB-STDP, our model greatly
accelerates the training of unsupervised spiking neural networks and improves
the performance of unsupervised SNNs on complex tasks. Our model achieves the
current state-of-the-art performance of unsupervised STDP-based SNNs in the
MNIST and FashionMNIST datasets. Further, we tested on the more complex CIFAR10
dataset, and the results fully illustrate the superiority of our algorithm. Our
model is also the first work to apply unsupervised STDP-based SNNs to CIFAR10.
At the same time, in the small-sample learning scenario, it will far exceed the
supervised ANN using the same structure.
| [
{
"version": "v1",
"created": "Wed, 6 Jul 2022 14:53:32 GMT"
},
{
"version": "v2",
"created": "Sat, 22 Apr 2023 05:44:22 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Dong",
"Yiting",
""
],
[
"Zhao",
"Dongcheng",
""
],
[
"Li",
"Yang",
""
],
[
"Zeng",
"Yi",
""
]
] | TITLE: An Unsupervised STDP-based Spiking Neural Network Inspired By
Biologically Plausible Learning Rules and Connections
ABSTRACT: The backpropagation algorithm has promoted the rapid development of deep
learning, but it relies on a large amount of labeled data and still has a large
gap with how humans learn. The human brain can quickly learn various conceptual
knowledge in a self-organized and unsupervised manner, accomplished through
coordinating various learning rules and structures in the human brain.
Spike-timing-dependent plasticity (STDP) is a general learning rule in the
brain, but spiking neural networks (SNNs) trained with STDP alone is
inefficient and perform poorly. In this paper, taking inspiration from
short-term synaptic plasticity, we design an adaptive synaptic filter and
introduce the adaptive spiking threshold as the neuron plasticity to enrich the
representation ability of SNNs. We also introduce an adaptive lateral
inhibitory connection to adjust the spikes balance dynamically to help the
network learn richer features. To speed up and stabilize the training of
unsupervised spiking neural networks, we design a samples temporal batch STDP
(STB-STDP), which updates weights based on multiple samples and moments. By
integrating the above three adaptive mechanisms and STB-STDP, our model greatly
accelerates the training of unsupervised spiking neural networks and improves
the performance of unsupervised SNNs on complex tasks. Our model achieves the
current state-of-the-art performance of unsupervised STDP-based SNNs in the
MNIST and FashionMNIST datasets. Further, we tested on the more complex CIFAR10
dataset, and the results fully illustrate the superiority of our algorithm. Our
model is also the first work to apply unsupervised STDP-based SNNs to CIFAR10.
At the same time, in the small-sample learning scenario, it will far exceed the
supervised ANN using the same structure.
| no_new_dataset | 0.709604 |
2207.05286 | Jayaraman J. Thiagarajan | Vivek Narayanaswamy, Yamen Mubarka, Rushil Anirudh, Deepta Rajan,
Andreas Spanias and Jayaraman J. Thiagarajan | Know Your Space: Inlier and Outlier Construction for Calibrating Medical
OOD Detectors | null | null | null | null | cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We focus on the problem of producing well-calibrated out-of-distribution
(OOD) detectors, in order to enable safe deployment of medical image
classifiers. Motivated by the difficulty of curating suitable calibration
datasets, synthetic augmentations have become highly prevalent for
inlier/outlier specification. While there have been rapid advances in data
augmentation techniques, this paper makes a striking finding that the space in
which the inliers and outliers are synthesized, in addition to the type of
augmentation, plays a critical role in calibrating OOD detectors. Using the
popular energy-based OOD detection framework, we find that the optimal protocol
is to synthesize latent-space inliers along with diverse pixel-space outliers.
Based on empirical studies with multiple medical imaging benchmarks, we
demonstrate that our approach consistently leads to superior OOD detection
($15\% - 35\%$ in AUROC) over the state-of-the-art in a variety of open-set
recognition settings.
| [
{
"version": "v1",
"created": "Tue, 12 Jul 2022 03:42:05 GMT"
},
{
"version": "v2",
"created": "Sat, 22 Apr 2023 15:31:55 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Narayanaswamy",
"Vivek",
""
],
[
"Mubarka",
"Yamen",
""
],
[
"Anirudh",
"Rushil",
""
],
[
"Rajan",
"Deepta",
""
],
[
"Spanias",
"Andreas",
""
],
[
"Thiagarajan",
"Jayaraman J.",
""
]
] | TITLE: Know Your Space: Inlier and Outlier Construction for Calibrating Medical
OOD Detectors
ABSTRACT: We focus on the problem of producing well-calibrated out-of-distribution
(OOD) detectors, in order to enable safe deployment of medical image
classifiers. Motivated by the difficulty of curating suitable calibration
datasets, synthetic augmentations have become highly prevalent for
inlier/outlier specification. While there have been rapid advances in data
augmentation techniques, this paper makes a striking finding that the space in
which the inliers and outliers are synthesized, in addition to the type of
augmentation, plays a critical role in calibrating OOD detectors. Using the
popular energy-based OOD detection framework, we find that the optimal protocol
is to synthesize latent-space inliers along with diverse pixel-space outliers.
Based on empirical studies with multiple medical imaging benchmarks, we
demonstrate that our approach consistently leads to superior OOD detection
($15\% - 35\%$ in AUROC) over the state-of-the-art in a variety of open-set
recognition settings.
| no_new_dataset | 0.711443 |
2207.05753 | \'Alvaro L\'opez Garc\'ia | Ignacio Heredia Cacha, Judith Sainz-Pardo D\'iaz, Mar\'ia Castrillo
Melguizo, \'Alvaro L\'opez Garc\'ia | Forecasting COVID-19 spreading trough an ensemble of classical and
machine learning models: Spain's case study | null | null | 10.1038/s41598-023-33795-8 | null | cs.LG cs.AI | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In this work we evaluate the applicability of an ensemble of population
models and machine learning models to predict the near future evolution of the
COVID-19 pandemic, with a particular use case in Spain. We rely solely in open
and public datasets, fusing incidence, vaccination, human mobility and weather
data to feed our machine learning models (Random Forest, Gradient Boosting,
k-Nearest Neighbours and Kernel Ridge Regression). We use the incidence data to
adjust classic population models (Gompertz, Logistic, Richards, Bertalanffy) in
order to be able to better capture the trend of the data. We then ensemble
these two families of models in order to obtain a more robust and accurate
prediction. Furthermore, we have observed an improvement in the predictions
obtained with machine learning models as we add new features (vaccines,
mobility, climatic conditions), analyzing the importance of each of them using
Shapley Additive Explanation values. As in any other modelling work, data and
predictions quality have several limitations and therefore they must be seen
from a critical standpoint, as we discuss in the text. Our work concludes that
the ensemble use of these models improves the individual predictions (using
only machine learning models or only population models) and can be applied,
with caution, in cases when compartmental models cannot be utilized due to the
lack of relevant data.
| [
{
"version": "v1",
"created": "Tue, 12 Jul 2022 08:16:44 GMT"
},
{
"version": "v2",
"created": "Fri, 12 Aug 2022 08:36:39 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Cacha",
"Ignacio Heredia",
""
],
[
"Díaz",
"Judith Sainz-Pardo",
""
],
[
"Melguizo",
"María Castrillo",
""
],
[
"García",
"Álvaro López",
""
]
] | TITLE: Forecasting COVID-19 spreading trough an ensemble of classical and
machine learning models: Spain's case study
ABSTRACT: In this work we evaluate the applicability of an ensemble of population
models and machine learning models to predict the near future evolution of the
COVID-19 pandemic, with a particular use case in Spain. We rely solely in open
and public datasets, fusing incidence, vaccination, human mobility and weather
data to feed our machine learning models (Random Forest, Gradient Boosting,
k-Nearest Neighbours and Kernel Ridge Regression). We use the incidence data to
adjust classic population models (Gompertz, Logistic, Richards, Bertalanffy) in
order to be able to better capture the trend of the data. We then ensemble
these two families of models in order to obtain a more robust and accurate
prediction. Furthermore, we have observed an improvement in the predictions
obtained with machine learning models as we add new features (vaccines,
mobility, climatic conditions), analyzing the importance of each of them using
Shapley Additive Explanation values. As in any other modelling work, data and
predictions quality have several limitations and therefore they must be seen
from a critical standpoint, as we discuss in the text. Our work concludes that
the ensemble use of these models improves the individual predictions (using
only machine learning models or only population models) and can be applied,
with caution, in cases when compartmental models cannot be utilized due to the
lack of relevant data.
| no_new_dataset | 0.709856 |
2207.12710 | Christoffer Loeffler | Christoffer Loeffler, Kion Fallah, Stefano Fenu, Dario Zanca, Bjoern
Eskofier, Christopher John Rozell, Christopher Mutschler | Active Learning of Ordinal Embeddings: A User Study on Football Data | 23 pages, 17 figures | Transactions on Machine Learning Research 04/2023
https://openreview.net/forum?id=oq3tx5kinu | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Humans innately measure distance between instances in an unlabeled dataset
using an unknown similarity function. Distance metrics can only serve as proxy
for similarity in information retrieval of similar instances. Learning a good
similarity function from human annotations improves the quality of retrievals.
This work uses deep metric learning to learn these user-defined similarity
functions from few annotations for a large football trajectory dataset. We
adapt an entropy-based active learning method with recent work from triplet
mining to collect easy-to-answer but still informative annotations from human
participants and use them to train a deep convolutional network that
generalizes to unseen samples. Our user study shows that our approach improves
the quality of the information retrieval compared to a previous deep metric
learning approach that relies on a Siamese network. Specifically, we shed light
on the strengths and weaknesses of passive sampling heuristics and active
learners alike by analyzing the participants' response efficacy. To this end,
we collect accuracy, algorithmic time complexity, the participants' fatigue and
time-to-response, qualitative self-assessment and statements, as well as the
effects of mixed-expertise annotators and their consistency on model
performance and transfer-learning.
| [
{
"version": "v1",
"created": "Tue, 26 Jul 2022 07:55:23 GMT"
},
{
"version": "v2",
"created": "Thu, 10 Nov 2022 09:49:18 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Loeffler",
"Christoffer",
""
],
[
"Fallah",
"Kion",
""
],
[
"Fenu",
"Stefano",
""
],
[
"Zanca",
"Dario",
""
],
[
"Eskofier",
"Bjoern",
""
],
[
"Rozell",
"Christopher John",
""
],
[
"Mutschler",
"Christopher",
""
]
] | TITLE: Active Learning of Ordinal Embeddings: A User Study on Football Data
ABSTRACT: Humans innately measure distance between instances in an unlabeled dataset
using an unknown similarity function. Distance metrics can only serve as proxy
for similarity in information retrieval of similar instances. Learning a good
similarity function from human annotations improves the quality of retrievals.
This work uses deep metric learning to learn these user-defined similarity
functions from few annotations for a large football trajectory dataset. We
adapt an entropy-based active learning method with recent work from triplet
mining to collect easy-to-answer but still informative annotations from human
participants and use them to train a deep convolutional network that
generalizes to unseen samples. Our user study shows that our approach improves
the quality of the information retrieval compared to a previous deep metric
learning approach that relies on a Siamese network. Specifically, we shed light
on the strengths and weaknesses of passive sampling heuristics and active
learners alike by analyzing the participants' response efficacy. To this end,
we collect accuracy, algorithmic time complexity, the participants' fatigue and
time-to-response, qualitative self-assessment and statements, as well as the
effects of mixed-expertise annotators and their consistency on model
performance and transfer-learning.
| no_new_dataset | 0.709422 |
2208.11126 | Weili Nie | Zichao Wang, Weili Nie, Zhuoran Qiao, Chaowei Xiao, Richard Baraniuk,
Anima Anandkumar | Retrieval-based Controllable Molecule Generation | ICLR 2023 | null | null | null | q-bio.QM cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Generating new molecules with specified chemical and biological properties
via generative models has emerged as a promising direction for drug discovery.
However, existing methods require extensive training/fine-tuning with a large
dataset, often unavailable in real-world generation tasks. In this work, we
propose a new retrieval-based framework for controllable molecule generation.
We use a small set of exemplar molecules, i.e., those that (partially) satisfy
the design criteria, to steer the pre-trained generative model towards
synthesizing molecules that satisfy the given design criteria. We design a
retrieval mechanism that retrieves and fuses the exemplar molecules with the
input molecule, which is trained by a new self-supervised objective that
predicts the nearest neighbor of the input molecule. We also propose an
iterative refinement process to dynamically update the generated molecules and
retrieval database for better generalization. Our approach is agnostic to the
choice of generative models and requires no task-specific fine-tuning. On
various tasks ranging from simple design criteria to a challenging real-world
scenario for designing lead compounds that bind to the SARS-CoV-2 main
protease, we demonstrate our approach extrapolates well beyond the retrieval
database, and achieves better performance and wider applicability than previous
methods. Code is available at https://github.com/NVlabs/RetMol.
| [
{
"version": "v1",
"created": "Tue, 23 Aug 2022 17:01:16 GMT"
},
{
"version": "v2",
"created": "Fri, 30 Sep 2022 16:47:41 GMT"
},
{
"version": "v3",
"created": "Mon, 24 Apr 2023 17:50:51 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Wang",
"Zichao",
""
],
[
"Nie",
"Weili",
""
],
[
"Qiao",
"Zhuoran",
""
],
[
"Xiao",
"Chaowei",
""
],
[
"Baraniuk",
"Richard",
""
],
[
"Anandkumar",
"Anima",
""
]
] | TITLE: Retrieval-based Controllable Molecule Generation
ABSTRACT: Generating new molecules with specified chemical and biological properties
via generative models has emerged as a promising direction for drug discovery.
However, existing methods require extensive training/fine-tuning with a large
dataset, often unavailable in real-world generation tasks. In this work, we
propose a new retrieval-based framework for controllable molecule generation.
We use a small set of exemplar molecules, i.e., those that (partially) satisfy
the design criteria, to steer the pre-trained generative model towards
synthesizing molecules that satisfy the given design criteria. We design a
retrieval mechanism that retrieves and fuses the exemplar molecules with the
input molecule, which is trained by a new self-supervised objective that
predicts the nearest neighbor of the input molecule. We also propose an
iterative refinement process to dynamically update the generated molecules and
retrieval database for better generalization. Our approach is agnostic to the
choice of generative models and requires no task-specific fine-tuning. On
various tasks ranging from simple design criteria to a challenging real-world
scenario for designing lead compounds that bind to the SARS-CoV-2 main
protease, we demonstrate our approach extrapolates well beyond the retrieval
database, and achieves better performance and wider applicability than previous
methods. Code is available at https://github.com/NVlabs/RetMol.
| no_new_dataset | 0.710672 |
2208.14889 | Jihye Park | Jihye Park, Sunwoo Kim, Soohyun Kim, Seokju Cho, Jaejun Yoo, Youngjung
Uh, Seungryong Kim | LANIT: Language-Driven Image-to-Image Translation for Unlabeled Data | Accepted to CVPR 2023. Project Page:
https://ku-cvlab.github.io/LANIT/ | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | Existing techniques for image-to-image translation commonly have suffered
from two critical problems: heavy reliance on per-sample domain annotation
and/or inability of handling multiple attributes per image. Recent
truly-unsupervised methods adopt clustering approaches to easily provide
per-sample one-hot domain labels. However, they cannot account for the
real-world setting: one sample may have multiple attributes. In addition, the
semantics of the clusters are not easily coupled to the human understanding. To
overcome these, we present a LANguage-driven Image-to-image Translation model,
dubbed LANIT. We leverage easy-to-obtain candidate attributes given in texts
for a dataset: the similarity between images and attributes indicates
per-sample domain labels. This formulation naturally enables multi-hot label so
that users can specify the target domain with a set of attributes in language.
To account for the case that the initial prompts are inaccurate, we also
present prompt learning. We further present domain regularization loss that
enforces translated images be mapped to the corresponding domain. Experiments
on several standard benchmarks demonstrate that LANIT achieves comparable or
superior performance to existing models.
| [
{
"version": "v1",
"created": "Wed, 31 Aug 2022 14:30:00 GMT"
},
{
"version": "v2",
"created": "Thu, 23 Feb 2023 05:39:21 GMT"
},
{
"version": "v3",
"created": "Sat, 4 Mar 2023 09:46:09 GMT"
},
{
"version": "v4",
"created": "Mon, 24 Apr 2023 08:14:41 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Park",
"Jihye",
""
],
[
"Kim",
"Sunwoo",
""
],
[
"Kim",
"Soohyun",
""
],
[
"Cho",
"Seokju",
""
],
[
"Yoo",
"Jaejun",
""
],
[
"Uh",
"Youngjung",
""
],
[
"Kim",
"Seungryong",
""
]
] | TITLE: LANIT: Language-Driven Image-to-Image Translation for Unlabeled Data
ABSTRACT: Existing techniques for image-to-image translation commonly have suffered
from two critical problems: heavy reliance on per-sample domain annotation
and/or inability of handling multiple attributes per image. Recent
truly-unsupervised methods adopt clustering approaches to easily provide
per-sample one-hot domain labels. However, they cannot account for the
real-world setting: one sample may have multiple attributes. In addition, the
semantics of the clusters are not easily coupled to the human understanding. To
overcome these, we present a LANguage-driven Image-to-image Translation model,
dubbed LANIT. We leverage easy-to-obtain candidate attributes given in texts
for a dataset: the similarity between images and attributes indicates
per-sample domain labels. This formulation naturally enables multi-hot label so
that users can specify the target domain with a set of attributes in language.
To account for the case that the initial prompts are inaccurate, we also
present prompt learning. We further present domain regularization loss that
enforces translated images be mapped to the corresponding domain. Experiments
on several standard benchmarks demonstrate that LANIT achieves comparable or
superior performance to existing models.
| no_new_dataset | 0.710597 |
2209.00068 | Chenkai Sun | Chenkai Sun, Tie Xu, ChengXiang Zhai, Heng Ji | Incorporating Task-specific Concept Knowledge into Script Learning | null | null | null | null | cs.CL cs.LG | http://creativecommons.org/licenses/by/4.0/ | In this paper, we present Tetris, a new task of Goal-Oriented Script
Completion. Unlike previous work, it considers a more realistic and general
setting, where the input includes not only the goal but also additional user
context, including preferences and history. To address this problem, we propose
a novel approach, which uses two techniques to improve performance: (1) concept
prompting, and (2) script-oriented contrastive learning that addresses step
repetition and hallucination problems. On our WikiHow-based dataset, we find
that both methods improve performance. The dataset, repository, and models will
be publicly available to facilitate further research on this new task.
| [
{
"version": "v1",
"created": "Wed, 31 Aug 2022 18:55:22 GMT"
},
{
"version": "v2",
"created": "Sat, 21 Jan 2023 01:23:00 GMT"
},
{
"version": "v3",
"created": "Mon, 24 Apr 2023 00:48:03 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Sun",
"Chenkai",
""
],
[
"Xu",
"Tie",
""
],
[
"Zhai",
"ChengXiang",
""
],
[
"Ji",
"Heng",
""
]
] | TITLE: Incorporating Task-specific Concept Knowledge into Script Learning
ABSTRACT: In this paper, we present Tetris, a new task of Goal-Oriented Script
Completion. Unlike previous work, it considers a more realistic and general
setting, where the input includes not only the goal but also additional user
context, including preferences and history. To address this problem, we propose
a novel approach, which uses two techniques to improve performance: (1) concept
prompting, and (2) script-oriented contrastive learning that addresses step
repetition and hallucination problems. On our WikiHow-based dataset, we find
that both methods improve performance. The dataset, repository, and models will
be publicly available to facilitate further research on this new task.
| new_dataset | 0.711638 |
2209.15052 | Yahia Zakaria | Yahia Zakaria, Magda Fayek, Mayada Hadhoud | Start Small: Training Controllable Game Level Generators without
Training Data by Learning at Multiple Sizes | 26 pages, 7 tables, 7 figures. Code:
https://github.com/yahiaetman/ms-level-gen | Alexandria Engineering Journal 72C (2023) pp. 479-494 | 10.1016/j.aej.2023.04.019 | null | cs.LG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | A level generator is a tool that generates game levels from noise. Training a
generator without a dataset suffers from feedback sparsity, since it is
unlikely to generate a playable level via random exploration. A common solution
is shaped rewards, which guides the generator to achieve subgoals towards level
playability, but they consume effort to design and require game-specific domain
knowledge. This paper proposes a novel approach to train generators without
datasets or shaped rewards by learning at multiple level sizes starting from
small sizes and up to the desired sizes. The denser feedback at small sizes
negates the need for shaped rewards. Additionally, the generators learn to
build levels at various sizes, including sizes they were not trained for. We
apply our approach to train recurrent auto-regressive generative flow networks
(GFlowNets) for controllable level generation. We also adapt diversity sampling
to be compatible with GFlowNets. The results show that our generators create
diverse playable levels at various sizes for Sokoban, Zelda, and Danger Dave.
When compared with controllable reinforcement learning level generators for
Sokoban, the results show that our generators achieve better controllability
and competitive diversity, while being 9x faster at training and level
generation.
| [
{
"version": "v1",
"created": "Thu, 29 Sep 2022 18:52:54 GMT"
},
{
"version": "v2",
"created": "Fri, 6 Jan 2023 14:36:23 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Zakaria",
"Yahia",
""
],
[
"Fayek",
"Magda",
""
],
[
"Hadhoud",
"Mayada",
""
]
] | TITLE: Start Small: Training Controllable Game Level Generators without
Training Data by Learning at Multiple Sizes
ABSTRACT: A level generator is a tool that generates game levels from noise. Training a
generator without a dataset suffers from feedback sparsity, since it is
unlikely to generate a playable level via random exploration. A common solution
is shaped rewards, which guides the generator to achieve subgoals towards level
playability, but they consume effort to design and require game-specific domain
knowledge. This paper proposes a novel approach to train generators without
datasets or shaped rewards by learning at multiple level sizes starting from
small sizes and up to the desired sizes. The denser feedback at small sizes
negates the need for shaped rewards. Additionally, the generators learn to
build levels at various sizes, including sizes they were not trained for. We
apply our approach to train recurrent auto-regressive generative flow networks
(GFlowNets) for controllable level generation. We also adapt diversity sampling
to be compatible with GFlowNets. The results show that our generators create
diverse playable levels at various sizes for Sokoban, Zelda, and Danger Dave.
When compared with controllable reinforcement learning level generators for
Sokoban, the results show that our generators achieve better controllability
and competitive diversity, while being 9x faster at training and level
generation.
| no_new_dataset | 0.711224 |
2210.01338 | Chongyang Gao | Xu Yang and Hanwang Zhang and Chongyang Gao and Jianfei Cai | Learning to Collocate Visual-Linguistic Neural Modules for Image
Captioning | Accepted to IJCV. Codes are available at
https://github.com/GCYZSL/CVLMN | null | null | null | cs.CV cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Humans tend to decompose a sentence into different parts like \textsc{sth do
sth at someplace} and then fill each part with certain content. Inspired by
this, we follow the \textit{principle of modular design} to propose a novel
image captioner: learning to Collocate Visual-Linguistic Neural Modules
(CVLNM). Unlike the \re{widely used} neural module networks in VQA, where the
language (\ie, question) is fully observable, \re{the task of collocating
visual-linguistic modules is more challenging.} This is because the language is
only partially observable, for which we need to dynamically collocate the
modules during the process of image captioning. To sum up, we make the
following technical contributions to design and train our CVLNM: 1)
\textit{distinguishable module design} -- \re{four modules in the encoder}
including one linguistic module for function words and three visual modules for
different content words (\ie, noun, adjective, and verb) and another linguistic
one in the decoder for commonsense reasoning, 2) a self-attention based
\textit{module controller} for robustifying the visual reasoning, 3) a
part-of-speech based \textit{syntax loss} imposed on the module controller for
further regularizing the training of our CVLNM. Extensive experiments on the
MS-COCO dataset show that our CVLNM is more effective, \eg, achieving a new
state-of-the-art 129.5 CIDEr-D, and more robust, \eg, being less likely to
overfit to dataset bias and suffering less when fewer training samples are
available. Codes are available at \url{https://github.com/GCYZSL/CVLMN}
| [
{
"version": "v1",
"created": "Tue, 4 Oct 2022 03:09:50 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Apr 2023 02:27:07 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Yang",
"Xu",
""
],
[
"Zhang",
"Hanwang",
""
],
[
"Gao",
"Chongyang",
""
],
[
"Cai",
"Jianfei",
""
]
] | TITLE: Learning to Collocate Visual-Linguistic Neural Modules for Image
Captioning
ABSTRACT: Humans tend to decompose a sentence into different parts like \textsc{sth do
sth at someplace} and then fill each part with certain content. Inspired by
this, we follow the \textit{principle of modular design} to propose a novel
image captioner: learning to Collocate Visual-Linguistic Neural Modules
(CVLNM). Unlike the \re{widely used} neural module networks in VQA, where the
language (\ie, question) is fully observable, \re{the task of collocating
visual-linguistic modules is more challenging.} This is because the language is
only partially observable, for which we need to dynamically collocate the
modules during the process of image captioning. To sum up, we make the
following technical contributions to design and train our CVLNM: 1)
\textit{distinguishable module design} -- \re{four modules in the encoder}
including one linguistic module for function words and three visual modules for
different content words (\ie, noun, adjective, and verb) and another linguistic
one in the decoder for commonsense reasoning, 2) a self-attention based
\textit{module controller} for robustifying the visual reasoning, 3) a
part-of-speech based \textit{syntax loss} imposed on the module controller for
further regularizing the training of our CVLNM. Extensive experiments on the
MS-COCO dataset show that our CVLNM is more effective, \eg, achieving a new
state-of-the-art 129.5 CIDEr-D, and more robust, \eg, being less likely to
overfit to dataset bias and suffering less when fewer training samples are
available. Codes are available at \url{https://github.com/GCYZSL/CVLMN}
| no_new_dataset | 0.709856 |
2210.08065 | Lev Grossman | Lev Grossman and Brian Plancher | Just Round: Quantized Observation Spaces Enable Memory Efficient
Learning of Dynamic Locomotion | Accepted to ICRA 2023 | null | null | null | cs.RO cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deep reinforcement learning (DRL) is one of the most powerful tools for
synthesizing complex robotic behaviors. But training DRL models is incredibly
compute and memory intensive, requiring large training datasets and replay
buffers to achieve performant results. This poses a challenge for the next
generation of field robots that will need to learn on the edge to adapt to
their environment. In this paper, we begin to address this issue through
observation space quantization. We evaluate our approach using four simulated
robot locomotion tasks and two state-of-the-art DRL algorithms, the on-policy
Proximal Policy Optimization (PPO) and off-policy Soft Actor-Critic (SAC) and
find that observation space quantization reduces overall memory costs by as
much as 4.2x without impacting learning performance.
| [
{
"version": "v1",
"created": "Fri, 14 Oct 2022 19:14:47 GMT"
},
{
"version": "v2",
"created": "Sat, 22 Apr 2023 16:33:15 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Grossman",
"Lev",
""
],
[
"Plancher",
"Brian",
""
]
] | TITLE: Just Round: Quantized Observation Spaces Enable Memory Efficient
Learning of Dynamic Locomotion
ABSTRACT: Deep reinforcement learning (DRL) is one of the most powerful tools for
synthesizing complex robotic behaviors. But training DRL models is incredibly
compute and memory intensive, requiring large training datasets and replay
buffers to achieve performant results. This poses a challenge for the next
generation of field robots that will need to learn on the edge to adapt to
their environment. In this paper, we begin to address this issue through
observation space quantization. We evaluate our approach using four simulated
robot locomotion tasks and two state-of-the-art DRL algorithms, the on-policy
Proximal Policy Optimization (PPO) and off-policy Soft Actor-Critic (SAC) and
find that observation space quantization reduces overall memory costs by as
much as 4.2x without impacting learning performance.
| no_new_dataset | 0.709636 |
2210.15496 | Md Ferdous Pervej | Md Ferdous Pervej, Richeng Jin and Huaiyu Dai | Resource Constrained Vehicular Edge Federated Learning with Highly
Mobile Connected Vehicles | Accepted for publication in the IEEE Journal on Selected Areas in
Communications (JSAC) | null | null | null | eess.SY cs.LG cs.NI cs.SY | http://creativecommons.org/licenses/by/4.0/ | This paper proposes a vehicular edge federated learning (VEFL) solution,
where an edge server leverages highly mobile connected vehicles' (CVs') onboard
central processing units (CPUs) and local datasets to train a global model.
Convergence analysis reveals that the VEFL training loss depends on the
successful receptions of the CVs' trained models over the intermittent
vehicle-to-infrastructure (V2I) wireless links. Owing to high mobility, in the
full device participation case (FDPC), the edge server aggregates client model
parameters based on a weighted combination according to the CVs' dataset sizes
and sojourn periods, while it selects a subset of CVs in the partial device
participation case (PDPC). We then devise joint VEFL and radio access
technology (RAT) parameters optimization problems under delay, energy and cost
constraints to maximize the probability of successful reception of the locally
trained models. Considering that the optimization problem is NP-hard, we
decompose it into a VEFL parameter optimization sub-problem, given the
estimated worst-case sojourn period, delay and energy expense, and an online
RAT parameter optimization sub-problem. Finally, extensive simulations are
conducted to validate the effectiveness of the proposed solutions with a
practical 5G new radio (5G-NR) RAT under a realistic microscopic mobility
model.
| [
{
"version": "v1",
"created": "Thu, 27 Oct 2022 14:33:06 GMT"
},
{
"version": "v2",
"created": "Mon, 20 Feb 2023 05:13:20 GMT"
},
{
"version": "v3",
"created": "Thu, 6 Apr 2023 16:37:52 GMT"
},
{
"version": "v4",
"created": "Sun, 23 Apr 2023 22:58:37 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Pervej",
"Md Ferdous",
""
],
[
"Jin",
"Richeng",
""
],
[
"Dai",
"Huaiyu",
""
]
] | TITLE: Resource Constrained Vehicular Edge Federated Learning with Highly
Mobile Connected Vehicles
ABSTRACT: This paper proposes a vehicular edge federated learning (VEFL) solution,
where an edge server leverages highly mobile connected vehicles' (CVs') onboard
central processing units (CPUs) and local datasets to train a global model.
Convergence analysis reveals that the VEFL training loss depends on the
successful receptions of the CVs' trained models over the intermittent
vehicle-to-infrastructure (V2I) wireless links. Owing to high mobility, in the
full device participation case (FDPC), the edge server aggregates client model
parameters based on a weighted combination according to the CVs' dataset sizes
and sojourn periods, while it selects a subset of CVs in the partial device
participation case (PDPC). We then devise joint VEFL and radio access
technology (RAT) parameters optimization problems under delay, energy and cost
constraints to maximize the probability of successful reception of the locally
trained models. Considering that the optimization problem is NP-hard, we
decompose it into a VEFL parameter optimization sub-problem, given the
estimated worst-case sojourn period, delay and energy expense, and an online
RAT parameter optimization sub-problem. Finally, extensive simulations are
conducted to validate the effectiveness of the proposed solutions with a
practical 5G new radio (5G-NR) RAT under a realistic microscopic mobility
model.
| no_new_dataset | 0.708862 |
2210.15578 | Dong Yang | Dong Yang, Peijun Qing, Yang Li, Haonan Lu, Xiaodong Lin | GammaE: Gamma Embeddings for Logical Queries on Knowledge Graphs | null | null | null | null | cs.LG cs.AI cs.LO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Embedding knowledge graphs (KGs) for multi-hop logical reasoning is a
challenging problem due to massive and complicated structures in many KGs.
Recently, many promising works projected entities and queries into a geometric
space to efficiently find answers. However, it remains challenging to model the
negation and union operator. The negation operator has no strict boundaries,
which generates overlapped embeddings and leads to obtaining ambiguous answers.
An additional limitation is that the union operator is non-closure, which
undermines the model to handle a series of union operators. To address these
problems, we propose a novel probabilistic embedding model, namely Gamma
Embeddings (GammaE), for encoding entities and queries to answer different
types of FOL queries on KGs. We utilize the linear property and strong boundary
support of the Gamma distribution to capture more features of entities and
queries, which dramatically reduces model uncertainty. Furthermore, GammaE
implements the Gamma mixture method to design the closed union operator. The
performance of GammaE is validated on three large logical query datasets.
Experimental results show that GammaE significantly outperforms
state-of-the-art models on public benchmarks.
| [
{
"version": "v1",
"created": "Thu, 27 Oct 2022 16:15:11 GMT"
},
{
"version": "v2",
"created": "Sat, 22 Apr 2023 14:54:08 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Yang",
"Dong",
""
],
[
"Qing",
"Peijun",
""
],
[
"Li",
"Yang",
""
],
[
"Lu",
"Haonan",
""
],
[
"Lin",
"Xiaodong",
""
]
] | TITLE: GammaE: Gamma Embeddings for Logical Queries on Knowledge Graphs
ABSTRACT: Embedding knowledge graphs (KGs) for multi-hop logical reasoning is a
challenging problem due to massive and complicated structures in many KGs.
Recently, many promising works projected entities and queries into a geometric
space to efficiently find answers. However, it remains challenging to model the
negation and union operator. The negation operator has no strict boundaries,
which generates overlapped embeddings and leads to obtaining ambiguous answers.
An additional limitation is that the union operator is non-closure, which
undermines the model to handle a series of union operators. To address these
problems, we propose a novel probabilistic embedding model, namely Gamma
Embeddings (GammaE), for encoding entities and queries to answer different
types of FOL queries on KGs. We utilize the linear property and strong boundary
support of the Gamma distribution to capture more features of entities and
queries, which dramatically reduces model uncertainty. Furthermore, GammaE
implements the Gamma mixture method to design the closed union operator. The
performance of GammaE is validated on three large logical query datasets.
Experimental results show that GammaE significantly outperforms
state-of-the-art models on public benchmarks.
| no_new_dataset | 0.708603 |
2210.15741 | Yassine Naji | Yassine Naji, Aleksandr Setkov, Ang\'elique Loesch, Mich\`ele
Gouiff\`es, Romaric Audigier | Spatio-temporal predictive tasks for abnormal event detection in videos | Accepted at the 18th IEEE International Conference on Advanced Video
and Signal Based Surveillance (AVSS), 2022 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Abnormal event detection in videos is a challenging problem, partly due to
the multiplicity of abnormal patterns and the lack of their corresponding
annotations. In this paper, we propose new constrained pretext tasks to learn
object level normality patterns. Our approach consists in learning a mapping
between down-scaled visual queries and their corresponding normal appearance
and motion characteristics at the original resolution. The proposed tasks are
more challenging than reconstruction and future frame prediction tasks which
are widely used in the literature, since our model learns to jointly predict
spatial and temporal features rather than reconstructing them. We believe that
more constrained pretext tasks induce a better learning of normality patterns.
Experiments on several benchmark datasets demonstrate the effectiveness of our
approach to localize and track anomalies as it outperforms or reaches the
current state-of-the-art on spatio-temporal evaluation metrics.
| [
{
"version": "v1",
"created": "Thu, 27 Oct 2022 19:45:12 GMT"
},
{
"version": "v2",
"created": "Sun, 23 Apr 2023 16:35:35 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Naji",
"Yassine",
""
],
[
"Setkov",
"Aleksandr",
""
],
[
"Loesch",
"Angélique",
""
],
[
"Gouiffès",
"Michèle",
""
],
[
"Audigier",
"Romaric",
""
]
] | TITLE: Spatio-temporal predictive tasks for abnormal event detection in videos
ABSTRACT: Abnormal event detection in videos is a challenging problem, partly due to
the multiplicity of abnormal patterns and the lack of their corresponding
annotations. In this paper, we propose new constrained pretext tasks to learn
object level normality patterns. Our approach consists in learning a mapping
between down-scaled visual queries and their corresponding normal appearance
and motion characteristics at the original resolution. The proposed tasks are
more challenging than reconstruction and future frame prediction tasks which
are widely used in the literature, since our model learns to jointly predict
spatial and temporal features rather than reconstructing them. We believe that
more constrained pretext tasks induce a better learning of normality patterns.
Experiments on several benchmark datasets demonstrate the effectiveness of our
approach to localize and track anomalies as it outperforms or reaches the
current state-of-the-art on spatio-temporal evaluation metrics.
| no_new_dataset | 0.711619 |
2211.00381 | Maliheh Izadi | Maliheh Izadi, Pooya Rostami Mazrae, Tom Mens, Arie van Deursen | An Empirical Study on Data Leakage and Generalizability of Link
Prediction Models for Issues and Commits | null | null | null | null | cs.SE cs.LG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | To enhance documentation and maintenance practices, developers conventionally
establish links between related software artifacts manually. Empirical research
has revealed that developers frequently overlook this practice, resulting in
significant information loss. To address this issue, automatic link recovery
techniques have been proposed. However, these approaches primarily focused on
improving prediction accuracy on randomly-split datasets, with limited
attention given to the impact of data leakage and the generalizability of the
predictive models. LinkFormer seeks to address these limitations. Our approach
not only preserves and improves the accuracy of existing predictions but also
enhances their alignment with real-world settings and their generalizability.
First, to better utilize contextual information for prediction, we employ the
Transformer architecture and fine-tune multiple pre-trained models on both
textual and metadata information of issues and commits. Next, to gauge the
effect of time on model performance, we employ two splitting policies during
both the training and testing phases; randomly- and temporally-split datasets.
Finally, in pursuit of a generic model that can demonstrate high performance
across a range of projects, we undertake additional fine-tuning of LinkFormer
within two distinct transfer-learning settings. Our findings support that to
simulate real-world scenarios effectively, researchers must maintain the
temporal flow of data when training models. Furthermore, the results
demonstrate that LinkFormer outperforms existing methodologies by a significant
margin, achieving a 48% improvement in F1-measure within a project-based
setting. Finally, the performance of LinkFormer in the cross-project setting is
comparable to its average performance within the project-based scenario.
| [
{
"version": "v1",
"created": "Tue, 1 Nov 2022 10:54:26 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Apr 2023 11:01:55 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Izadi",
"Maliheh",
""
],
[
"Mazrae",
"Pooya Rostami",
""
],
[
"Mens",
"Tom",
""
],
[
"van Deursen",
"Arie",
""
]
] | TITLE: An Empirical Study on Data Leakage and Generalizability of Link
Prediction Models for Issues and Commits
ABSTRACT: To enhance documentation and maintenance practices, developers conventionally
establish links between related software artifacts manually. Empirical research
has revealed that developers frequently overlook this practice, resulting in
significant information loss. To address this issue, automatic link recovery
techniques have been proposed. However, these approaches primarily focused on
improving prediction accuracy on randomly-split datasets, with limited
attention given to the impact of data leakage and the generalizability of the
predictive models. LinkFormer seeks to address these limitations. Our approach
not only preserves and improves the accuracy of existing predictions but also
enhances their alignment with real-world settings and their generalizability.
First, to better utilize contextual information for prediction, we employ the
Transformer architecture and fine-tune multiple pre-trained models on both
textual and metadata information of issues and commits. Next, to gauge the
effect of time on model performance, we employ two splitting policies during
both the training and testing phases; randomly- and temporally-split datasets.
Finally, in pursuit of a generic model that can demonstrate high performance
across a range of projects, we undertake additional fine-tuning of LinkFormer
within two distinct transfer-learning settings. Our findings support that to
simulate real-world scenarios effectively, researchers must maintain the
temporal flow of data when training models. Furthermore, the results
demonstrate that LinkFormer outperforms existing methodologies by a significant
margin, achieving a 48% improvement in F1-measure within a project-based
setting. Finally, the performance of LinkFormer in the cross-project setting is
comparable to its average performance within the project-based scenario.
| no_new_dataset | 0.709044 |
2211.11082 | Zhengqi Li | Zhengqi Li, Qianqian Wang, Forrester Cole, Richard Tucker, Noah
Snavely | DynIBaR: Neural Dynamic Image-Based Rendering | Award Candidate, CVPR 2023 Project page: dynibar.github.io | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We address the problem of synthesizing novel views from a monocular video
depicting a complex dynamic scene. State-of-the-art methods based on temporally
varying Neural Radiance Fields (aka dynamic NeRFs) have shown impressive
results on this task. However, for long videos with complex object motions and
uncontrolled camera trajectories, these methods can produce blurry or
inaccurate renderings, hampering their use in real-world applications. Instead
of encoding the entire dynamic scene within the weights of MLPs, we present a
new approach that addresses these limitations by adopting a volumetric
image-based rendering framework that synthesizes new viewpoints by aggregating
features from nearby views in a scene-motion-aware manner. Our system retains
the advantages of prior methods in its ability to model complex scenes and
view-dependent effects, but also enables synthesizing photo-realistic novel
views from long videos featuring complex scene dynamics with unconstrained
camera trajectories. We demonstrate significant improvements over
state-of-the-art methods on dynamic scene datasets, and also apply our approach
to in-the-wild videos with challenging camera and object motion, where prior
methods fail to produce high-quality renderings. Our project webpage is at
dynibar.github.io.
| [
{
"version": "v1",
"created": "Sun, 20 Nov 2022 20:57:02 GMT"
},
{
"version": "v2",
"created": "Mon, 28 Nov 2022 17:29:18 GMT"
},
{
"version": "v3",
"created": "Mon, 24 Apr 2023 16:42:08 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Li",
"Zhengqi",
""
],
[
"Wang",
"Qianqian",
""
],
[
"Cole",
"Forrester",
""
],
[
"Tucker",
"Richard",
""
],
[
"Snavely",
"Noah",
""
]
] | TITLE: DynIBaR: Neural Dynamic Image-Based Rendering
ABSTRACT: We address the problem of synthesizing novel views from a monocular video
depicting a complex dynamic scene. State-of-the-art methods based on temporally
varying Neural Radiance Fields (aka dynamic NeRFs) have shown impressive
results on this task. However, for long videos with complex object motions and
uncontrolled camera trajectories, these methods can produce blurry or
inaccurate renderings, hampering their use in real-world applications. Instead
of encoding the entire dynamic scene within the weights of MLPs, we present a
new approach that addresses these limitations by adopting a volumetric
image-based rendering framework that synthesizes new viewpoints by aggregating
features from nearby views in a scene-motion-aware manner. Our system retains
the advantages of prior methods in its ability to model complex scenes and
view-dependent effects, but also enables synthesizing photo-realistic novel
views from long videos featuring complex scene dynamics with unconstrained
camera trajectories. We demonstrate significant improvements over
state-of-the-art methods on dynamic scene datasets, and also apply our approach
to in-the-wild videos with challenging camera and object motion, where prior
methods fail to produce high-quality renderings. Our project webpage is at
dynibar.github.io.
| no_new_dataset | 0.712201 |
2211.15279 | Pengwei Yang | Pengwei Yang, Chongyangzi Teng and Jack George Mangos | Establishment of Neural Networks Robust to Label Noise | 11 pages, 7 figures | null | null | null | cs.LG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Label noise is a significant obstacle in deep learning model training. It can
have a considerable impact on the performance of image classification models,
particularly deep neural networks, which are especially susceptible because
they have a strong propensity to memorise noisy labels. In this paper, we have
examined the fundamental concept underlying related label noise approaches. A
transition matrix estimator has been created, and its effectiveness against the
actual transition matrix has been demonstrated. In addition, we examined the
label noise robustness of two convolutional neural network classifiers with
LeNet and AlexNet designs. The two FashionMINIST datasets have revealed the
robustness of both models. We are not efficiently able to demonstrate the
influence of the transition matrix noise correction on robustness enhancements
due to our inability to correctly tune the complex convolutional neural network
model due to time and computing resource constraints. There is a need for
additional effort to fine-tune the neural network model and explore the
precision of the estimated transition model in future research.
| [
{
"version": "v1",
"created": "Mon, 28 Nov 2022 13:07:23 GMT"
},
{
"version": "v2",
"created": "Sun, 4 Dec 2022 05:23:13 GMT"
},
{
"version": "v3",
"created": "Mon, 24 Apr 2023 02:43:38 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Yang",
"Pengwei",
""
],
[
"Teng",
"Chongyangzi",
""
],
[
"Mangos",
"Jack George",
""
]
] | TITLE: Establishment of Neural Networks Robust to Label Noise
ABSTRACT: Label noise is a significant obstacle in deep learning model training. It can
have a considerable impact on the performance of image classification models,
particularly deep neural networks, which are especially susceptible because
they have a strong propensity to memorise noisy labels. In this paper, we have
examined the fundamental concept underlying related label noise approaches. A
transition matrix estimator has been created, and its effectiveness against the
actual transition matrix has been demonstrated. In addition, we examined the
label noise robustness of two convolutional neural network classifiers with
LeNet and AlexNet designs. The two FashionMINIST datasets have revealed the
robustness of both models. We are not efficiently able to demonstrate the
influence of the transition matrix noise correction on robustness enhancements
due to our inability to correctly tune the complex convolutional neural network
model due to time and computing resource constraints. There is a need for
additional effort to fine-tune the neural network model and explore the
precision of the estimated transition model in future research.
| no_new_dataset | 0.708584 |
2212.03475 | Xun Jiao | Ruixuan Wang, Fred Lin, Daniel Moore, Sriram Sankar, Xun Jiao | PyGFI: Analyzing and Enhancing Robustness of Graph Neural Networks
Against Hardware Errors | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Graph neural networks (GNNs) have recently emerged as a promising learning
paradigm in learning graph-structured data and have demonstrated wide success
across various domains such as recommendation systems, social networks, and
electronic design automation (EDA). Like other deep learning (DL) methods, GNNs
are being deployed in sophisticated modern hardware systems, as well as
dedicated accelerators. However, despite the popularity of GNNs and the recent
efforts of bringing GNNs to hardware, the fault tolerance and resilience of
GNNs have generally been overlooked. Inspired by the inherent algorithmic
resilience of DL methods, this paper conducts, for the first time, a
large-scale and empirical study of GNN resilience, aiming to understand the
relationship between hardware faults and GNN accuracy. By developing a
customized fault injection tool on top of PyTorch, we perform extensive fault
injection experiments on various GNN models and application datasets. We
observe that the error resilience of GNN models varies by orders of magnitude
with respect to different models and application datasets. Further, we explore
a low-cost error mitigation mechanism for GNN to enhance its resilience. This
GNN resilience study aims to open up new directions and opportunities for
future GNN accelerator design and architectural optimization.
| [
{
"version": "v1",
"created": "Wed, 7 Dec 2022 06:14:14 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Apr 2023 15:38:27 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Wang",
"Ruixuan",
""
],
[
"Lin",
"Fred",
""
],
[
"Moore",
"Daniel",
""
],
[
"Sankar",
"Sriram",
""
],
[
"Jiao",
"Xun",
""
]
] | TITLE: PyGFI: Analyzing and Enhancing Robustness of Graph Neural Networks
Against Hardware Errors
ABSTRACT: Graph neural networks (GNNs) have recently emerged as a promising learning
paradigm in learning graph-structured data and have demonstrated wide success
across various domains such as recommendation systems, social networks, and
electronic design automation (EDA). Like other deep learning (DL) methods, GNNs
are being deployed in sophisticated modern hardware systems, as well as
dedicated accelerators. However, despite the popularity of GNNs and the recent
efforts of bringing GNNs to hardware, the fault tolerance and resilience of
GNNs have generally been overlooked. Inspired by the inherent algorithmic
resilience of DL methods, this paper conducts, for the first time, a
large-scale and empirical study of GNN resilience, aiming to understand the
relationship between hardware faults and GNN accuracy. By developing a
customized fault injection tool on top of PyTorch, we perform extensive fault
injection experiments on various GNN models and application datasets. We
observe that the error resilience of GNN models varies by orders of magnitude
with respect to different models and application datasets. Further, we explore
a low-cost error mitigation mechanism for GNN to enhance its resilience. This
GNN resilience study aims to open up new directions and opportunities for
future GNN accelerator design and architectural optimization.
| no_new_dataset | 0.706368 |
2212.12061 | Nuno Fachada | Alina Petukhova, Nuno Fachada | MN-DS: A Multilabeled News Dataset for News Articles Hierarchical
Classification | The peer-reviewed version of this paper is published in Data at
https://doi.org/10.3390/data8050074. This version is typeset by the authors
and differs only in pagination and typographical detail | Data, 8(5), 74, 2023 | 10.3390/data8050074 | null | cs.CL cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | This article presents a dataset of 10,917 news articles with hierarchical
news categories collected between 1 January 2019 and 31 December 2019. We
manually labeled the articles based on a hierarchical taxonomy with 17
first-level and 109 second-level categories. This dataset can be used to train
machine learning models for automatically classifying news articles by topic.
This dataset can be helpful for researchers working on news structuring,
classification, and predicting future events based on released news.
| [
{
"version": "v1",
"created": "Thu, 22 Dec 2022 22:27:26 GMT"
},
{
"version": "v2",
"created": "Sun, 19 Mar 2023 12:10:02 GMT"
},
{
"version": "v3",
"created": "Sun, 23 Apr 2023 14:49:44 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Petukhova",
"Alina",
""
],
[
"Fachada",
"Nuno",
""
]
] | TITLE: MN-DS: A Multilabeled News Dataset for News Articles Hierarchical
Classification
ABSTRACT: This article presents a dataset of 10,917 news articles with hierarchical
news categories collected between 1 January 2019 and 31 December 2019. We
manually labeled the articles based on a hierarchical taxonomy with 17
first-level and 109 second-level categories. This dataset can be used to train
machine learning models for automatically classifying news articles by topic.
This dataset can be helpful for researchers working on news structuring,
classification, and predicting future events based on released news.
| new_dataset | 0.711243 |
2301.01395 | Ronak Buch | Ronak Buch | COST of Graph Processing Using Actors | null | null | null | null | cs.DC cs.PF | http://creativecommons.org/licenses/by-sa/4.0/ | Graph processing is an increasingly important domain of computer science,
with applications in data and network analysis, among others. Target graphs in
these applications are often large, leading to the creation of "big data"
systems designed to provide the scalability needed to analyze these graphs
using parallel processing. However, researchers have shown that while these
systems often provide scalability, they also often introduce overheads that
exceed the benefits they provide, sometimes resulting in lower absolute
performance than even simple serial implementations. This report studies the
viability and performance of the actor model to implement scalable concurrent
programs to perform common graph computations. We show that relatively simple
actor-based implementations outperform both dedicated graph processing systems
and the benchmark serial implementations.
| [
{
"version": "v1",
"created": "Wed, 4 Jan 2023 01:03:04 GMT"
},
{
"version": "v2",
"created": "Sat, 22 Apr 2023 00:22:24 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Buch",
"Ronak",
""
]
] | TITLE: COST of Graph Processing Using Actors
ABSTRACT: Graph processing is an increasingly important domain of computer science,
with applications in data and network analysis, among others. Target graphs in
these applications are often large, leading to the creation of "big data"
systems designed to provide the scalability needed to analyze these graphs
using parallel processing. However, researchers have shown that while these
systems often provide scalability, they also often introduce overheads that
exceed the benefits they provide, sometimes resulting in lower absolute
performance than even simple serial implementations. This report studies the
viability and performance of the actor model to implement scalable concurrent
programs to perform common graph computations. We show that relatively simple
actor-based implementations outperform both dedicated graph processing systems
and the benchmark serial implementations.
| no_new_dataset | 0.708055 |
2302.00539 | Nils Lukas | Nils Lukas, Ahmed Salem, Robert Sim, Shruti Tople, Lukas Wutschitz and
Santiago Zanella-B\'eguelin | Analyzing Leakage of Personally Identifiable Information in Language
Models | IEEE Symposium on Security and Privacy (S&P) 2023 | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Language Models (LMs) have been shown to leak information about training data
through sentence-level membership inference and reconstruction attacks.
Understanding the risk of LMs leaking Personally Identifiable Information (PII)
has received less attention, which can be attributed to the false assumption
that dataset curation techniques such as scrubbing are sufficient to prevent
PII leakage. Scrubbing techniques reduce but do not prevent the risk of PII
leakage: in practice scrubbing is imperfect and must balance the trade-off
between minimizing disclosure and preserving the utility of the dataset. On the
other hand, it is unclear to which extent algorithmic defenses such as
differential privacy, designed to guarantee sentence- or user-level privacy,
prevent PII disclosure. In this work, we introduce rigorous game-based
definitions for three types of PII leakage via black-box extraction, inference,
and reconstruction attacks with only API access to an LM. We empirically
evaluate the attacks against GPT-2 models fine-tuned with and without defenses
in three domains: case law, health care, and e-mails. Our main contributions
are (i) novel attacks that can extract up to 10$\times$ more PII sequences than
existing attacks, (ii) showing that sentence-level differential privacy reduces
the risk of PII disclosure but still leaks about 3% of PII sequences, and (iii)
a subtle connection between record-level membership inference and PII
reconstruction. Code to reproduce all experiments in the paper is available at
https://github.com/microsoft/analysing_pii_leakage.
| [
{
"version": "v1",
"created": "Wed, 1 Feb 2023 16:04:48 GMT"
},
{
"version": "v2",
"created": "Tue, 4 Apr 2023 15:07:57 GMT"
},
{
"version": "v3",
"created": "Thu, 6 Apr 2023 14:16:36 GMT"
},
{
"version": "v4",
"created": "Sun, 23 Apr 2023 22:20:47 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Lukas",
"Nils",
""
],
[
"Salem",
"Ahmed",
""
],
[
"Sim",
"Robert",
""
],
[
"Tople",
"Shruti",
""
],
[
"Wutschitz",
"Lukas",
""
],
[
"Zanella-Béguelin",
"Santiago",
""
]
] | TITLE: Analyzing Leakage of Personally Identifiable Information in Language
Models
ABSTRACT: Language Models (LMs) have been shown to leak information about training data
through sentence-level membership inference and reconstruction attacks.
Understanding the risk of LMs leaking Personally Identifiable Information (PII)
has received less attention, which can be attributed to the false assumption
that dataset curation techniques such as scrubbing are sufficient to prevent
PII leakage. Scrubbing techniques reduce but do not prevent the risk of PII
leakage: in practice scrubbing is imperfect and must balance the trade-off
between minimizing disclosure and preserving the utility of the dataset. On the
other hand, it is unclear to which extent algorithmic defenses such as
differential privacy, designed to guarantee sentence- or user-level privacy,
prevent PII disclosure. In this work, we introduce rigorous game-based
definitions for three types of PII leakage via black-box extraction, inference,
and reconstruction attacks with only API access to an LM. We empirically
evaluate the attacks against GPT-2 models fine-tuned with and without defenses
in three domains: case law, health care, and e-mails. Our main contributions
are (i) novel attacks that can extract up to 10$\times$ more PII sequences than
existing attacks, (ii) showing that sentence-level differential privacy reduces
the risk of PII disclosure but still leaks about 3% of PII sequences, and (iii)
a subtle connection between record-level membership inference and PII
reconstruction. Code to reproduce all experiments in the paper is available at
https://github.com/microsoft/analysing_pii_leakage.
| no_new_dataset | 0.709806 |
2302.01259 | Eivind Meyer | Eivind Meyer, Maurice Brenner, Bowen Zhang, Max Schickert, Bilal
Musani, and Matthias Althoff | Geometric Deep Learning for Autonomous Driving: Unlocking the Power of
Graph Neural Networks With CommonRoad-Geometric | Presented at IV 2023 | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Heterogeneous graphs offer powerful data representations for traffic, given
their ability to model the complex interaction effects among a varying number
of traffic participants and the underlying road infrastructure. With the recent
advent of graph neural networks (GNNs) as the accompanying deep learning
framework, the graph structure can be efficiently leveraged for various machine
learning applications such as trajectory prediction. As a first of its kind,
our proposed Python framework offers an easy-to-use and fully customizable data
processing pipeline to extract standardized graph datasets from traffic
scenarios. Providing a platform for GNN-based autonomous driving research, it
improves comparability between approaches and allows researchers to focus on
model implementation instead of dataset curation.
| [
{
"version": "v1",
"created": "Thu, 2 Feb 2023 17:45:02 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Apr 2023 08:30:35 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Meyer",
"Eivind",
""
],
[
"Brenner",
"Maurice",
""
],
[
"Zhang",
"Bowen",
""
],
[
"Schickert",
"Max",
""
],
[
"Musani",
"Bilal",
""
],
[
"Althoff",
"Matthias",
""
]
] | TITLE: Geometric Deep Learning for Autonomous Driving: Unlocking the Power of
Graph Neural Networks With CommonRoad-Geometric
ABSTRACT: Heterogeneous graphs offer powerful data representations for traffic, given
their ability to model the complex interaction effects among a varying number
of traffic participants and the underlying road infrastructure. With the recent
advent of graph neural networks (GNNs) as the accompanying deep learning
framework, the graph structure can be efficiently leveraged for various machine
learning applications such as trajectory prediction. As a first of its kind,
our proposed Python framework offers an easy-to-use and fully customizable data
processing pipeline to extract standardized graph datasets from traffic
scenarios. Providing a platform for GNN-based autonomous driving research, it
improves comparability between approaches and allows researchers to focus on
model implementation instead of dataset curation.
| no_new_dataset | 0.706817 |
2302.04774 | Boris N Oreshkin | Boris N. Oreshkin | 3D Human Pose and Shape Estimation via HybrIK-Transformer | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-sa/4.0/ | HybrIK relies on a combination of analytical inverse kinematics and deep
learning to produce more accurate 3D pose estimation from 2D monocular images.
HybrIK has three major components: (1) pretrained convolution backbone, (2)
deconvolution to lift 3D pose from 2D convolution features, (3) analytical
inverse kinematics pass correcting deep learning prediction using learned
distribution of plausible twist and swing angles. In this paper we propose an
enhancement of the 2D to 3D lifting module, replacing deconvolution with
Transformer, resulting in accuracy and computational efficiency improvement
relative to the original HybrIK method. We demonstrate our results on commonly
used H36M, PW3D, COCO and HP3D datasets. Our code is publicly available
https://github.com/boreshkinai/hybrik-transformer.
| [
{
"version": "v1",
"created": "Thu, 9 Feb 2023 17:08:43 GMT"
},
{
"version": "v2",
"created": "Mon, 20 Feb 2023 02:02:07 GMT"
},
{
"version": "v3",
"created": "Mon, 27 Mar 2023 15:33:35 GMT"
},
{
"version": "v4",
"created": "Sat, 22 Apr 2023 18:11:30 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Oreshkin",
"Boris N.",
""
]
] | TITLE: 3D Human Pose and Shape Estimation via HybrIK-Transformer
ABSTRACT: HybrIK relies on a combination of analytical inverse kinematics and deep
learning to produce more accurate 3D pose estimation from 2D monocular images.
HybrIK has three major components: (1) pretrained convolution backbone, (2)
deconvolution to lift 3D pose from 2D convolution features, (3) analytical
inverse kinematics pass correcting deep learning prediction using learned
distribution of plausible twist and swing angles. In this paper we propose an
enhancement of the 2D to 3D lifting module, replacing deconvolution with
Transformer, resulting in accuracy and computational efficiency improvement
relative to the original HybrIK method. We demonstrate our results on commonly
used H36M, PW3D, COCO and HP3D datasets. Our code is publicly available
https://github.com/boreshkinai/hybrik-transformer.
| no_new_dataset | 0.710245 |
2302.06086 | Linyi Li | Linyi Li, Yuhao Zhang, Luyao Ren, Yingfei Xiong, Tao Xie | Reliability Assurance for Deep Neural Network Architectures Against
Numerical Defects | To appear at 45th International Conference on Software Engineering
(ICSE 2023), camera-ready version, tiny typo fixed | null | null | null | cs.SE cs.LG cs.PL | http://creativecommons.org/licenses/by/4.0/ | With the widespread deployment of deep neural networks (DNNs), ensuring the
reliability of DNN-based systems is of great importance. Serious reliability
issues such as system failures can be caused by numerical defects, one of the
most frequent defects in DNNs. To assure high reliability against numerical
defects, in this paper, we propose the RANUM approach including novel
techniques for three reliability assurance tasks: detection of potential
numerical defects, confirmation of potential-defect feasibility, and suggestion
of defect fixes. To the best of our knowledge, RANUM is the first approach that
confirms potential-defect feasibility with failure-exhibiting tests and
suggests fixes automatically. Extensive experiments on the benchmarks of 63
real-world DNN architectures show that RANUM outperforms state-of-the-art
approaches across the three reliability assurance tasks. In addition, when the
RANUM-generated fixes are compared with developers' fixes on open-source
projects, in 37 out of 40 cases, RANUM-generated fixes are equivalent to or
even better than human fixes.
| [
{
"version": "v1",
"created": "Mon, 13 Feb 2023 04:09:34 GMT"
},
{
"version": "v2",
"created": "Sun, 26 Mar 2023 18:42:57 GMT"
},
{
"version": "v3",
"created": "Sun, 23 Apr 2023 18:13:01 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Li",
"Linyi",
""
],
[
"Zhang",
"Yuhao",
""
],
[
"Ren",
"Luyao",
""
],
[
"Xiong",
"Yingfei",
""
],
[
"Xie",
"Tao",
""
]
] | TITLE: Reliability Assurance for Deep Neural Network Architectures Against
Numerical Defects
ABSTRACT: With the widespread deployment of deep neural networks (DNNs), ensuring the
reliability of DNN-based systems is of great importance. Serious reliability
issues such as system failures can be caused by numerical defects, one of the
most frequent defects in DNNs. To assure high reliability against numerical
defects, in this paper, we propose the RANUM approach including novel
techniques for three reliability assurance tasks: detection of potential
numerical defects, confirmation of potential-defect feasibility, and suggestion
of defect fixes. To the best of our knowledge, RANUM is the first approach that
confirms potential-defect feasibility with failure-exhibiting tests and
suggests fixes automatically. Extensive experiments on the benchmarks of 63
real-world DNN architectures show that RANUM outperforms state-of-the-art
approaches across the three reliability assurance tasks. In addition, when the
RANUM-generated fixes are compared with developers' fixes on open-source
projects, in 37 out of 40 cases, RANUM-generated fixes are equivalent to or
even better than human fixes.
| no_new_dataset | 0.710603 |
2302.06745 | Hormoz Shahrzad | Hormoz Shahrzad, Risto Miikkulainen | Accelerating Evolution Through Gene Masking and Distributed Search | null | null | 10.1145/3583131.3590508 | null | cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In building practical applications of evolutionary computation (EC), two
optimizations are essential. First, the parameters of the search method need to
be tuned to the domain in order to balance exploration and exploitation
effectively. Second, the search method needs to be distributed to take
advantage of parallel computing resources. This paper presents BLADE (BLAnket
Distributed Evolution) as an approach to achieving both goals simultaneously.
BLADE uses blankets (i.e., masks on the genetic representation) to tune the
evolutionary operators during the search, and implements the search through
hub-and-spoke distribution. In the paper, (1) the blanket method is formalized
for the (1 + 1)EA case as a Markov chain process. Its effectiveness is then
demonstrated by analyzing dominant and subdominant eigenvalues of stochastic
matrices, suggesting a generalizable theory; (2) the fitness-level theory is
used to analyze the distribution method; and (3) these insights are verified
experimentally on three benchmark problems, showing that both blankets and
distribution lead to accelerated evolution. Moreover, a surprising synergy
emerges between them: When combined with distribution, the blanket approach
achieves more than $n$-fold speedup with $n$ clients in some cases. The work
thus highlights the importance and potential of optimizing evolutionary
computation in practical applications.
| [
{
"version": "v1",
"created": "Mon, 13 Feb 2023 23:11:25 GMT"
},
{
"version": "v2",
"created": "Mon, 10 Apr 2023 22:42:17 GMT"
},
{
"version": "v3",
"created": "Sun, 23 Apr 2023 06:26:16 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Shahrzad",
"Hormoz",
""
],
[
"Miikkulainen",
"Risto",
""
]
] | TITLE: Accelerating Evolution Through Gene Masking and Distributed Search
ABSTRACT: In building practical applications of evolutionary computation (EC), two
optimizations are essential. First, the parameters of the search method need to
be tuned to the domain in order to balance exploration and exploitation
effectively. Second, the search method needs to be distributed to take
advantage of parallel computing resources. This paper presents BLADE (BLAnket
Distributed Evolution) as an approach to achieving both goals simultaneously.
BLADE uses blankets (i.e., masks on the genetic representation) to tune the
evolutionary operators during the search, and implements the search through
hub-and-spoke distribution. In the paper, (1) the blanket method is formalized
for the (1 + 1)EA case as a Markov chain process. Its effectiveness is then
demonstrated by analyzing dominant and subdominant eigenvalues of stochastic
matrices, suggesting a generalizable theory; (2) the fitness-level theory is
used to analyze the distribution method; and (3) these insights are verified
experimentally on three benchmark problems, showing that both blankets and
distribution lead to accelerated evolution. Moreover, a surprising synergy
emerges between them: When combined with distribution, the blanket approach
achieves more than $n$-fold speedup with $n$ clients in some cases. The work
thus highlights the importance and potential of optimizing evolutionary
computation in practical applications.
| no_new_dataset | 0.711982 |
2302.09566 | David Shulman | David Shulman | Optimization Methods in Deep Learning: A Comprehensive Overview | We have decided to withdraw our manuscript 'Optimization Methods in
Deep Learning: A Comprehensive Overview' from arXiv as we are now pursuing
different research directions. We apologize for any inconvenience and
appreciate your understanding | null | null | null | cs.LG math.OC | http://creativecommons.org/licenses/by/4.0/ | In recent years, deep learning has achieved remarkable success in various
fields such as image recognition, natural language processing, and speech
recognition. The effectiveness of deep learning largely depends on the
optimization methods used to train deep neural networks. In this paper, we
provide an overview of first-order optimization methods such as Stochastic
Gradient Descent, Adagrad, Adadelta, and RMSprop, as well as recent
momentum-based and adaptive gradient methods such as Nesterov accelerated
gradient, Adam, Nadam, AdaMax, and AMSGrad. We also discuss the challenges
associated with optimization in deep learning and explore techniques for
addressing these challenges, including weight initialization, batch
normalization, and layer normalization. Finally, we provide recommendations for
selecting optimization methods for different deep learning tasks and datasets.
This paper serves as a comprehensive guide to optimization methods in deep
learning and can be used as a reference for researchers and practitioners in
the field.
| [
{
"version": "v1",
"created": "Sun, 19 Feb 2023 13:01:53 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Apr 2023 12:45:04 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Shulman",
"David",
""
]
] | TITLE: Optimization Methods in Deep Learning: A Comprehensive Overview
ABSTRACT: In recent years, deep learning has achieved remarkable success in various
fields such as image recognition, natural language processing, and speech
recognition. The effectiveness of deep learning largely depends on the
optimization methods used to train deep neural networks. In this paper, we
provide an overview of first-order optimization methods such as Stochastic
Gradient Descent, Adagrad, Adadelta, and RMSprop, as well as recent
momentum-based and adaptive gradient methods such as Nesterov accelerated
gradient, Adam, Nadam, AdaMax, and AMSGrad. We also discuss the challenges
associated with optimization in deep learning and explore techniques for
addressing these challenges, including weight initialization, batch
normalization, and layer normalization. Finally, we provide recommendations for
selecting optimization methods for different deep learning tasks and datasets.
This paper serves as a comprehensive guide to optimization methods in deep
learning and can be used as a reference for researchers and practitioners in
the field.
| no_new_dataset | 0.711982 |
2303.00138 | Allen Jiang | Yalong Jiang | Texture-Based Input Feature Selection for Action Recognition | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The performance of video action recognition has been significantly boosted by
using motion representations within a two-stream Convolutional Neural Network
(CNN) architecture. However, there are a few challenging problems in action
recognition in real scenarios, e.g., the variations in viewpoints and poses,
and the changes in backgrounds. The domain discrepancy between the training
data and the test data causes the performance drop. To improve the model
robustness, we propose a novel method to determine the task-irrelevant content
in inputs which increases the domain discrepancy. The method is based on a
human parsing model (HP model) which jointly conducts dense correspondence
labelling and semantic part segmentation. The predictions from the HP model
also function as re-rendering the human regions in each video using the same
set of textures to make humans appearances in all classes be the same. A
revised dataset is generated for training and testing and makes the action
recognition model exhibit invariance to the irrelevant content in the inputs.
Moreover, the predictions from the HP model are used to enrich the inputs to
the AR model during both training and testing. Experimental results show that
our proposed model is superior to existing models for action recognition on the
HMDB-51 dataset and the Penn Action dataset.
| [
{
"version": "v1",
"created": "Tue, 28 Feb 2023 23:56:31 GMT"
},
{
"version": "v2",
"created": "Sat, 4 Mar 2023 14:12:25 GMT"
},
{
"version": "v3",
"created": "Sun, 23 Apr 2023 09:00:53 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Jiang",
"Yalong",
""
]
] | TITLE: Texture-Based Input Feature Selection for Action Recognition
ABSTRACT: The performance of video action recognition has been significantly boosted by
using motion representations within a two-stream Convolutional Neural Network
(CNN) architecture. However, there are a few challenging problems in action
recognition in real scenarios, e.g., the variations in viewpoints and poses,
and the changes in backgrounds. The domain discrepancy between the training
data and the test data causes the performance drop. To improve the model
robustness, we propose a novel method to determine the task-irrelevant content
in inputs which increases the domain discrepancy. The method is based on a
human parsing model (HP model) which jointly conducts dense correspondence
labelling and semantic part segmentation. The predictions from the HP model
also function as re-rendering the human regions in each video using the same
set of textures to make humans appearances in all classes be the same. A
revised dataset is generated for training and testing and makes the action
recognition model exhibit invariance to the irrelevant content in the inputs.
Moreover, the predictions from the HP model are used to enrich the inputs to
the AR model during both training and testing. Experimental results show that
our proposed model is superior to existing models for action recognition on the
HMDB-51 dataset and the Penn Action dataset.
| no_new_dataset | 0.710478 |
2303.02001 | Jingyi Xu | Jingyi Xu, Hieu Le, Vu Nguyen, Viresh Ranjan, and Dimitris Samaras | Zero-shot Object Counting | CVPR 2023 | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Class-agnostic object counting aims to count object instances of an arbitrary
class at test time. It is challenging but also enables many potential
applications. Current methods require human-annotated exemplars as inputs which
are often unavailable for novel categories, especially for autonomous systems.
Thus, we propose zero-shot object counting (ZSC), a new setting where only the
class name is available during test time. Such a counting system does not
require human annotators in the loop and can operate automatically. Starting
from a class name, we propose a method that can accurately identify the optimal
patches which can then be used as counting exemplars. Specifically, we first
construct a class prototype to select the patches that are likely to contain
the objects of interest, namely class-relevant patches. Furthermore, we
introduce a model that can quantitatively measure how suitable an arbitrary
patch is as a counting exemplar. By applying this model to all the candidate
patches, we can select the most suitable patches as exemplars for counting.
Experimental results on a recent class-agnostic counting dataset, FSC-147,
validate the effectiveness of our method. Code is available at
https://github.com/cvlab-stonybrook/zero-shot-counting
| [
{
"version": "v1",
"created": "Fri, 3 Mar 2023 15:14:36 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Apr 2023 15:51:01 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Xu",
"Jingyi",
""
],
[
"Le",
"Hieu",
""
],
[
"Nguyen",
"Vu",
""
],
[
"Ranjan",
"Viresh",
""
],
[
"Samaras",
"Dimitris",
""
]
] | TITLE: Zero-shot Object Counting
ABSTRACT: Class-agnostic object counting aims to count object instances of an arbitrary
class at test time. It is challenging but also enables many potential
applications. Current methods require human-annotated exemplars as inputs which
are often unavailable for novel categories, especially for autonomous systems.
Thus, we propose zero-shot object counting (ZSC), a new setting where only the
class name is available during test time. Such a counting system does not
require human annotators in the loop and can operate automatically. Starting
from a class name, we propose a method that can accurately identify the optimal
patches which can then be used as counting exemplars. Specifically, we first
construct a class prototype to select the patches that are likely to contain
the objects of interest, namely class-relevant patches. Furthermore, we
introduce a model that can quantitatively measure how suitable an arbitrary
patch is as a counting exemplar. By applying this model to all the candidate
patches, we can select the most suitable patches as exemplars for counting.
Experimental results on a recent class-agnostic counting dataset, FSC-147,
validate the effectiveness of our method. Code is available at
https://github.com/cvlab-stonybrook/zero-shot-counting
| new_dataset | 0.50116 |
2303.09314 | Linhao Zhang | Linhao Zhang, Li Jin, Xian Sun, Guangluan Xu, Zequn Zhang, Xiaoyu Li,
Nayu Liu, Qing Liu, Shiyao Yan | TOT: Topology-Aware Optimal Transport For Multimodal Hate Detection | accepted at AAAI23 | null | null | null | cs.CL cs.AI cs.MM | http://creativecommons.org/licenses/by/4.0/ | Multimodal hate detection, which aims to identify harmful content online such
as memes, is crucial for building a wholesome internet environment. Previous
work has made enlightening exploration in detecting explicit hate remarks.
However, most of their approaches neglect the analysis of implicit harm, which
is particularly challenging as explicit text markers and demographic visual
cues are often twisted or missing. The leveraged cross-modal attention
mechanisms also suffer from the distributional modality gap and lack logical
interpretability. To address these semantic gaps issues, we propose TOT: a
topology-aware optimal transport framework to decipher the implicit harm in
memes scenario, which formulates the cross-modal aligning problem as solutions
for optimal transportation plans. Specifically, we leverage an optimal
transport kernel method to capture complementary information from multiple
modalities. The kernel embedding provides a non-linear transformation ability
to reproduce a kernel Hilbert space (RKHS), which reflects significance for
eliminating the distributional modality gap. Moreover, we perceive the topology
information based on aligned representations to conduct bipartite graph path
reasoning. The newly achieved state-of-the-art performance on two publicly
available benchmark datasets, together with further visual analysis,
demonstrate the superiority of TOT in capturing implicit cross-modal alignment.
| [
{
"version": "v1",
"created": "Mon, 27 Feb 2023 06:58:19 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Apr 2023 09:23:25 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Zhang",
"Linhao",
""
],
[
"Jin",
"Li",
""
],
[
"Sun",
"Xian",
""
],
[
"Xu",
"Guangluan",
""
],
[
"Zhang",
"Zequn",
""
],
[
"Li",
"Xiaoyu",
""
],
[
"Liu",
"Nayu",
""
],
[
"Liu",
"Qing",
""
],
[
"Yan",
"Shiyao",
""
]
] | TITLE: TOT: Topology-Aware Optimal Transport For Multimodal Hate Detection
ABSTRACT: Multimodal hate detection, which aims to identify harmful content online such
as memes, is crucial for building a wholesome internet environment. Previous
work has made enlightening exploration in detecting explicit hate remarks.
However, most of their approaches neglect the analysis of implicit harm, which
is particularly challenging as explicit text markers and demographic visual
cues are often twisted or missing. The leveraged cross-modal attention
mechanisms also suffer from the distributional modality gap and lack logical
interpretability. To address these semantic gaps issues, we propose TOT: a
topology-aware optimal transport framework to decipher the implicit harm in
memes scenario, which formulates the cross-modal aligning problem as solutions
for optimal transportation plans. Specifically, we leverage an optimal
transport kernel method to capture complementary information from multiple
modalities. The kernel embedding provides a non-linear transformation ability
to reproduce a kernel Hilbert space (RKHS), which reflects significance for
eliminating the distributional modality gap. Moreover, we perceive the topology
information based on aligned representations to conduct bipartite graph path
reasoning. The newly achieved state-of-the-art performance on two publicly
available benchmark datasets, together with further visual analysis,
demonstrate the superiority of TOT in capturing implicit cross-modal alignment.
| no_new_dataset | 0.709044 |
2303.12092 | Yu Dong | Yu Dong and Christy Jie Liang and Yi Chen and Jie Hua | A Visual Modeling Method for Spatiotemporal and Multidimensional
Features in Epidemiological Analysis: Applied COVID-19 Aggregated Datasets | null | null | null | null | stat.AP cs.GR | http://creativecommons.org/licenses/by/4.0/ | The visual modeling method enables flexible interactions with rich graphical
depictions of data and supports the exploration of the complexities of
epidemiological analysis. However, most epidemiology visualizations do not
support the combined analysis of objective factors that might influence the
transmission situation, resulting in a lack of quantitative and qualitative
evidence. To address this issue, we have developed a portrait-based visual
modeling method called +msRNAer. This method considers the spatiotemporal
features of virus transmission patterns and the multidimensional features of
objective risk factors in communities, enabling portrait-based exploration and
comparison in epidemiological analysis. We applied +msRNAer to aggregate
COVID-19-related datasets in New South Wales, Australia, which combined
COVID-19 case number trends, geo-information, intervention events, and
expert-supervised risk factors extracted from LGA-based censuses. We perfected
the +msRNAer workflow with collaborative views and evaluated its feasibility,
effectiveness, and usefulness through one user study and three subject-driven
case studies. Positive feedback from experts indicates that +msRNAer provides a
general understanding of analyzing comprehension that not only compares
relationships between cases in time-varying and risk factors through portraits
but also supports navigation in fundamental geographical, timeline, and other
factor comparisons. By adopting interactions, experts discovered functional and
practical implications for potential patterns of long-standing community
factors against the vulnerability faced by the pandemic. Experts confirmed that
+msRNAer is expected to deliver visual modeling benefits with spatiotemporal
and multidimensional features in other epidemiological analysis scenarios.
| [
{
"version": "v1",
"created": "Tue, 21 Mar 2023 12:16:43 GMT"
},
{
"version": "v2",
"created": "Tue, 28 Mar 2023 06:01:49 GMT"
},
{
"version": "v3",
"created": "Sun, 23 Apr 2023 00:20:51 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Dong",
"Yu",
""
],
[
"Liang",
"Christy Jie",
""
],
[
"Chen",
"Yi",
""
],
[
"Hua",
"Jie",
""
]
] | TITLE: A Visual Modeling Method for Spatiotemporal and Multidimensional
Features in Epidemiological Analysis: Applied COVID-19 Aggregated Datasets
ABSTRACT: The visual modeling method enables flexible interactions with rich graphical
depictions of data and supports the exploration of the complexities of
epidemiological analysis. However, most epidemiology visualizations do not
support the combined analysis of objective factors that might influence the
transmission situation, resulting in a lack of quantitative and qualitative
evidence. To address this issue, we have developed a portrait-based visual
modeling method called +msRNAer. This method considers the spatiotemporal
features of virus transmission patterns and the multidimensional features of
objective risk factors in communities, enabling portrait-based exploration and
comparison in epidemiological analysis. We applied +msRNAer to aggregate
COVID-19-related datasets in New South Wales, Australia, which combined
COVID-19 case number trends, geo-information, intervention events, and
expert-supervised risk factors extracted from LGA-based censuses. We perfected
the +msRNAer workflow with collaborative views and evaluated its feasibility,
effectiveness, and usefulness through one user study and three subject-driven
case studies. Positive feedback from experts indicates that +msRNAer provides a
general understanding of analyzing comprehension that not only compares
relationships between cases in time-varying and risk factors through portraits
but also supports navigation in fundamental geographical, timeline, and other
factor comparisons. By adopting interactions, experts discovered functional and
practical implications for potential patterns of long-standing community
factors against the vulnerability faced by the pandemic. Experts confirmed that
+msRNAer is expected to deliver visual modeling benefits with spatiotemporal
and multidimensional features in other epidemiological analysis scenarios.
| no_new_dataset | 0.711406 |
2303.14778 | Yabin Zhu | Yabin Zhu, Chenglong Li, Xiao Wang, Jin Tang, Zhixiang Huang | RGBT Tracking via Progressive Fusion Transformer with Dynamically Guided
Learning | Incorrect experimental setup on the LasHeR dataset | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Existing Transformer-based RGBT tracking methods either use cross-attention
to fuse the two modalities, or use self-attention and cross-attention to model
both modality-specific and modality-sharing information. However, the
significant appearance gap between modalities limits the feature representation
ability of certain modalities during the fusion process. To address this
problem, we propose a novel Progressive Fusion Transformer called ProFormer,
which progressively integrates single-modality information into the multimodal
representation for robust RGBT tracking. In particular, ProFormer first uses a
self-attention module to collaboratively extract the multimodal representation,
and then uses two cross-attention modules to interact it with the features of
the dual modalities respectively. In this way, the modality-specific
information can well be activated in the multimodal representation. Finally, a
feed-forward network is used to fuse two interacted multimodal representations
for the further enhancement of the final multimodal representation. In
addition, existing learning methods of RGBT trackers either fuse multimodal
features into one for final classification, or exploit the relationship between
unimodal branches and fused branch through a competitive learning strategy.
However, they either ignore the learning of single-modality branches or result
in one branch failing to be well optimized. To solve these problems, we propose
a dynamically guided learning algorithm that adaptively uses well-performing
branches to guide the learning of other branches, for enhancing the
representation ability of each branch. Extensive experiments demonstrate that
our proposed ProFormer sets a new state-of-the-art performance on RGBT210,
RGBT234, LasHeR, and VTUAV datasets.
| [
{
"version": "v1",
"created": "Sun, 26 Mar 2023 16:55:58 GMT"
},
{
"version": "v2",
"created": "Sat, 22 Apr 2023 08:55:55 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Zhu",
"Yabin",
""
],
[
"Li",
"Chenglong",
""
],
[
"Wang",
"Xiao",
""
],
[
"Tang",
"Jin",
""
],
[
"Huang",
"Zhixiang",
""
]
] | TITLE: RGBT Tracking via Progressive Fusion Transformer with Dynamically Guided
Learning
ABSTRACT: Existing Transformer-based RGBT tracking methods either use cross-attention
to fuse the two modalities, or use self-attention and cross-attention to model
both modality-specific and modality-sharing information. However, the
significant appearance gap between modalities limits the feature representation
ability of certain modalities during the fusion process. To address this
problem, we propose a novel Progressive Fusion Transformer called ProFormer,
which progressively integrates single-modality information into the multimodal
representation for robust RGBT tracking. In particular, ProFormer first uses a
self-attention module to collaboratively extract the multimodal representation,
and then uses two cross-attention modules to interact it with the features of
the dual modalities respectively. In this way, the modality-specific
information can well be activated in the multimodal representation. Finally, a
feed-forward network is used to fuse two interacted multimodal representations
for the further enhancement of the final multimodal representation. In
addition, existing learning methods of RGBT trackers either fuse multimodal
features into one for final classification, or exploit the relationship between
unimodal branches and fused branch through a competitive learning strategy.
However, they either ignore the learning of single-modality branches or result
in one branch failing to be well optimized. To solve these problems, we propose
a dynamically guided learning algorithm that adaptively uses well-performing
branches to guide the learning of other branches, for enhancing the
representation ability of each branch. Extensive experiments demonstrate that
our proposed ProFormer sets a new state-of-the-art performance on RGBT210,
RGBT234, LasHeR, and VTUAV datasets.
| no_new_dataset | 0.708421 |
2303.17972 | Jesujoba Alabi | Idris Akinade, Jesujoba Alabi, David Adelani, Clement Odoje and
Dietrich Klakow | $\varepsilon$ K\'U <MASK>: Integrating Yor\`ub\'a cultural greetings
into machine translation | C3NLP Workshop @ EACL2023 and AfricaNLP workshop @ ICLR2023 | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | This paper investigates the performance of massively multilingual neural
machine translation (NMT) systems in translating Yor\`ub\'a greetings
($\varepsilon$ k\'u [MASK]), which are a big part of Yor\`ub\'a language and
culture, into English. To evaluate these models, we present IkiniYor\`ub\'a, a
Yor\`ub\'a-English translation dataset containing some Yor\`ub\'a greetings,
and sample use cases. We analysed the performance of different multilingual NMT
systems including Google and NLLB and show that these models struggle to
accurately translate Yor\`ub\'a greetings into English. In addition, we trained
a Yor\`ub\'a-English model by finetuning an existing NMT model on the training
split of IkiniYor\`ub\'a and this achieved better performance when compared to
the pre-trained multilingual NMT models, although they were trained on a large
volume of data.
| [
{
"version": "v1",
"created": "Fri, 31 Mar 2023 11:16:20 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Apr 2023 16:29:38 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Akinade",
"Idris",
""
],
[
"Alabi",
"Jesujoba",
""
],
[
"Adelani",
"David",
""
],
[
"Odoje",
"Clement",
""
],
[
"Klakow",
"Dietrich",
""
]
] | TITLE: $\varepsilon$ K\'U <MASK>: Integrating Yor\`ub\'a cultural greetings
into machine translation
ABSTRACT: This paper investigates the performance of massively multilingual neural
machine translation (NMT) systems in translating Yor\`ub\'a greetings
($\varepsilon$ k\'u [MASK]), which are a big part of Yor\`ub\'a language and
culture, into English. To evaluate these models, we present IkiniYor\`ub\'a, a
Yor\`ub\'a-English translation dataset containing some Yor\`ub\'a greetings,
and sample use cases. We analysed the performance of different multilingual NMT
systems including Google and NLLB and show that these models struggle to
accurately translate Yor\`ub\'a greetings into English. In addition, we trained
a Yor\`ub\'a-English model by finetuning an existing NMT model on the training
split of IkiniYor\`ub\'a and this achieved better performance when compared to
the pre-trained multilingual NMT models, although they were trained on a large
volume of data.
| new_dataset | 0.717173 |
2304.00325 | Chenbin Pan | Chenbin Pan, Rui Hou, Hanchao Yu, Qifan Wang, Senem Velipasalar, and
Madian Khabsa | SVT: Supertoken Video Transformer for Efficient Video Understanding | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Whether by processing videos with fixed resolution from start to end or
incorporating pooling and down-scaling strategies, existing video transformers
process the whole video content throughout the network without specially
handling the large portions of redundant information. In this paper, we present
a Supertoken Video Transformer (SVT) that incorporates a Semantic Pooling
Module (SPM) to aggregate latent representations along the depth of visual
transformer based on their semantics, and thus, reduces redundancy inherent in
video inputs.~Qualitative results show that our method can effectively reduce
redundancy by merging latent representations with similar semantics and thus
increase the proportion of salient information for downstream
tasks.~Quantitatively, our method improves the performance of both ViT and MViT
while requiring significantly less computations on the Kinectics and
Something-Something-V2 benchmarks.~More specifically, with our SPM, we improve
the accuracy of MAE-pretrained ViT-B and ViT-L by 1.5% with 33% less GFLOPs and
by 0.2% with 55% less FLOPs, respectively, on the Kinectics-400 benchmark, and
improve the accuracy of MViTv2-B by 0.2% and 0.3% with 22% less GFLOPs on
Kinectics-400 and Something-Something-V2, respectively.
| [
{
"version": "v1",
"created": "Sat, 1 Apr 2023 14:31:56 GMT"
},
{
"version": "v2",
"created": "Sun, 23 Apr 2023 21:42:25 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Pan",
"Chenbin",
""
],
[
"Hou",
"Rui",
""
],
[
"Yu",
"Hanchao",
""
],
[
"Wang",
"Qifan",
""
],
[
"Velipasalar",
"Senem",
""
],
[
"Khabsa",
"Madian",
""
]
] | TITLE: SVT: Supertoken Video Transformer for Efficient Video Understanding
ABSTRACT: Whether by processing videos with fixed resolution from start to end or
incorporating pooling and down-scaling strategies, existing video transformers
process the whole video content throughout the network without specially
handling the large portions of redundant information. In this paper, we present
a Supertoken Video Transformer (SVT) that incorporates a Semantic Pooling
Module (SPM) to aggregate latent representations along the depth of visual
transformer based on their semantics, and thus, reduces redundancy inherent in
video inputs.~Qualitative results show that our method can effectively reduce
redundancy by merging latent representations with similar semantics and thus
increase the proportion of salient information for downstream
tasks.~Quantitatively, our method improves the performance of both ViT and MViT
while requiring significantly less computations on the Kinectics and
Something-Something-V2 benchmarks.~More specifically, with our SPM, we improve
the accuracy of MAE-pretrained ViT-B and ViT-L by 1.5% with 33% less GFLOPs and
by 0.2% with 55% less FLOPs, respectively, on the Kinectics-400 benchmark, and
improve the accuracy of MViTv2-B by 0.2% and 0.3% with 22% less GFLOPs on
Kinectics-400 and Something-Something-V2, respectively.
| no_new_dataset | 0.710603 |
2304.00792 | Suho Lee | Suho Lee, Seungwon Seo, Jihyo Kim, Yejin Lee, Sangheum Hwang | Few-shot Fine-tuning is All You Need for Source-free Domain Adaptation | The first two authors contributed equally | null | null | null | cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recently, source-free unsupervised domain adaptation (SFUDA) has emerged as a
more practical and feasible approach compared to unsupervised domain adaptation
(UDA) which assumes that labeled source data are always accessible. However,
significant limitations associated with SFUDA approaches are often overlooked,
which limits their practicality in real-world applications. These limitations
include a lack of principled ways to determine optimal hyperparameters and
performance degradation when the unlabeled target data fail to meet certain
requirements such as a closed-set and identical label distribution to the
source data. All these limitations stem from the fact that SFUDA entirely
relies on unlabeled target data. We empirically demonstrate the limitations of
existing SFUDA methods in real-world scenarios including out-of-distribution
and label distribution shifts in target data, and verify that none of these
methods can be safely applied to real-world settings. Based on our experimental
results, we claim that fine-tuning a source pretrained model with a few labeled
data (e.g., 1- or 3-shot) is a practical and reliable solution to circumvent
the limitations of SFUDA. Contrary to common belief, we find that carefully
fine-tuned models do not suffer from overfitting even when trained with only a
few labeled data, and also show little change in performance due to sampling
bias. Our experimental results on various domain adaptation benchmarks
demonstrate that the few-shot fine-tuning approach performs comparatively under
the standard SFUDA settings, and outperforms comparison methods under realistic
scenarios. Our code is available at https://github.com/daintlab/fewshot-SFDA .
| [
{
"version": "v1",
"created": "Mon, 3 Apr 2023 08:24:40 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Apr 2023 13:23:53 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Lee",
"Suho",
""
],
[
"Seo",
"Seungwon",
""
],
[
"Kim",
"Jihyo",
""
],
[
"Lee",
"Yejin",
""
],
[
"Hwang",
"Sangheum",
""
]
] | TITLE: Few-shot Fine-tuning is All You Need for Source-free Domain Adaptation
ABSTRACT: Recently, source-free unsupervised domain adaptation (SFUDA) has emerged as a
more practical and feasible approach compared to unsupervised domain adaptation
(UDA) which assumes that labeled source data are always accessible. However,
significant limitations associated with SFUDA approaches are often overlooked,
which limits their practicality in real-world applications. These limitations
include a lack of principled ways to determine optimal hyperparameters and
performance degradation when the unlabeled target data fail to meet certain
requirements such as a closed-set and identical label distribution to the
source data. All these limitations stem from the fact that SFUDA entirely
relies on unlabeled target data. We empirically demonstrate the limitations of
existing SFUDA methods in real-world scenarios including out-of-distribution
and label distribution shifts in target data, and verify that none of these
methods can be safely applied to real-world settings. Based on our experimental
results, we claim that fine-tuning a source pretrained model with a few labeled
data (e.g., 1- or 3-shot) is a practical and reliable solution to circumvent
the limitations of SFUDA. Contrary to common belief, we find that carefully
fine-tuned models do not suffer from overfitting even when trained with only a
few labeled data, and also show little change in performance due to sampling
bias. Our experimental results on various domain adaptation benchmarks
demonstrate that the few-shot fine-tuning approach performs comparatively under
the standard SFUDA settings, and outperforms comparison methods under realistic
scenarios. Our code is available at https://github.com/daintlab/fewshot-SFDA .
| no_new_dataset | 0.710622 |
2304.09254 | Radhika Tibrewala | Radhika Tibrewala, Tarun Dutt, Angela Tong, Luke Ginocchio, Mahesh B
Keerthivasan, Steven H Baete, Sumit Chopra, Yvonne W Lui, Daniel K Sodickson,
Hersh Chandarana, Patricia M Johnson | FastMRI Prostate: A Publicly Available, Biparametric MRI Dataset to
Advance Machine Learning for Prostate Cancer Imaging | 4 pages, 1 figure | null | null | null | physics.med-ph cs.LG eess.IV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The fastMRI brain and knee dataset has enabled significant advances in
exploring reconstruction methods for improving speed and image quality for
Magnetic Resonance Imaging (MRI) via novel, clinically relevant reconstruction
approaches. In this study, we describe the April 2023 expansion of the fastMRI
dataset to include biparametric prostate MRI data acquired on a clinical
population. The dataset consists of raw k-space and reconstructed images for
T2-weighted and diffusion-weighted sequences along with slice-level labels that
indicate the presence and grade of prostate cancer. As has been the case with
fastMRI, increasing accessibility to raw prostate MRI data will further
facilitate research in MR image reconstruction and evaluation with the larger
goal of improving the utility of MRI for prostate cancer detection and
evaluation. The dataset is available at https://fastmri.med.nyu.edu.
| [
{
"version": "v1",
"created": "Tue, 18 Apr 2023 19:34:28 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Tibrewala",
"Radhika",
""
],
[
"Dutt",
"Tarun",
""
],
[
"Tong",
"Angela",
""
],
[
"Ginocchio",
"Luke",
""
],
[
"Keerthivasan",
"Mahesh B",
""
],
[
"Baete",
"Steven H",
""
],
[
"Chopra",
"Sumit",
""
],
[
"Lui",
"Yvonne W",
""
],
[
"Sodickson",
"Daniel K",
""
],
[
"Chandarana",
"Hersh",
""
],
[
"Johnson",
"Patricia M",
""
]
] | TITLE: FastMRI Prostate: A Publicly Available, Biparametric MRI Dataset to
Advance Machine Learning for Prostate Cancer Imaging
ABSTRACT: The fastMRI brain and knee dataset has enabled significant advances in
exploring reconstruction methods for improving speed and image quality for
Magnetic Resonance Imaging (MRI) via novel, clinically relevant reconstruction
approaches. In this study, we describe the April 2023 expansion of the fastMRI
dataset to include biparametric prostate MRI data acquired on a clinical
population. The dataset consists of raw k-space and reconstructed images for
T2-weighted and diffusion-weighted sequences along with slice-level labels that
indicate the presence and grade of prostate cancer. As has been the case with
fastMRI, increasing accessibility to raw prostate MRI data will further
facilitate research in MR image reconstruction and evaluation with the larger
goal of improving the utility of MRI for prostate cancer detection and
evaluation. The dataset is available at https://fastmri.med.nyu.edu.
| no_new_dataset | 0.61086 |
2304.10145 | Yiming Zhu | Yiming Zhu, Peixian Zhang, Ehsan-Ul Haq, Pan Hui, Gareth Tyson | Can ChatGPT Reproduce Human-Generated Labels? A Study of Social
Computing Tasks | null | null | null | null | cs.AI cs.CL | http://creativecommons.org/licenses/by-nc-sa/4.0/ | The release of ChatGPT has uncovered a range of possibilities whereby large
language models (LLMs) can substitute human intelligence. In this paper, we
seek to understand whether ChatGPT has the potential to reproduce
human-generated label annotations in social computing tasks. Such an
achievement could significantly reduce the cost and complexity of social
computing research. As such, we use ChatGPT to relabel five seminal datasets
covering stance detection (2x), sentiment analysis, hate speech, and bot
detection. Our results highlight that ChatGPT does have the potential to handle
these data annotation tasks, although a number of challenges remain. ChatGPT
obtains an average accuracy 0.609. Performance is highest for the sentiment
analysis dataset, with ChatGPT correctly annotating 64.9% of tweets. Yet, we
show that performance varies substantially across individual labels. We believe
this work can open up new lines of analysis and act as a basis for future
research into the exploitation of ChatGPT for human annotation tasks.
| [
{
"version": "v1",
"created": "Thu, 20 Apr 2023 08:08:12 GMT"
},
{
"version": "v2",
"created": "Sat, 22 Apr 2023 08:55:33 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Zhu",
"Yiming",
""
],
[
"Zhang",
"Peixian",
""
],
[
"Haq",
"Ehsan-Ul",
""
],
[
"Hui",
"Pan",
""
],
[
"Tyson",
"Gareth",
""
]
] | TITLE: Can ChatGPT Reproduce Human-Generated Labels? A Study of Social
Computing Tasks
ABSTRACT: The release of ChatGPT has uncovered a range of possibilities whereby large
language models (LLMs) can substitute human intelligence. In this paper, we
seek to understand whether ChatGPT has the potential to reproduce
human-generated label annotations in social computing tasks. Such an
achievement could significantly reduce the cost and complexity of social
computing research. As such, we use ChatGPT to relabel five seminal datasets
covering stance detection (2x), sentiment analysis, hate speech, and bot
detection. Our results highlight that ChatGPT does have the potential to handle
these data annotation tasks, although a number of challenges remain. ChatGPT
obtains an average accuracy 0.609. Performance is highest for the sentiment
analysis dataset, with ChatGPT correctly annotating 64.9% of tweets. Yet, we
show that performance varies substantially across individual labels. We believe
this work can open up new lines of analysis and act as a basis for future
research into the exploitation of ChatGPT for human annotation tasks.
| no_new_dataset | 0.708011 |
2304.10179 | Yushuang Wu | Yushuang Wu, Zizheng Yan, Ce Chen, Lai Wei, Xiao Li, Guanbin Li, Yihao
Li, Shuguang Cui, Xiaoguang Han | SCoDA: Domain Adaptive Shape Completion for Real Scans | null | CVPR 2023 | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | 3D shape completion from point clouds is a challenging task, especially from
scans of real-world objects. Considering the paucity of 3D shape ground truths
for real scans, existing works mainly focus on benchmarking this task on
synthetic data, e.g. 3D computer-aided design models. However, the domain gap
between synthetic and real data limits the generalizability of these methods.
Thus, we propose a new task, SCoDA, for the domain adaptation of real scan
shape completion from synthetic data. A new dataset, ScanSalon, is contributed
with a bunch of elaborate 3D models created by skillful artists according to
scans. To address this new task, we propose a novel cross-domain feature fusion
method for knowledge transfer and a novel volume-consistent self-training
framework for robust learning from real data. Extensive experiments prove our
method is effective to bring an improvement of 6%~7% mIoU.
| [
{
"version": "v1",
"created": "Thu, 20 Apr 2023 09:38:26 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Apr 2023 06:31:59 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Wu",
"Yushuang",
""
],
[
"Yan",
"Zizheng",
""
],
[
"Chen",
"Ce",
""
],
[
"Wei",
"Lai",
""
],
[
"Li",
"Xiao",
""
],
[
"Li",
"Guanbin",
""
],
[
"Li",
"Yihao",
""
],
[
"Cui",
"Shuguang",
""
],
[
"Han",
"Xiaoguang",
""
]
] | TITLE: SCoDA: Domain Adaptive Shape Completion for Real Scans
ABSTRACT: 3D shape completion from point clouds is a challenging task, especially from
scans of real-world objects. Considering the paucity of 3D shape ground truths
for real scans, existing works mainly focus on benchmarking this task on
synthetic data, e.g. 3D computer-aided design models. However, the domain gap
between synthetic and real data limits the generalizability of these methods.
Thus, we propose a new task, SCoDA, for the domain adaptation of real scan
shape completion from synthetic data. A new dataset, ScanSalon, is contributed
with a bunch of elaborate 3D models created by skillful artists according to
scans. To address this new task, we propose a novel cross-domain feature fusion
method for knowledge transfer and a novel volume-consistent self-training
framework for robust learning from real data. Extensive experiments prove our
method is effective to bring an improvement of 6%~7% mIoU.
| new_dataset | 0.715474 |
2304.10244 | Hang Wang | Hang Wang, Xuanhong Chen, Bingbing Ni, Yutian Liu, Jinfan Liu | Omni Aggregation Networks for Lightweight Image Super-Resolution | Accepted by CVPR2023. Code is available at
\url{https://github.com/Francis0625/Omni-SR} | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While lightweight ViT framework has made tremendous progress in image
super-resolution, its uni-dimensional self-attention modeling, as well as
homogeneous aggregation scheme, limit its effective receptive field (ERF) to
include more comprehensive interactions from both spatial and channel
dimensions. To tackle these drawbacks, this work proposes two enhanced
components under a new Omni-SR architecture. First, an Omni Self-Attention
(OSA) block is proposed based on dense interaction principle, which can
simultaneously model pixel-interaction from both spatial and channel
dimensions, mining the potential correlations across omni-axis (i.e., spatial
and channel). Coupling with mainstream window partitioning strategies, OSA can
achieve superior performance with compelling computational budgets. Second, a
multi-scale interaction scheme is proposed to mitigate sub-optimal ERF (i.e.,
premature saturation) in shallow models, which facilitates local propagation
and meso-/global-scale interactions, rendering an omni-scale aggregation
building block. Extensive experiments demonstrate that Omni-SR achieves
record-high performance on lightweight super-resolution benchmarks (e.g., 26.95
dB@Urban100 $\times 4$ with only 792K parameters). Our code is available at
\url{https://github.com/Francis0625/Omni-SR}.
| [
{
"version": "v1",
"created": "Thu, 20 Apr 2023 12:05:14 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Apr 2023 09:03:56 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Wang",
"Hang",
""
],
[
"Chen",
"Xuanhong",
""
],
[
"Ni",
"Bingbing",
""
],
[
"Liu",
"Yutian",
""
],
[
"Liu",
"Jinfan",
""
]
] | TITLE: Omni Aggregation Networks for Lightweight Image Super-Resolution
ABSTRACT: While lightweight ViT framework has made tremendous progress in image
super-resolution, its uni-dimensional self-attention modeling, as well as
homogeneous aggregation scheme, limit its effective receptive field (ERF) to
include more comprehensive interactions from both spatial and channel
dimensions. To tackle these drawbacks, this work proposes two enhanced
components under a new Omni-SR architecture. First, an Omni Self-Attention
(OSA) block is proposed based on dense interaction principle, which can
simultaneously model pixel-interaction from both spatial and channel
dimensions, mining the potential correlations across omni-axis (i.e., spatial
and channel). Coupling with mainstream window partitioning strategies, OSA can
achieve superior performance with compelling computational budgets. Second, a
multi-scale interaction scheme is proposed to mitigate sub-optimal ERF (i.e.,
premature saturation) in shallow models, which facilitates local propagation
and meso-/global-scale interactions, rendering an omni-scale aggregation
building block. Extensive experiments demonstrate that Omni-SR achieves
record-high performance on lightweight super-resolution benchmarks (e.g., 26.95
dB@Urban100 $\times 4$ with only 792K parameters). Our code is available at
\url{https://github.com/Francis0625/Omni-SR}.
| no_new_dataset | 0.708824 |
2304.10558 | Yihao Zhang | Yihao Zhang, Zeming Wei, Xiyue Zhang, Meng Sun | Using Z3 for Formal Modeling and Verification of FNN Global Robustness | Accepted By SEKE 2023 | null | null | null | cs.LG cs.AI cs.LO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While Feedforward Neural Networks (FNNs) have achieved remarkable success in
various tasks, they are vulnerable to adversarial examples. Several techniques
have been developed to verify the adversarial robustness of FNNs, but most of
them focus on robustness verification against the local perturbation
neighborhood of a single data point. There is still a large research gap in
global robustness analysis. The global-robustness verifiable framework
DeepGlobal has been proposed to identify \textit{all} possible Adversarial
Dangerous Regions (ADRs) of FNNs, not limited to data samples in a test set. In
this paper, we propose a complete specification and implementation of
DeepGlobal utilizing the SMT solver Z3 for more explicit definition, and
propose several improvements to DeepGlobal for more efficient verification. To
evaluate the effectiveness of our implementation and improvements, we conduct
extensive experiments on a set of benchmark datasets. Visualization of our
experiment results shows the validity and effectiveness of the approach.
| [
{
"version": "v1",
"created": "Thu, 20 Apr 2023 15:40:22 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Apr 2023 05:53:03 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Zhang",
"Yihao",
""
],
[
"Wei",
"Zeming",
""
],
[
"Zhang",
"Xiyue",
""
],
[
"Sun",
"Meng",
""
]
] | TITLE: Using Z3 for Formal Modeling and Verification of FNN Global Robustness
ABSTRACT: While Feedforward Neural Networks (FNNs) have achieved remarkable success in
various tasks, they are vulnerable to adversarial examples. Several techniques
have been developed to verify the adversarial robustness of FNNs, but most of
them focus on robustness verification against the local perturbation
neighborhood of a single data point. There is still a large research gap in
global robustness analysis. The global-robustness verifiable framework
DeepGlobal has been proposed to identify \textit{all} possible Adversarial
Dangerous Regions (ADRs) of FNNs, not limited to data samples in a test set. In
this paper, we propose a complete specification and implementation of
DeepGlobal utilizing the SMT solver Z3 for more explicit definition, and
propose several improvements to DeepGlobal for more efficient verification. To
evaluate the effectiveness of our implementation and improvements, we conduct
extensive experiments on a set of benchmark datasets. Visualization of our
experiment results shows the validity and effectiveness of the approach.
| no_new_dataset | 0.709799 |
2304.10744 | Jieming Bian | Jieming Bian, Cong Shen, Jie Xu | Joint Client Assignment and UAV Route Planning for
Indirect-Communication Federated Learning | 31 pages, under review, extended version of conference paper:
arXiv:2302.07323 | null | null | null | cs.DC cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Federated Learning (FL) is a machine learning approach that enables the
creation of shared models for powerful applications while allowing data to
remain on devices. This approach provides benefits such as improved data
privacy, security, and reduced latency. However, in some systems, direct
communication between clients and servers may not be possible, such as remote
areas without proper communication infrastructure. To overcome this challenge,
a new framework called FedEx (Federated Learning via Model Express Delivery) is
proposed. This framework employs mobile transporters, such as UAVs, to
establish indirect communication channels between the server and clients. These
transporters act as intermediaries and allow for model information exchange.
The use of indirect communication presents new challenges for convergence
analysis and optimization, as the delay introduced by the transporters'
movement creates issues for both global model dissemination and local model
collection. To address this, two algorithms, FedEx-Sync and FedEx-Async, are
proposed for synchronized and asynchronized learning at the transporter level.
Additionally, a bi-level optimization algorithm is proposed to solve the joint
client assignment and route planning problem. Experimental validation using two
public datasets in a simulated network demonstrates consistent results with the
theory, proving the efficacy of FedEx.
| [
{
"version": "v1",
"created": "Fri, 21 Apr 2023 04:47:54 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Apr 2023 14:57:18 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Bian",
"Jieming",
""
],
[
"Shen",
"Cong",
""
],
[
"Xu",
"Jie",
""
]
] | TITLE: Joint Client Assignment and UAV Route Planning for
Indirect-Communication Federated Learning
ABSTRACT: Federated Learning (FL) is a machine learning approach that enables the
creation of shared models for powerful applications while allowing data to
remain on devices. This approach provides benefits such as improved data
privacy, security, and reduced latency. However, in some systems, direct
communication between clients and servers may not be possible, such as remote
areas without proper communication infrastructure. To overcome this challenge,
a new framework called FedEx (Federated Learning via Model Express Delivery) is
proposed. This framework employs mobile transporters, such as UAVs, to
establish indirect communication channels between the server and clients. These
transporters act as intermediaries and allow for model information exchange.
The use of indirect communication presents new challenges for convergence
analysis and optimization, as the delay introduced by the transporters'
movement creates issues for both global model dissemination and local model
collection. To address this, two algorithms, FedEx-Sync and FedEx-Async, are
proposed for synchronized and asynchronized learning at the transporter level.
Additionally, a bi-level optimization algorithm is proposed to solve the joint
client assignment and route planning problem. Experimental validation using two
public datasets in a simulated network demonstrates consistent results with the
theory, proving the efficacy of FedEx.
| no_new_dataset | 0.710829 |
2304.11164 | Jose Hernandez-Orallo | Anthony G Cohn, Jose Hernandez-Orallo | Dialectical language model evaluation: An initial appraisal of the
commonsense spatial reasoning abilities of LLMs | 11 pages in main paper + 71 pages in appendix | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Language models have become very popular recently and many claims have been
made about their abilities, including for commonsense reasoning. Given the
increasingly better results of current language models on previous static
benchmarks for commonsense reasoning, we explore an alternative dialectical
evaluation. The goal of this kind of evaluation is not to obtain an aggregate
performance value but to find failures and map the boundaries of the system.
Dialoguing with the system gives the opportunity to check for consistency and
get more reassurance of these boundaries beyond anecdotal evidence. In this
paper we conduct some qualitative investigations of this kind of evaluation for
the particular case of spatial reasoning (which is a fundamental aspect of
commonsense reasoning). We conclude with some suggestions for future work both
to improve the capabilities of language models and to systematise this kind of
dialectical evaluation.
| [
{
"version": "v1",
"created": "Sat, 22 Apr 2023 06:28:46 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Cohn",
"Anthony G",
""
],
[
"Hernandez-Orallo",
"Jose",
""
]
] | TITLE: Dialectical language model evaluation: An initial appraisal of the
commonsense spatial reasoning abilities of LLMs
ABSTRACT: Language models have become very popular recently and many claims have been
made about their abilities, including for commonsense reasoning. Given the
increasingly better results of current language models on previous static
benchmarks for commonsense reasoning, we explore an alternative dialectical
evaluation. The goal of this kind of evaluation is not to obtain an aggregate
performance value but to find failures and map the boundaries of the system.
Dialoguing with the system gives the opportunity to check for consistency and
get more reassurance of these boundaries beyond anecdotal evidence. In this
paper we conduct some qualitative investigations of this kind of evaluation for
the particular case of spatial reasoning (which is a fundamental aspect of
commonsense reasoning). We conclude with some suggestions for future work both
to improve the capabilities of language models and to systematise this kind of
dialectical evaluation.
| no_new_dataset | 0.710201 |
2304.11168 | Prakash Chandra Chhipa | Ekta Gupta, Varun Gupta, Muskaan Chopra, Prakash Chandra Chhipa and
Marcus Liwicki | Learning Self-Supervised Representations for Label Efficient
Cross-Domain Knowledge Transfer on Diabetic Retinopathy Fundus Images | Accepted to International Joint Conference on Neural Networks (IJCNN)
2023 | null | null | null | eess.IV cs.CV | http://creativecommons.org/licenses/by/4.0/ | This work presents a novel label-efficient selfsupervised representation
learning-based approach for classifying diabetic retinopathy (DR) images in
cross-domain settings. Most of the existing DR image classification methods are
based on supervised learning which requires a lot of time-consuming and
expensive medical domain experts-annotated data for training. The proposed
approach uses the prior learning from the source DR image dataset to classify
images drawn from the target datasets. The image representations learned from
the unlabeled source domain dataset through contrastive learning are used to
classify DR images from the target domain dataset. Moreover, the proposed
approach requires a few labeled images to perform successfully on DR image
classification tasks in cross-domain settings. The proposed work experiments
with four publicly available datasets: EyePACS, APTOS 2019, MESSIDOR-I, and
Fundus Images for self-supervised representation learning-based DR image
classification in cross-domain settings. The proposed method achieves
state-of-the-art results on binary and multiclassification of DR images, even
in cross-domain settings. The proposed method outperforms the existing DR image
binary and multi-class classification methods proposed in the literature. The
proposed method is also validated qualitatively using class activation maps,
revealing that the method can learn explainable image representations. The
source code and trained models are published on GitHub.
| [
{
"version": "v1",
"created": "Thu, 20 Apr 2023 12:46:34 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Gupta",
"Ekta",
""
],
[
"Gupta",
"Varun",
""
],
[
"Chopra",
"Muskaan",
""
],
[
"Chhipa",
"Prakash Chandra",
""
],
[
"Liwicki",
"Marcus",
""
]
] | TITLE: Learning Self-Supervised Representations for Label Efficient
Cross-Domain Knowledge Transfer on Diabetic Retinopathy Fundus Images
ABSTRACT: This work presents a novel label-efficient selfsupervised representation
learning-based approach for classifying diabetic retinopathy (DR) images in
cross-domain settings. Most of the existing DR image classification methods are
based on supervised learning which requires a lot of time-consuming and
expensive medical domain experts-annotated data for training. The proposed
approach uses the prior learning from the source DR image dataset to classify
images drawn from the target datasets. The image representations learned from
the unlabeled source domain dataset through contrastive learning are used to
classify DR images from the target domain dataset. Moreover, the proposed
approach requires a few labeled images to perform successfully on DR image
classification tasks in cross-domain settings. The proposed work experiments
with four publicly available datasets: EyePACS, APTOS 2019, MESSIDOR-I, and
Fundus Images for self-supervised representation learning-based DR image
classification in cross-domain settings. The proposed method achieves
state-of-the-art results on binary and multiclassification of DR images, even
in cross-domain settings. The proposed method outperforms the existing DR image
binary and multi-class classification methods proposed in the literature. The
proposed method is also validated qualitatively using class activation maps,
revealing that the method can learn explainable image representations. The
source code and trained models are published on GitHub.
| no_new_dataset | 0.71345 |
2304.11193 | Amir Masoud Ghalamzan Esfahani | Willow Mandil and Amir Ghalamzan-E | Combining Vision and Tactile Sensation for Video Prediction | This paper submitted to TRO | null | null | null | cs.RO cs.AI cs.CV | http://creativecommons.org/licenses/by/4.0/ | In this paper, we explore the impact of adding tactile sensation to video
prediction models for physical robot interactions. Predicting the impact of
robotic actions on the environment is a fundamental challenge in robotics.
Current methods leverage visual and robot action data to generate video
predictions over a given time period, which can then be used to adjust robot
actions. However, humans rely on both visual and tactile feedback to develop
and maintain a mental model of their physical surroundings. In this paper, we
investigate the impact of integrating tactile feedback into video prediction
models for physical robot interactions. We propose three multi-modal
integration approaches and compare the performance of these tactile-enhanced
video prediction models. Additionally, we introduce two new datasets of robot
pushing that use a magnetic-based tactile sensor for unsupervised learning. The
first dataset contains visually identical objects with different physical
properties, while the second dataset mimics existing robot-pushing datasets of
household object clusters. Our results demonstrate that incorporating tactile
feedback into video prediction models improves scene prediction accuracy and
enhances the agent's perception of physical interactions and understanding of
cause-effect relationships during physical robot interactions.
| [
{
"version": "v1",
"created": "Fri, 21 Apr 2023 18:02:15 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Mandil",
"Willow",
""
],
[
"Ghalamzan-E",
"Amir",
""
]
] | TITLE: Combining Vision and Tactile Sensation for Video Prediction
ABSTRACT: In this paper, we explore the impact of adding tactile sensation to video
prediction models for physical robot interactions. Predicting the impact of
robotic actions on the environment is a fundamental challenge in robotics.
Current methods leverage visual and robot action data to generate video
predictions over a given time period, which can then be used to adjust robot
actions. However, humans rely on both visual and tactile feedback to develop
and maintain a mental model of their physical surroundings. In this paper, we
investigate the impact of integrating tactile feedback into video prediction
models for physical robot interactions. We propose three multi-modal
integration approaches and compare the performance of these tactile-enhanced
video prediction models. Additionally, we introduce two new datasets of robot
pushing that use a magnetic-based tactile sensor for unsupervised learning. The
first dataset contains visually identical objects with different physical
properties, while the second dataset mimics existing robot-pushing datasets of
household object clusters. Our results demonstrate that incorporating tactile
feedback into video prediction models improves scene prediction accuracy and
enhances the agent's perception of physical interactions and understanding of
cause-effect relationships during physical robot interactions.
| new_dataset | 0.714858 |
2304.11196 | Saeejith Nair | Alexander Wong, Yifan Wu, Saad Abbasi, Saeejith Nair, Yuhao Chen,
Mohammad Javad Shafiee | Fast GraspNeXt: A Fast Self-Attention Neural Network Architecture for
Multi-task Learning in Computer Vision Tasks for Robotic Grasping on the Edge | Accepted at CVPR-NAS 2023 Workshop | null | null | null | cs.CV cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Multi-task learning has shown considerable promise for improving the
performance of deep learning-driven vision systems for the purpose of robotic
grasping. However, high architectural and computational complexity can result
in poor suitability for deployment on embedded devices that are typically
leveraged in robotic arms for real-world manufacturing and warehouse
environments. As such, the design of highly efficient multi-task deep neural
network architectures tailored for computer vision tasks for robotic grasping
on the edge is highly desired for widespread adoption in manufacturing
environments. Motivated by this, we propose Fast GraspNeXt, a fast
self-attention neural network architecture tailored for embedded multi-task
learning in computer vision tasks for robotic grasping. To build Fast
GraspNeXt, we leverage a generative network architecture search strategy with a
set of architectural constraints customized to achieve a strong balance between
multi-task learning performance and embedded inference efficiency. Experimental
results on the MetaGraspNet benchmark dataset show that the Fast GraspNeXt
network design achieves the highest performance (average precision (AP),
accuracy, and mean squared error (MSE)) across multiple computer vision tasks
when compared to other efficient multi-task network architecture designs, while
having only 17.8M parameters (about >5x smaller), 259 GFLOPs (as much as >5x
lower) and as much as >3.15x faster on a NVIDIA Jetson TX2 embedded processor.
| [
{
"version": "v1",
"created": "Fri, 21 Apr 2023 18:07:14 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Wong",
"Alexander",
""
],
[
"Wu",
"Yifan",
""
],
[
"Abbasi",
"Saad",
""
],
[
"Nair",
"Saeejith",
""
],
[
"Chen",
"Yuhao",
""
],
[
"Shafiee",
"Mohammad Javad",
""
]
] | TITLE: Fast GraspNeXt: A Fast Self-Attention Neural Network Architecture for
Multi-task Learning in Computer Vision Tasks for Robotic Grasping on the Edge
ABSTRACT: Multi-task learning has shown considerable promise for improving the
performance of deep learning-driven vision systems for the purpose of robotic
grasping. However, high architectural and computational complexity can result
in poor suitability for deployment on embedded devices that are typically
leveraged in robotic arms for real-world manufacturing and warehouse
environments. As such, the design of highly efficient multi-task deep neural
network architectures tailored for computer vision tasks for robotic grasping
on the edge is highly desired for widespread adoption in manufacturing
environments. Motivated by this, we propose Fast GraspNeXt, a fast
self-attention neural network architecture tailored for embedded multi-task
learning in computer vision tasks for robotic grasping. To build Fast
GraspNeXt, we leverage a generative network architecture search strategy with a
set of architectural constraints customized to achieve a strong balance between
multi-task learning performance and embedded inference efficiency. Experimental
results on the MetaGraspNet benchmark dataset show that the Fast GraspNeXt
network design achieves the highest performance (average precision (AP),
accuracy, and mean squared error (MSE)) across multiple computer vision tasks
when compared to other efficient multi-task network architecture designs, while
having only 17.8M parameters (about >5x smaller), 259 GFLOPs (as much as >5x
lower) and as much as >3.15x faster on a NVIDIA Jetson TX2 embedded processor.
| no_new_dataset | 0.710785 |
2304.11205 | Camille Coti | Camille Coti and Kevin Huck and Allen D. Malony | STaKTAU: profiling HPC applications' operating system usage | null | null | null | null | cs.DC | http://creativecommons.org/licenses/by/4.0/ | This paper presents a approach for measuring the time spent by HPC
applications in the operating system's kernel. We use the SystemTap interface
to insert timers before and after system calls, and take advantage of its
stability to design a tool that can be used with multiple versions of the
kernel. We evaluate its performance overhead, using an OS-intensive
mini-benchmark and a raytracing mini app.
| [
{
"version": "v1",
"created": "Fri, 21 Apr 2023 18:27:57 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Coti",
"Camille",
""
],
[
"Huck",
"Kevin",
""
],
[
"Malony",
"Allen D.",
""
]
] | TITLE: STaKTAU: profiling HPC applications' operating system usage
ABSTRACT: This paper presents a approach for measuring the time spent by HPC
applications in the operating system's kernel. We use the SystemTap interface
to insert timers before and after system calls, and take advantage of its
stability to design a tool that can be used with multiple versions of the
kernel. We evaluate its performance overhead, using an OS-intensive
mini-benchmark and a raytracing mini app.
| no_new_dataset | 0.706165 |
2304.11219 | Rishov Sarkar | Rishov Sarkar, Cong Hao | LightningSim: Fast and Accurate Trace-Based Simulation for High-Level
Synthesis | 11 pages, 7 figures. Accepted at FCCM 2023 | null | null | null | cs.PF cs.AR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | High-Level Synthesis allows hardware designers to create complex RTL designs
using C/C++. The traditional HLS workflow involves iterations of C/C++
simulation for partial functional verification and HLS synthesis for coarse
timing estimates. However, neither C/C++ simulation nor HLS synthesis estimates
can account for complex behaviors like FIFO interactions and pipeline stalls,
thereby obscuring problems like deadlocks and latency overheads. Such problems
are revealed only through C/RTL co-simulation, which is typically orders of
magnitude slower than either C/C++ simulation or HLS synthesis, far too slow to
integrate into the edit-run development cycle. Addressing this, we propose
LightningSim, a fast simulation tool for HLS that combines the speed of native
C/C++ with the accuracy of C/RTL co-simulation. LightningSim directly operates
on the LLVM intermediate representation (IR) code and accurately simulates a
hardware design's dynamic behavior. First, it traces LLVM IR execution to
capture the run-time information; second, it maps the static HLS scheduling
information to the trace to simulate the dynamic behavior; third, it calculates
stalls and deadlocks from inter-function interactions to get precise cycle
counts. Evaluated on 33 benchmarks, LightningSim produces 99.9%-accurate timing
estimates up to 95x faster than RTL simulation. Our code is publicly available
on GitHub.
| [
{
"version": "v1",
"created": "Fri, 21 Apr 2023 18:58:54 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Sarkar",
"Rishov",
""
],
[
"Hao",
"Cong",
""
]
] | TITLE: LightningSim: Fast and Accurate Trace-Based Simulation for High-Level
Synthesis
ABSTRACT: High-Level Synthesis allows hardware designers to create complex RTL designs
using C/C++. The traditional HLS workflow involves iterations of C/C++
simulation for partial functional verification and HLS synthesis for coarse
timing estimates. However, neither C/C++ simulation nor HLS synthesis estimates
can account for complex behaviors like FIFO interactions and pipeline stalls,
thereby obscuring problems like deadlocks and latency overheads. Such problems
are revealed only through C/RTL co-simulation, which is typically orders of
magnitude slower than either C/C++ simulation or HLS synthesis, far too slow to
integrate into the edit-run development cycle. Addressing this, we propose
LightningSim, a fast simulation tool for HLS that combines the speed of native
C/C++ with the accuracy of C/RTL co-simulation. LightningSim directly operates
on the LLVM intermediate representation (IR) code and accurately simulates a
hardware design's dynamic behavior. First, it traces LLVM IR execution to
capture the run-time information; second, it maps the static HLS scheduling
information to the trace to simulate the dynamic behavior; third, it calculates
stalls and deadlocks from inter-function interactions to get precise cycle
counts. Evaluated on 33 benchmarks, LightningSim produces 99.9%-accurate timing
estimates up to 95x faster than RTL simulation. Our code is publicly available
on GitHub.
| no_new_dataset | 0.711224 |
2304.11237 | Kai Jia | Kai Jia, Martin Rinard | Effective Neural Network $L_0$ Regularization With BinMask | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | $L_0$ regularization of neural networks is a fundamental problem. In addition
to regularizing models for better generalizability, $L_0$ regularization also
applies to selecting input features and training sparse neural networks. There
is a large body of research on related topics, some with quite complicated
methods. In this paper, we show that a straightforward formulation, BinMask,
which multiplies weights with deterministic binary masks and uses the identity
straight-through estimator for backpropagation, is an effective $L_0$
regularizer. We evaluate BinMask on three tasks: feature selection, network
sparsification, and model regularization. Despite its simplicity, BinMask
achieves competitive performance on all the benchmarks without task-specific
tuning compared to methods designed for each task. Our results suggest that
decoupling weights from mask optimization, which has been widely adopted by
previous work, is a key component for effective $L_0$ regularization.
| [
{
"version": "v1",
"created": "Fri, 21 Apr 2023 20:08:57 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Jia",
"Kai",
""
],
[
"Rinard",
"Martin",
""
]
] | TITLE: Effective Neural Network $L_0$ Regularization With BinMask
ABSTRACT: $L_0$ regularization of neural networks is a fundamental problem. In addition
to regularizing models for better generalizability, $L_0$ regularization also
applies to selecting input features and training sparse neural networks. There
is a large body of research on related topics, some with quite complicated
methods. In this paper, we show that a straightforward formulation, BinMask,
which multiplies weights with deterministic binary masks and uses the identity
straight-through estimator for backpropagation, is an effective $L_0$
regularizer. We evaluate BinMask on three tasks: feature selection, network
sparsification, and model regularization. Despite its simplicity, BinMask
achieves competitive performance on all the benchmarks without task-specific
tuning compared to methods designed for each task. Our results suggest that
decoupling weights from mask optimization, which has been widely adopted by
previous work, is a key component for effective $L_0$ regularization.
| no_new_dataset | 0.708799 |
2304.11238 | Aniket Pramanik | Aniket Pramanik, Sampada Bhave, Saurav Sajib, Samir D. Sharma, Mathews
Jacob | Adapting model-based deep learning to multiple acquisition conditions:
Ada-MoDL | null | null | null | null | eess.IV cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Purpose: The aim of this work is to introduce a single model-based deep
network that can provide high-quality reconstructions from undersampled
parallel MRI data acquired with multiple sequences, acquisition settings and
field strengths.
Methods: A single unrolled architecture, which offers good reconstructions
for multiple acquisition settings, is introduced. The proposed scheme adapts
the model to each setting by scaling the CNN features and the regularization
parameter with appropriate weights. The scaling weights and regularization
parameter are derived using a multi-layer perceptron model from conditional
vectors, which represents the specific acquisition setting. The perceptron
parameters and the CNN weights are jointly trained using data from multiple
acquisition settings, including differences in field strengths, acceleration,
and contrasts. The conditional network is validated using datasets acquired
with different acquisition settings.
Results: The comparison of the adaptive framework, which trains a single
model using the data from all the settings, shows that it can offer
consistently improved performance for each acquisition condition. The
comparison of the proposed scheme with networks that are trained independently
for each acquisition setting shows that it requires less training data per
acquisition setting to offer good performance.
Conclusion: The Ada-MoDL framework enables the use of a single model-based
unrolled network for multiple acquisition settings. In addition to eliminating
the need to train and store multiple networks for different acquisition
settings, this approach reduces the training data needed for each acquisition
setting.
| [
{
"version": "v1",
"created": "Fri, 21 Apr 2023 20:12:59 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Pramanik",
"Aniket",
""
],
[
"Bhave",
"Sampada",
""
],
[
"Sajib",
"Saurav",
""
],
[
"Sharma",
"Samir D.",
""
],
[
"Jacob",
"Mathews",
""
]
] | TITLE: Adapting model-based deep learning to multiple acquisition conditions:
Ada-MoDL
ABSTRACT: Purpose: The aim of this work is to introduce a single model-based deep
network that can provide high-quality reconstructions from undersampled
parallel MRI data acquired with multiple sequences, acquisition settings and
field strengths.
Methods: A single unrolled architecture, which offers good reconstructions
for multiple acquisition settings, is introduced. The proposed scheme adapts
the model to each setting by scaling the CNN features and the regularization
parameter with appropriate weights. The scaling weights and regularization
parameter are derived using a multi-layer perceptron model from conditional
vectors, which represents the specific acquisition setting. The perceptron
parameters and the CNN weights are jointly trained using data from multiple
acquisition settings, including differences in field strengths, acceleration,
and contrasts. The conditional network is validated using datasets acquired
with different acquisition settings.
Results: The comparison of the adaptive framework, which trains a single
model using the data from all the settings, shows that it can offer
consistently improved performance for each acquisition condition. The
comparison of the proposed scheme with networks that are trained independently
for each acquisition setting shows that it requires less training data per
acquisition setting to offer good performance.
Conclusion: The Ada-MoDL framework enables the use of a single model-based
unrolled network for multiple acquisition settings. In addition to eliminating
the need to train and store multiple networks for different acquisition
settings, this approach reduces the training data needed for each acquisition
setting.
| no_new_dataset | 0.710427 |
2304.11272 | Dmitry Zinoviev | Dmitry Zinoviev, Arkapravo Sarkar, Pelin Bicen | The "Non-Musk Effect" at Twitter | 17 pages, 7 figures | null | null | null | cs.SI cs.CY | http://creativecommons.org/licenses/by/4.0/ | Elon Musk has long been known to significantly impact Wall Street through his
controversial statements and actions, particularly through his own use of
social media. An innovator and visionary entrepreneur, Musk is often considered
a poster boy for all entrepreneurs worldwide. It is, thus, interesting to
examine the effect that Musk might have on Main Street, i.e., on the social
media activity of other entrepreneurs. In this research, we study and quantify
this "Musk Effect," i.e., the impact of Musk's recent and highly publicized
acquisition of Twitter on the tweeting activity of entrepreneurs. Using a
dataset consisting of 9.94 million actual tweets from 47,190 self-declared
entrepreneurs from seven English-speaking countries (US, Australia, New
Zealand, UK, Canada, South Africa, and Ireland) spanning 71 weeks and
encompassing the entire period from the rumor that Musk may buy Twitter till
the completion of the acquisition, we find that only about 2.5% of the
entrepreneurs display a significant change in their tweeting behavior over the
time.
We believe that our study is one of the first works to examine the effect of
Musk's acquisition of Twitter on the actual tweeting behavior of Twitter users
(entrepreneurs). By quantifying the impact of the Musk Effect on Main Street,
we provide a comparison with the effect Musk's actions have on Wall Street.
Finally, our systematic identification of the characteristics of entrepreneurs
most affected by the Musk Effect has practical implications for academics and
practitioners alike.
| [
{
"version": "v1",
"created": "Fri, 21 Apr 2023 23:37:18 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Zinoviev",
"Dmitry",
""
],
[
"Sarkar",
"Arkapravo",
""
],
[
"Bicen",
"Pelin",
""
]
] | TITLE: The "Non-Musk Effect" at Twitter
ABSTRACT: Elon Musk has long been known to significantly impact Wall Street through his
controversial statements and actions, particularly through his own use of
social media. An innovator and visionary entrepreneur, Musk is often considered
a poster boy for all entrepreneurs worldwide. It is, thus, interesting to
examine the effect that Musk might have on Main Street, i.e., on the social
media activity of other entrepreneurs. In this research, we study and quantify
this "Musk Effect," i.e., the impact of Musk's recent and highly publicized
acquisition of Twitter on the tweeting activity of entrepreneurs. Using a
dataset consisting of 9.94 million actual tweets from 47,190 self-declared
entrepreneurs from seven English-speaking countries (US, Australia, New
Zealand, UK, Canada, South Africa, and Ireland) spanning 71 weeks and
encompassing the entire period from the rumor that Musk may buy Twitter till
the completion of the acquisition, we find that only about 2.5% of the
entrepreneurs display a significant change in their tweeting behavior over the
time.
We believe that our study is one of the first works to examine the effect of
Musk's acquisition of Twitter on the actual tweeting behavior of Twitter users
(entrepreneurs). By quantifying the impact of the Musk Effect on Main Street,
we provide a comparison with the effect Musk's actions have on Wall Street.
Finally, our systematic identification of the characteristics of entrepreneurs
most affected by the Musk Effect has practical implications for academics and
practitioners alike.
| new_dataset | 0.505859 |
2304.11273 | Krishna Kumar | Chahak Mehta, Krishna Kumar | Enabling knowledge discovery in natural hazard engineering datasets on
DesignSafe | null | null | null | null | physics.geo-ph cs.IR | http://creativecommons.org/licenses/by/4.0/ | Data-driven discoveries require identifying relevant data relationships from
a sea of complex, unstructured, and heterogeneous scientific data. We propose a
hybrid methodology that extracts metadata and leverages scientific domain
knowledge to synthesize a new dataset from the original to construct knowledge
graphs. We demonstrate our approach's effectiveness through a case study on the
natural hazard engineering dataset on ``LEAP Liquefaction'' hosted on
DesignSafe. Traditional lexical search on DesignSafe is limited in uncovering
hidden relationships within the data. Our knowledge graph enables complex
queries and fosters new scientific insights by accurately identifying relevant
entities and establishing their relationships within the dataset. This
innovative implementation can transform the landscape of data-driven
discoveries across various scientific domains.
| [
{
"version": "v1",
"created": "Fri, 21 Apr 2023 23:40:57 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Mehta",
"Chahak",
""
],
[
"Kumar",
"Krishna",
""
]
] | TITLE: Enabling knowledge discovery in natural hazard engineering datasets on
DesignSafe
ABSTRACT: Data-driven discoveries require identifying relevant data relationships from
a sea of complex, unstructured, and heterogeneous scientific data. We propose a
hybrid methodology that extracts metadata and leverages scientific domain
knowledge to synthesize a new dataset from the original to construct knowledge
graphs. We demonstrate our approach's effectiveness through a case study on the
natural hazard engineering dataset on ``LEAP Liquefaction'' hosted on
DesignSafe. Traditional lexical search on DesignSafe is limited in uncovering
hidden relationships within the data. Our knowledge graph enables complex
queries and fosters new scientific insights by accurately identifying relevant
entities and establishing their relationships within the dataset. This
innovative implementation can transform the landscape of data-driven
discoveries across various scientific domains.
| no_new_dataset | 0.710635 |
2304.11278 | Kaustav Bhattacharjee | Kaustav Bhattacharjee and Aritra Dasgupta | Power to the Data Defenders: Human-Centered Disclosure Risk Calibration
of Open Data | In Proceedings of the Symposium on Usable Security and Privacy (USEC)
2023 | Proceedings of Symposium on Usable Security and Privacy (USEC)
2023 | 10.14722/usec.2023.237256 | null | cs.CR | http://creativecommons.org/licenses/by/4.0/ | The open data ecosystem is susceptible to vulnerabilities due to disclosure
risks. Though the datasets are anonymized during release, the prevalence of the
release-and-forget model makes the data defenders blind to privacy issues
arising after the dataset release. One such issue can be the disclosure risks
in the presence of newly released datasets which may compromise the privacy of
the data subjects of the anonymous open datasets. In this paper, we first
examine some of these pitfalls through the examples we observed during a red
teaming exercise and then envision other possible vulnerabilities in this
context. We also discuss proactive risk monitoring, including developing a
collection of highly susceptible open datasets and a visual analytic workflow
that empowers data defenders towards undertaking dynamic risk calibration
strategies.
| [
{
"version": "v1",
"created": "Fri, 21 Apr 2023 23:53:08 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Bhattacharjee",
"Kaustav",
""
],
[
"Dasgupta",
"Aritra",
""
]
] | TITLE: Power to the Data Defenders: Human-Centered Disclosure Risk Calibration
of Open Data
ABSTRACT: The open data ecosystem is susceptible to vulnerabilities due to disclosure
risks. Though the datasets are anonymized during release, the prevalence of the
release-and-forget model makes the data defenders blind to privacy issues
arising after the dataset release. One such issue can be the disclosure risks
in the presence of newly released datasets which may compromise the privacy of
the data subjects of the anonymous open datasets. In this paper, we first
examine some of these pitfalls through the examples we observed during a red
teaming exercise and then envision other possible vulnerabilities in this
context. We also discuss proactive risk monitoring, including developing a
collection of highly susceptible open datasets and a visual analytic workflow
that empowers data defenders towards undertaking dynamic risk calibration
strategies.
| no_new_dataset | 0.674908 |
2304.11291 | Noreen Anwar | Noreen Anwar, Philippe Duplessis-Guindon, Guillaume-Alexandre Bilodeau
and Wassim Bouachir | VisiTherS: Visible-thermal infrared stereo disparity estimation of human
silhouette | 8 pages,3 Figures,CVPR workshop | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents a novel approach for visible-thermal infrared
stereoscopy, focusing on the estimation of disparities of human silhouettes.
Visible-thermal infrared stereo poses several challenges, including occlusions
and differently textured matching regions in both spectra. Finding matches
between two spectra with varying colors, textures, and shapes adds further
complexity to the task. To address the aforementioned challenges, this paper
proposes a novel approach where a high-resolution convolutional neural network
is used to better capture relationships between the two spectra. To do so, a
modified HRNet backbone is used for feature extraction. This HRNet backbone is
capable of capturing fine details and textures as it extracts features at
multiple scales, thereby enabling the utilization of both local and global
information. For matching visible and thermal infrared regions, our method
extracts features on each patch using two modified HRNet streams. Features from
the two streams are then combined for predicting the disparities by
concatenation and correlation. Results on public datasets demonstrate the
effectiveness of the proposed approach by improving the results by
approximately 18 percentage points on the $\leq$ 1 pixel error, highlighting
its potential for improving accuracy in this task. The code of VisiTherS is
available on GitHub at the following link
https://github.com/philippeDG/VisiTherS.
| [
{
"version": "v1",
"created": "Sat, 22 Apr 2023 01:53:28 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Anwar",
"Noreen",
""
],
[
"Duplessis-Guindon",
"Philippe",
""
],
[
"Bilodeau",
"Guillaume-Alexandre",
""
],
[
"Bouachir",
"Wassim",
""
]
] | TITLE: VisiTherS: Visible-thermal infrared stereo disparity estimation of human
silhouette
ABSTRACT: This paper presents a novel approach for visible-thermal infrared
stereoscopy, focusing on the estimation of disparities of human silhouettes.
Visible-thermal infrared stereo poses several challenges, including occlusions
and differently textured matching regions in both spectra. Finding matches
between two spectra with varying colors, textures, and shapes adds further
complexity to the task. To address the aforementioned challenges, this paper
proposes a novel approach where a high-resolution convolutional neural network
is used to better capture relationships between the two spectra. To do so, a
modified HRNet backbone is used for feature extraction. This HRNet backbone is
capable of capturing fine details and textures as it extracts features at
multiple scales, thereby enabling the utilization of both local and global
information. For matching visible and thermal infrared regions, our method
extracts features on each patch using two modified HRNet streams. Features from
the two streams are then combined for predicting the disparities by
concatenation and correlation. Results on public datasets demonstrate the
effectiveness of the proposed approach by improving the results by
approximately 18 percentage points on the $\leq$ 1 pixel error, highlighting
its potential for improving accuracy in this task. The code of VisiTherS is
available on GitHub at the following link
https://github.com/philippeDG/VisiTherS.
| no_new_dataset | 0.712057 |
2304.11295 | Ding Li | Ding Li and Qichao Zhang and Zhongpu Xia and Kuan Zhang and Menglong
Yi and Wenda Jin and Dongbin Zhao | Planning-inspired Hierarchical Trajectory Prediction for Autonomous
Driving | 9 pages, 4 figures | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recently, anchor-based trajectory prediction methods have shown promising
performance, which directly selects a final set of anchors as future intents in
the spatio-temporal coupled space. However, such methods typically neglect a
deeper semantic interpretation of path intents and suffer from inferior
performance under the imperfect High-Definition (HD) map. To address this
challenge, we propose a novel Planning-inspired Hierarchical (PiH) trajectory
prediction framework that selects path and speed intents through a hierarchical
lateral and longitudinal decomposition. Especially, a hybrid lateral predictor
is presented to select a set of fixed-distance lateral paths from map-based
road-following and cluster-based free-move path candidates. {Then, the
subsequent longitudinal predictor selects plausible goals sampled from a set of
lateral paths as speed intents.} Finally, a trajectory decoder is given to
generate future trajectories conditioned on a categorical distribution over
lateral-longitudinal intents. Experiments demonstrate that PiH achieves
competitive and more balanced results against state-of-the-art methods on the
Argoverse motion forecasting benchmark and has the strongest robustness under
the imperfect HD map.
| [
{
"version": "v1",
"created": "Sat, 22 Apr 2023 02:17:52 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Li",
"Ding",
""
],
[
"Zhang",
"Qichao",
""
],
[
"Xia",
"Zhongpu",
""
],
[
"Zhang",
"Kuan",
""
],
[
"Yi",
"Menglong",
""
],
[
"Jin",
"Wenda",
""
],
[
"Zhao",
"Dongbin",
""
]
] | TITLE: Planning-inspired Hierarchical Trajectory Prediction for Autonomous
Driving
ABSTRACT: Recently, anchor-based trajectory prediction methods have shown promising
performance, which directly selects a final set of anchors as future intents in
the spatio-temporal coupled space. However, such methods typically neglect a
deeper semantic interpretation of path intents and suffer from inferior
performance under the imperfect High-Definition (HD) map. To address this
challenge, we propose a novel Planning-inspired Hierarchical (PiH) trajectory
prediction framework that selects path and speed intents through a hierarchical
lateral and longitudinal decomposition. Especially, a hybrid lateral predictor
is presented to select a set of fixed-distance lateral paths from map-based
road-following and cluster-based free-move path candidates. {Then, the
subsequent longitudinal predictor selects plausible goals sampled from a set of
lateral paths as speed intents.} Finally, a trajectory decoder is given to
generate future trajectories conditioned on a categorical distribution over
lateral-longitudinal intents. Experiments demonstrate that PiH achieves
competitive and more balanced results against state-of-the-art methods on the
Argoverse motion forecasting benchmark and has the strongest robustness under
the imperfect HD map.
| no_new_dataset | 0.71027 |
2304.11341 | Ziyang Song | Ziyang Song, Zheng Shi, Jiaji Su, Qingping Dou, Guanghua Yang,
Haichuan Ding, and Shaodan Ma | Performance Analysis and Optimal Design of HARQ-IR-Aided Terahertz
Communications | Blockage, hybrid automatic repeat request (HARQ), outage probability,
terahertz (THz) communications | null | null | null | cs.IT eess.SP math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Terahertz (THz) communications are envisioned to be a promising technology
for 6G thanks to its broad bandwidth. However, the large path loss, antenna
misalignment, and atmospheric influence of THz communications severely
deteriorate its reliability. To address this, hybrid automatic repeat request
(HARQ) is recognized as an effective technique to ensure reliable THz
communications. This paper delves into the performance analysis of HARQ with
incremental redundancy (HARQ-IR)-aided THz communications in the
presence/absence of blockage. More specifically, the analytical expression of
the outage probability of HARQ-IR-aided THz communications is derived, with
which the asymptotic outage analysis is enabled to gain meaningful insights,
including diversity order, power allocation gain, modulation and coding gain,
etc. Then the long term average throughput (LTAT) is expressed in terms of the
outage probability based on renewal theory. Moreover, to combat the blockage
effects, a multi-hop HARQ-IR-aided THz communication scheme is proposed and its
performance is examined. To demonstrate the superiority of the proposed scheme,
the other two HARQ-aided schemes, i.e., Type-I HARQ and HARQ with chase
combining (HARQ-CC), are used for benchmarking in the simulations. In addition,
a deep neural network (DNN) based outage evaluation framework with low
computational complexity is devised to reap the benefits of using both
asymptotic and simulation results in low and high outage regimes, respectively.
This novel outage evaluation framework is finally employed for the optimal rate
selection, which outperforms the asymptotic based optimization.
| [
{
"version": "v1",
"created": "Sat, 22 Apr 2023 07:44:06 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Song",
"Ziyang",
""
],
[
"Shi",
"Zheng",
""
],
[
"Su",
"Jiaji",
""
],
[
"Dou",
"Qingping",
""
],
[
"Yang",
"Guanghua",
""
],
[
"Ding",
"Haichuan",
""
],
[
"Ma",
"Shaodan",
""
]
] | TITLE: Performance Analysis and Optimal Design of HARQ-IR-Aided Terahertz
Communications
ABSTRACT: Terahertz (THz) communications are envisioned to be a promising technology
for 6G thanks to its broad bandwidth. However, the large path loss, antenna
misalignment, and atmospheric influence of THz communications severely
deteriorate its reliability. To address this, hybrid automatic repeat request
(HARQ) is recognized as an effective technique to ensure reliable THz
communications. This paper delves into the performance analysis of HARQ with
incremental redundancy (HARQ-IR)-aided THz communications in the
presence/absence of blockage. More specifically, the analytical expression of
the outage probability of HARQ-IR-aided THz communications is derived, with
which the asymptotic outage analysis is enabled to gain meaningful insights,
including diversity order, power allocation gain, modulation and coding gain,
etc. Then the long term average throughput (LTAT) is expressed in terms of the
outage probability based on renewal theory. Moreover, to combat the blockage
effects, a multi-hop HARQ-IR-aided THz communication scheme is proposed and its
performance is examined. To demonstrate the superiority of the proposed scheme,
the other two HARQ-aided schemes, i.e., Type-I HARQ and HARQ with chase
combining (HARQ-CC), are used for benchmarking in the simulations. In addition,
a deep neural network (DNN) based outage evaluation framework with low
computational complexity is devised to reap the benefits of using both
asymptotic and simulation results in low and high outage regimes, respectively.
This novel outage evaluation framework is finally employed for the optimal rate
selection, which outperforms the asymptotic based optimization.
| no_new_dataset | 0.709239 |
2304.11367 | Zhenkun Zhou | Xingyu Peng, Zhenkun Zhou, Chong Zhang, Ke Xu | Detecting Political Opinions in Tweets through Bipartite Graph Analysis:
A Skip Aggregation Graph Convolution Approach | null | null | null | null | cs.SI cs.AI | http://creativecommons.org/licenses/by/4.0/ | Public opinion is a crucial factor in shaping political decision-making.
Nowadays, social media has become an essential platform for individuals to
engage in political discussions and express their political views, presenting
researchers with an invaluable resource for analyzing public opinion. In this
paper, we focus on the 2020 US presidential election and create a large-scale
dataset from Twitter. To detect political opinions in tweets, we build a
user-tweet bipartite graph based on users' posting and retweeting behaviors and
convert the task into a Graph Neural Network (GNN)-based node classification
problem. Then, we introduce a novel skip aggregation mechanism that makes tweet
nodes aggregate information from second-order neighbors, which are also tweet
nodes due to the graph's bipartite nature, effectively leveraging user
behavioral information. The experimental results show that our proposed model
significantly outperforms several competitive baselines. Further analyses
demonstrate the significance of user behavioral information and the
effectiveness of skip aggregation.
| [
{
"version": "v1",
"created": "Sat, 22 Apr 2023 10:38:35 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Peng",
"Xingyu",
""
],
[
"Zhou",
"Zhenkun",
""
],
[
"Zhang",
"Chong",
""
],
[
"Xu",
"Ke",
""
]
] | TITLE: Detecting Political Opinions in Tweets through Bipartite Graph Analysis:
A Skip Aggregation Graph Convolution Approach
ABSTRACT: Public opinion is a crucial factor in shaping political decision-making.
Nowadays, social media has become an essential platform for individuals to
engage in political discussions and express their political views, presenting
researchers with an invaluable resource for analyzing public opinion. In this
paper, we focus on the 2020 US presidential election and create a large-scale
dataset from Twitter. To detect political opinions in tweets, we build a
user-tweet bipartite graph based on users' posting and retweeting behaviors and
convert the task into a Graph Neural Network (GNN)-based node classification
problem. Then, we introduce a novel skip aggregation mechanism that makes tweet
nodes aggregate information from second-order neighbors, which are also tweet
nodes due to the graph's bipartite nature, effectively leveraging user
behavioral information. The experimental results show that our proposed model
significantly outperforms several competitive baselines. Further analyses
demonstrate the significance of user behavioral information and the
effectiveness of skip aggregation.
| no_new_dataset | 0.703537 |
2304.11375 | Yuxing Chen | Yuxing Chen, Lorenzo Bruzzone | Unsupervised CD in satellite image time series by contrastive learning
and feature tracking | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | While unsupervised change detection using contrastive learning has been
significantly improved the performance of literature techniques, at present, it
only focuses on the bi-temporal change detection scenario. Previous
state-of-the-art models for image time-series change detection often use
features obtained by learning for clustering or training a model from scratch
using pseudo labels tailored to each scene. However, these approaches fail to
exploit the spatial-temporal information of image time-series or generalize to
unseen scenarios. In this work, we propose a two-stage approach to unsupervised
change detection in satellite image time-series using contrastive learning with
feature tracking. By deriving pseudo labels from pre-trained models and using
feature tracking to propagate them among the image time-series, we improve the
consistency of our pseudo labels and address the challenges of seasonal changes
in long-term remote sensing image time-series. We adopt the self-training
algorithm with ConvLSTM on the obtained pseudo labels, where we first use
supervised contrastive loss and contrastive random walks to further improve the
feature correspondence in space-time. Then a fully connected layer is
fine-tuned on the pre-trained multi-temporal features for generating the final
change maps. Through comprehensive experiments on two datasets, we demonstrate
consistent improvements in accuracy on fitting and inference scenarios.
| [
{
"version": "v1",
"created": "Sat, 22 Apr 2023 11:19:19 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Chen",
"Yuxing",
""
],
[
"Bruzzone",
"Lorenzo",
""
]
] | TITLE: Unsupervised CD in satellite image time series by contrastive learning
and feature tracking
ABSTRACT: While unsupervised change detection using contrastive learning has been
significantly improved the performance of literature techniques, at present, it
only focuses on the bi-temporal change detection scenario. Previous
state-of-the-art models for image time-series change detection often use
features obtained by learning for clustering or training a model from scratch
using pseudo labels tailored to each scene. However, these approaches fail to
exploit the spatial-temporal information of image time-series or generalize to
unseen scenarios. In this work, we propose a two-stage approach to unsupervised
change detection in satellite image time-series using contrastive learning with
feature tracking. By deriving pseudo labels from pre-trained models and using
feature tracking to propagate them among the image time-series, we improve the
consistency of our pseudo labels and address the challenges of seasonal changes
in long-term remote sensing image time-series. We adopt the self-training
algorithm with ConvLSTM on the obtained pseudo labels, where we first use
supervised contrastive loss and contrastive random walks to further improve the
feature correspondence in space-time. Then a fully connected layer is
fine-tuned on the pre-trained multi-temporal features for generating the final
change maps. Through comprehensive experiments on two datasets, we demonstrate
consistent improvements in accuracy on fitting and inference scenarios.
| no_new_dataset | 0.710641 |
2304.11377 | Sibi Chakkaravarthy S | Sibi Chakkaravarthy Sethuraman, Gaurav Reddy Tadkapally, Athresh
Kiran, Saraju P. Mohanty, Anitha Subramanian | SimplyMime: A Control at Our Fingertips | null | null | null | null | cs.HC cs.AI | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The utilization of consumer electronics, such as televisions, set-top boxes,
home theaters, and air conditioners, has become increasingly prevalent in
modern society as technology continues to evolve. As new devices enter our
homes each year, the accumulation of multiple infrared remote controls to
operate them not only results in a waste of energy and resources, but also
creates a cumbersome and cluttered environment for the user. This paper
presents a novel system, named SimplyMime, which aims to eliminate the need for
multiple remote controls for consumer electronics and provide the user with
intuitive control without the need for additional devices. SimplyMime leverages
a dynamic hand gesture recognition architecture, incorporating Artificial
Intelligence and Human-Computer Interaction, to create a sophisticated system
that enables users to interact with a vast majority of consumer electronics
with ease. Additionally, SimplyMime has a security aspect where it can verify
and authenticate the user utilising the palmprint, which ensures that only
authorized users can control the devices. The performance of the proposed
method for detecting and recognizing gestures in a stream of motion was
thoroughly tested and validated using multiple benchmark datasets, resulting in
commendable accuracy levels. One of the distinct advantages of the proposed
method is its minimal computational power requirements, making it highly
adaptable and reliable in a wide range of circumstances. The paper proposes
incorporating this technology into all consumer electronic devices that
currently require a secondary remote for operation, thus promoting a more
efficient and sustainable living environment.
| [
{
"version": "v1",
"created": "Sat, 22 Apr 2023 11:25:19 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Sethuraman",
"Sibi Chakkaravarthy",
""
],
[
"Tadkapally",
"Gaurav Reddy",
""
],
[
"Kiran",
"Athresh",
""
],
[
"Mohanty",
"Saraju P.",
""
],
[
"Subramanian",
"Anitha",
""
]
] | TITLE: SimplyMime: A Control at Our Fingertips
ABSTRACT: The utilization of consumer electronics, such as televisions, set-top boxes,
home theaters, and air conditioners, has become increasingly prevalent in
modern society as technology continues to evolve. As new devices enter our
homes each year, the accumulation of multiple infrared remote controls to
operate them not only results in a waste of energy and resources, but also
creates a cumbersome and cluttered environment for the user. This paper
presents a novel system, named SimplyMime, which aims to eliminate the need for
multiple remote controls for consumer electronics and provide the user with
intuitive control without the need for additional devices. SimplyMime leverages
a dynamic hand gesture recognition architecture, incorporating Artificial
Intelligence and Human-Computer Interaction, to create a sophisticated system
that enables users to interact with a vast majority of consumer electronics
with ease. Additionally, SimplyMime has a security aspect where it can verify
and authenticate the user utilising the palmprint, which ensures that only
authorized users can control the devices. The performance of the proposed
method for detecting and recognizing gestures in a stream of motion was
thoroughly tested and validated using multiple benchmark datasets, resulting in
commendable accuracy levels. One of the distinct advantages of the proposed
method is its minimal computational power requirements, making it highly
adaptable and reliable in a wide range of circumstances. The paper proposes
incorporating this technology into all consumer electronic devices that
currently require a secondary remote for operation, thus promoting a more
efficient and sustainable living environment.
| no_new_dataset | 0.708219 |
2304.11381 | Yuxing Chen | Yuxing Chen, Maofan Zhao, Lorenzo Bruzzone | Incomplete Multimodal Learning for Remote Sensing Data Fusion | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | The mechanism of connecting multimodal signals through self-attention
operation is a key factor in the success of multimodal Transformer networks in
remote sensing data fusion tasks. However, traditional approaches assume access
to all modalities during both training and inference, which can lead to severe
degradation when dealing with modal-incomplete inputs in downstream
applications. To address this limitation, our proposed approach introduces a
novel model for incomplete multimodal learning in the context of remote sensing
data fusion. This approach can be used in both supervised and self-supervised
pretraining paradigms and leverages the additional learned fusion tokens in
combination with Bi-LSTM attention and masked self-attention mechanisms to
collect multimodal signals. The proposed approach employs reconstruction and
contrastive loss to facilitate fusion in pre-training while allowing for random
modality combinations as inputs in network training. Our approach delivers
state-of-the-art performance on two multimodal datasets for tasks such as
building instance / semantic segmentation and land-cover mapping tasks when
dealing with incomplete inputs during inference.
| [
{
"version": "v1",
"created": "Sat, 22 Apr 2023 12:16:52 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Chen",
"Yuxing",
""
],
[
"Zhao",
"Maofan",
""
],
[
"Bruzzone",
"Lorenzo",
""
]
] | TITLE: Incomplete Multimodal Learning for Remote Sensing Data Fusion
ABSTRACT: The mechanism of connecting multimodal signals through self-attention
operation is a key factor in the success of multimodal Transformer networks in
remote sensing data fusion tasks. However, traditional approaches assume access
to all modalities during both training and inference, which can lead to severe
degradation when dealing with modal-incomplete inputs in downstream
applications. To address this limitation, our proposed approach introduces a
novel model for incomplete multimodal learning in the context of remote sensing
data fusion. This approach can be used in both supervised and self-supervised
pretraining paradigms and leverages the additional learned fusion tokens in
combination with Bi-LSTM attention and masked self-attention mechanisms to
collect multimodal signals. The proposed approach employs reconstruction and
contrastive loss to facilitate fusion in pre-training while allowing for random
modality combinations as inputs in network training. Our approach delivers
state-of-the-art performance on two multimodal datasets for tasks such as
building instance / semantic segmentation and land-cover mapping tasks when
dealing with incomplete inputs during inference.
| no_new_dataset | 0.710239 |
2304.11391 | Zhenhao Li | Zhenhao Li, Chuan Luo, Tse-Hsun Chen, Weiyi Shang, Shilin He, Qingwei
Lin, Dongmei Zhang | Did We Miss Something Important? Studying and Exploring Variable-Aware
Log Abstraction | null | null | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Due to the sheer size of software logs, developers rely on automated
techniques for log analysis. One of the first and most important steps of
automated log analysis is log abstraction, which parses the raw logs into a
structured format. Prior log abstraction techniques aim to identify and
abstract all the dynamic variables in logs and output a static log template for
automated log analysis. However, these abstracted dynamic variables may also
contain important information that is useful to different tasks in log
analysis. In this paper, we investigate the characteristics of dynamic
variables and their importance in practice, and explore the potential of a
variable-aware log abstraction technique. Through manual investigations and
surveys with practitioners, we find that different categories of dynamic
variables record various information that can be important depending on the
given tasks, the distinction of dynamic variables in log abstraction can
further assist in log analysis. We then propose a deep learning based log
abstraction approach, named VALB, which can identify different categories of
dynamic variables and preserve the value of specified categories of dynamic
variables along with the log templates (i.e., variable-aware log abstraction).
Through the evaluation on a widely used log abstraction benchmark, we find that
VALB outperforms other state-of-the-art log abstraction techniques on general
log abstraction (i.e., when abstracting all the dynamic variables) and also
achieves a high variable-aware log abstraction accuracy that further identifies
the category of the dynamic variables. Our study highlights the potential of
leveraging the important information recorded in the dynamic variables to
further improve the process of log analysis.
| [
{
"version": "v1",
"created": "Sat, 22 Apr 2023 12:55:00 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Li",
"Zhenhao",
""
],
[
"Luo",
"Chuan",
""
],
[
"Chen",
"Tse-Hsun",
""
],
[
"Shang",
"Weiyi",
""
],
[
"He",
"Shilin",
""
],
[
"Lin",
"Qingwei",
""
],
[
"Zhang",
"Dongmei",
""
]
] | TITLE: Did We Miss Something Important? Studying and Exploring Variable-Aware
Log Abstraction
ABSTRACT: Due to the sheer size of software logs, developers rely on automated
techniques for log analysis. One of the first and most important steps of
automated log analysis is log abstraction, which parses the raw logs into a
structured format. Prior log abstraction techniques aim to identify and
abstract all the dynamic variables in logs and output a static log template for
automated log analysis. However, these abstracted dynamic variables may also
contain important information that is useful to different tasks in log
analysis. In this paper, we investigate the characteristics of dynamic
variables and their importance in practice, and explore the potential of a
variable-aware log abstraction technique. Through manual investigations and
surveys with practitioners, we find that different categories of dynamic
variables record various information that can be important depending on the
given tasks, the distinction of dynamic variables in log abstraction can
further assist in log analysis. We then propose a deep learning based log
abstraction approach, named VALB, which can identify different categories of
dynamic variables and preserve the value of specified categories of dynamic
variables along with the log templates (i.e., variable-aware log abstraction).
Through the evaluation on a widely used log abstraction benchmark, we find that
VALB outperforms other state-of-the-art log abstraction techniques on general
log abstraction (i.e., when abstracting all the dynamic variables) and also
achieves a high variable-aware log abstraction accuracy that further identifies
the category of the dynamic variables. Our study highlights the potential of
leveraging the important information recorded in the dynamic variables to
further improve the process of log analysis.
| no_new_dataset | 0.710791 |
2304.11393 | Feng Jiang | Feng Jiang, Heng Gao, Shoumeng Qiu, Haiqiang Zhang, Ru Wan and Jian Pu | Knowledge Distillation from 3D to Bird's-Eye-View for LiDAR Semantic
Segmentation | ICME 2023 Accepted | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | LiDAR point cloud segmentation is one of the most fundamental tasks for
autonomous driving scene understanding. However, it is difficult for existing
models to achieve both high inference speed and accuracy simultaneously. For
example, voxel-based methods perform well in accuracy, while Bird's-Eye-View
(BEV)-based methods can achieve real-time inference. To overcome this issue, we
develop an effective 3D-to-BEV knowledge distillation method that transfers
rich knowledge from 3D voxel-based models to BEV-based models. Our framework
mainly consists of two modules: the voxel-to-pillar distillation module and the
label-weight distillation module. Voxel-to-pillar distillation distills sparse
3D features to BEV features for middle layers to make the BEV-based model aware
of more structural and geometric information. Label-weight distillation helps
the model pay more attention to regions with more height information. Finally,
we conduct experiments on the SemanticKITTI dataset and Paris-Lille-3D. The
results on SemanticKITTI show more than 5% improvement on the test set,
especially for classes such as motorcycle and person, with more than 15%
improvement. The code can be accessed at
https://github.com/fengjiang5/Knowledge-Distillation-from-Cylinder3D-to-PolarNet.
| [
{
"version": "v1",
"created": "Sat, 22 Apr 2023 13:03:19 GMT"
}
] | 2023-04-25T00:00:00 | [
[
"Jiang",
"Feng",
""
],
[
"Gao",
"Heng",
""
],
[
"Qiu",
"Shoumeng",
""
],
[
"Zhang",
"Haiqiang",
""
],
[
"Wan",
"Ru",
""
],
[
"Pu",
"Jian",
""
]
] | TITLE: Knowledge Distillation from 3D to Bird's-Eye-View for LiDAR Semantic
Segmentation
ABSTRACT: LiDAR point cloud segmentation is one of the most fundamental tasks for
autonomous driving scene understanding. However, it is difficult for existing
models to achieve both high inference speed and accuracy simultaneously. For
example, voxel-based methods perform well in accuracy, while Bird's-Eye-View
(BEV)-based methods can achieve real-time inference. To overcome this issue, we
develop an effective 3D-to-BEV knowledge distillation method that transfers
rich knowledge from 3D voxel-based models to BEV-based models. Our framework
mainly consists of two modules: the voxel-to-pillar distillation module and the
label-weight distillation module. Voxel-to-pillar distillation distills sparse
3D features to BEV features for middle layers to make the BEV-based model aware
of more structural and geometric information. Label-weight distillation helps
the model pay more attention to regions with more height information. Finally,
we conduct experiments on the SemanticKITTI dataset and Paris-Lille-3D. The
results on SemanticKITTI show more than 5% improvement on the test set,
especially for classes such as motorcycle and person, with more than 15%
improvement. The code can be accessed at
https://github.com/fengjiang5/Knowledge-Distillation-from-Cylinder3D-to-PolarNet.
| no_new_dataset | 0.709453 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.