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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2306.09989 | Achyut Tiwari | Achyut Tiwari, Aryan Chugh, Aman Sharma | Ensemble Framework for Cardiovascular Disease Prediction | null | Computers in Biology and Medicine Volume 146, July 2022, 105624 | 10.1016/j.compbiomed.2022.105624 | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Heart disease is the major cause of non-communicable and silent death
worldwide. Heart diseases or cardiovascular diseases are classified into four
types: coronary heart disease, heart failure, congenital heart disease, and
cardiomyopathy. It is vital to diagnose heart disease early and accurately in
order to avoid further injury and save patients' lives. As a result, we need a
system that can predict cardiovascular disease before it becomes a critical
situation. Machine learning has piqued the interest of researchers in the field
of medical sciences. For heart disease prediction, researchers implement a
variety of machine learning methods and approaches. In this work, to the best
of our knowledge, we have used the dataset from IEEE Data Port which is one of
the online available largest datasets for cardiovascular diseases individuals.
The dataset isa combination of Hungarian, Cleveland, Long Beach VA, Switzerland
& Statlog datasets with important features such as Maximum Heart Rate Achieved,
Serum Cholesterol, Chest Pain Type, Fasting blood sugar, and so on. To assess
the efficacy and strength of the developed model, several performance measures
are used, such as ROC, AUC curve, specificity, F1-score, sensitivity, MCC, and
accuracy. In this study, we have proposed a framework with a stacked ensemble
classifier using several machine learning algorithms including ExtraTrees
Classifier, Random Forest, XGBoost, and so on. Our proposed framework attained
an accuracy of 92.34% which is higher than the existing literature.
| [
{
"version": "v1",
"created": "Fri, 16 Jun 2023 17:37:43 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Tiwari",
"Achyut",
""
],
[
"Chugh",
"Aryan",
""
],
[
"Sharma",
"Aman",
""
]
] | TITLE: Ensemble Framework for Cardiovascular Disease Prediction
ABSTRACT: Heart disease is the major cause of non-communicable and silent death
worldwide. Heart diseases or cardiovascular diseases are classified into four
types: coronary heart disease, heart failure, congenital heart disease, and
cardiomyopathy. It is vital to diagnose heart disease early and accurately in
order to avoid further injury and save patients' lives. As a result, we need a
system that can predict cardiovascular disease before it becomes a critical
situation. Machine learning has piqued the interest of researchers in the field
of medical sciences. For heart disease prediction, researchers implement a
variety of machine learning methods and approaches. In this work, to the best
of our knowledge, we have used the dataset from IEEE Data Port which is one of
the online available largest datasets for cardiovascular diseases individuals.
The dataset isa combination of Hungarian, Cleveland, Long Beach VA, Switzerland
& Statlog datasets with important features such as Maximum Heart Rate Achieved,
Serum Cholesterol, Chest Pain Type, Fasting blood sugar, and so on. To assess
the efficacy and strength of the developed model, several performance measures
are used, such as ROC, AUC curve, specificity, F1-score, sensitivity, MCC, and
accuracy. In this study, we have proposed a framework with a stacked ensemble
classifier using several machine learning algorithms including ExtraTrees
Classifier, Random Forest, XGBoost, and so on. Our proposed framework attained
an accuracy of 92.34% which is higher than the existing literature.
|
2306.11547 | Matthew McDermott | Matthew B. A. McDermott, Bret Nestor, Peniel Argaw, Isaac Kohane | Event Stream GPT: A Data Pre-processing and Modeling Library for
Generative, Pre-trained Transformers over Continuous-time Sequences of
Complex Events | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by-sa/4.0/ | Generative, pre-trained transformers (GPTs, a.k.a. "Foundation Models") have
reshaped natural language processing (NLP) through their versatility in diverse
downstream tasks. However, their potential extends far beyond NLP. This paper
provides a software utility to help realize this potential, extending the
applicability of GPTs to continuous-time sequences of complex events with
internal dependencies, such as medical record datasets. Despite their
potential, the adoption of foundation models in these domains has been hampered
by the lack of suitable tools for model construction and evaluation. To bridge
this gap, we introduce Event Stream GPT (ESGPT), an open-source library
designed to streamline the end-to-end process for building GPTs for
continuous-time event sequences. ESGPT allows users to (1) build flexible,
foundation-model scale input datasets by specifying only a minimal
configuration file, (2) leverage a Hugging Face compatible modeling API for
GPTs over this modality that incorporates intra-event causal dependency
structures and autoregressive generation capabilities, and (3) evaluate models
via standardized processes that can assess few and even zero-shot performance
of pre-trained models on user-specified fine-tuning tasks.
| [
{
"version": "v1",
"created": "Tue, 20 Jun 2023 14:01:29 GMT"
},
{
"version": "v2",
"created": "Wed, 21 Jun 2023 14:02:02 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"McDermott",
"Matthew B. A.",
""
],
[
"Nestor",
"Bret",
""
],
[
"Argaw",
"Peniel",
""
],
[
"Kohane",
"Isaac",
""
]
] | TITLE: Event Stream GPT: A Data Pre-processing and Modeling Library for
Generative, Pre-trained Transformers over Continuous-time Sequences of
Complex Events
ABSTRACT: Generative, pre-trained transformers (GPTs, a.k.a. "Foundation Models") have
reshaped natural language processing (NLP) through their versatility in diverse
downstream tasks. However, their potential extends far beyond NLP. This paper
provides a software utility to help realize this potential, extending the
applicability of GPTs to continuous-time sequences of complex events with
internal dependencies, such as medical record datasets. Despite their
potential, the adoption of foundation models in these domains has been hampered
by the lack of suitable tools for model construction and evaluation. To bridge
this gap, we introduce Event Stream GPT (ESGPT), an open-source library
designed to streamline the end-to-end process for building GPTs for
continuous-time event sequences. ESGPT allows users to (1) build flexible,
foundation-model scale input datasets by specifying only a minimal
configuration file, (2) leverage a Hugging Face compatible modeling API for
GPTs over this modality that incorporates intra-event causal dependency
structures and autoregressive generation capabilities, and (3) evaluate models
via standardized processes that can assess few and even zero-shot performance
of pre-trained models on user-specified fine-tuning tasks.
|
2306.11734 | Yuntian Chen | Qinglong Cao, Yuntian Chen, Chao Ma and Xiaokang Yang | Few-Shot Rotation-Invariant Aerial Image Semantic Segmentation | null | null | null | null | cs.CV eess.IV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Few-shot aerial image segmentation is a challenging task that involves
precisely parsing objects in query aerial images with limited annotated
support. Conventional matching methods without consideration of varying object
orientations can fail to activate same-category objects with different
orientations. Moreover, conventional algorithms can lead to false recognition
of lower-scored rotated semantic objects. In response to these challenges, the
authors propose a novel few-shot rotation-invariant aerial semantic
segmentation network (FRINet). FRINet matches each query feature
rotation-adaptively with orientation-varying yet category-consistent support
information. The segmentation predictions from different orientations are
supervised by the same label, and the backbones are pre-trained in the base
category to boost segmentation performance. Experimental results demonstrate
that FRINet achieves state-of-the-art performance in few-shot aerial semantic
segmentation benchmark.
| [
{
"version": "v1",
"created": "Mon, 29 May 2023 09:28:34 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Cao",
"Qinglong",
""
],
[
"Chen",
"Yuntian",
""
],
[
"Ma",
"Chao",
""
],
[
"Yang",
"Xiaokang",
""
]
] | TITLE: Few-Shot Rotation-Invariant Aerial Image Semantic Segmentation
ABSTRACT: Few-shot aerial image segmentation is a challenging task that involves
precisely parsing objects in query aerial images with limited annotated
support. Conventional matching methods without consideration of varying object
orientations can fail to activate same-category objects with different
orientations. Moreover, conventional algorithms can lead to false recognition
of lower-scored rotated semantic objects. In response to these challenges, the
authors propose a novel few-shot rotation-invariant aerial semantic
segmentation network (FRINet). FRINet matches each query feature
rotation-adaptively with orientation-varying yet category-consistent support
information. The segmentation predictions from different orientations are
supervised by the same label, and the backbones are pre-trained in the base
category to boost segmentation performance. Experimental results demonstrate
that FRINet achieves state-of-the-art performance in few-shot aerial semantic
segmentation benchmark.
|
2306.11746 | Jun Li | Jun Li, Yi Bin, Liang Peng, Yang Yang, Yangyang Li, Hao Jin, and Zi
Huang | Focusing on Relevant Responses for Multi-modal Rumor Detection | Submitted to TKDE | null | null | null | cs.SI cs.MM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In the absence of an authoritative statement about a rumor, people may expose
the truth behind such rumor through their responses on social media. Most rumor
detection methods aggregate the information of all the responses and have made
great progress. However, due to the different backgrounds of users, the
responses have different relevance for discovering th suspicious points hidden
in a rumor claim. The methods that focus on all the responding tweets would
dilute the effect of the critical ones. Moreover, for a multi-modal rumor
claim, the focus of a user may be on several words in the text or an object in
the image, so the different modalities should be considered to select the
relevant responses and verify the claim. In this paper, we propose a novel
multi-modal rumor detection model, termed Focal Reasoning Model (FoRM), to
filter out the irrelevant responses and further conduct fine-grained reasoning
with the multi-modal claim and corresponding responses. Concretely, there are
two main components in our FoRM: the coarse-grained selection and the
fine-grained reasoning. The coarse-grained selection component leverages the
post-level features of the responses to verify the claim and learns a relevant
score of each response. Based on the relevant scores, the most relevant
responses are reserved as the critical ones to the further reasoning. In the
fine-grained reasoning component, we design a relation attention module to
explore the fine-grained relations, i.e., token-to-token and token-to-object
relations, between the reserved responses and the multi-modal claim for finding
out the valuable clues. Extensive experiments have been conducted on two
real-world datasets, and the results demonstrate that our proposed model
outperforms all the baselines.
| [
{
"version": "v1",
"created": "Sun, 18 Jun 2023 04:39:46 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Li",
"Jun",
""
],
[
"Bin",
"Yi",
""
],
[
"Peng",
"Liang",
""
],
[
"Yang",
"Yang",
""
],
[
"Li",
"Yangyang",
""
],
[
"Jin",
"Hao",
""
],
[
"Huang",
"Zi",
""
]
] | TITLE: Focusing on Relevant Responses for Multi-modal Rumor Detection
ABSTRACT: In the absence of an authoritative statement about a rumor, people may expose
the truth behind such rumor through their responses on social media. Most rumor
detection methods aggregate the information of all the responses and have made
great progress. However, due to the different backgrounds of users, the
responses have different relevance for discovering th suspicious points hidden
in a rumor claim. The methods that focus on all the responding tweets would
dilute the effect of the critical ones. Moreover, for a multi-modal rumor
claim, the focus of a user may be on several words in the text or an object in
the image, so the different modalities should be considered to select the
relevant responses and verify the claim. In this paper, we propose a novel
multi-modal rumor detection model, termed Focal Reasoning Model (FoRM), to
filter out the irrelevant responses and further conduct fine-grained reasoning
with the multi-modal claim and corresponding responses. Concretely, there are
two main components in our FoRM: the coarse-grained selection and the
fine-grained reasoning. The coarse-grained selection component leverages the
post-level features of the responses to verify the claim and learns a relevant
score of each response. Based on the relevant scores, the most relevant
responses are reserved as the critical ones to the further reasoning. In the
fine-grained reasoning component, we design a relation attention module to
explore the fine-grained relations, i.e., token-to-token and token-to-object
relations, between the reserved responses and the multi-modal claim for finding
out the valuable clues. Extensive experiments have been conducted on two
real-world datasets, and the results demonstrate that our proposed model
outperforms all the baselines.
|
2306.11754 | Kamil Adamczewski | Kamil Adamczewski, Yingchen He, Mijung Park | Pre-Pruning and Gradient-Dropping Improve Differentially Private Image
Classification | arXiv admin note: text overlap with arXiv:2303.04612 | null | null | null | cs.CV cs.CR cs.LG | http://creativecommons.org/licenses/by-sa/4.0/ | Scalability is a significant challenge when it comes to applying differential
privacy to training deep neural networks. The commonly used DP-SGD algorithm
struggles to maintain a high level of privacy protection while achieving high
accuracy on even moderately sized models. To tackle this challenge, we take
advantage of the fact that neural networks are overparameterized, which allows
us to improve neural network training with differential privacy. Specifically,
we introduce a new training paradigm that uses \textit{pre-pruning} and
\textit{gradient-dropping} to reduce the parameter space and improve
scalability. The process starts with pre-pruning the parameters of the original
network to obtain a smaller model that is then trained with DP-SGD. During
training, less important gradients are dropped, and only selected gradients are
updated. Our training paradigm introduces a tension between the rates of
pre-pruning and gradient-dropping, privacy loss, and classification accuracy.
Too much pre-pruning and gradient-dropping reduces the model's capacity and
worsens accuracy, while training a smaller model requires less privacy budget
for achieving good accuracy. We evaluate the interplay between these factors
and demonstrate the effectiveness of our training paradigm for both training
from scratch and fine-tuning pre-trained networks on several benchmark image
classification datasets. The tools can also be readily incorporated into
existing training paradigms.
| [
{
"version": "v1",
"created": "Mon, 19 Jun 2023 14:35:28 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Adamczewski",
"Kamil",
""
],
[
"He",
"Yingchen",
""
],
[
"Park",
"Mijung",
""
]
] | TITLE: Pre-Pruning and Gradient-Dropping Improve Differentially Private Image
Classification
ABSTRACT: Scalability is a significant challenge when it comes to applying differential
privacy to training deep neural networks. The commonly used DP-SGD algorithm
struggles to maintain a high level of privacy protection while achieving high
accuracy on even moderately sized models. To tackle this challenge, we take
advantage of the fact that neural networks are overparameterized, which allows
us to improve neural network training with differential privacy. Specifically,
we introduce a new training paradigm that uses \textit{pre-pruning} and
\textit{gradient-dropping} to reduce the parameter space and improve
scalability. The process starts with pre-pruning the parameters of the original
network to obtain a smaller model that is then trained with DP-SGD. During
training, less important gradients are dropped, and only selected gradients are
updated. Our training paradigm introduces a tension between the rates of
pre-pruning and gradient-dropping, privacy loss, and classification accuracy.
Too much pre-pruning and gradient-dropping reduces the model's capacity and
worsens accuracy, while training a smaller model requires less privacy budget
for achieving good accuracy. We evaluate the interplay between these factors
and demonstrate the effectiveness of our training paradigm for both training
from scratch and fine-tuning pre-trained networks on several benchmark image
classification datasets. The tools can also be readily incorporated into
existing training paradigms.
|
2306.11762 | Dongoo Lee Ph.D | Seunghan Park, Dongoo Lee, Yeonju Choi, SungTae Moon | MultiEarth 2023 Deforestation Challenge -- Team FOREVER | CVPR 2023, MultiEarth 2023, Deforestation Estimation Challenge | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | It is important problem to accurately estimate deforestation of satellite
imagery since this approach can analyse extensive area without direct human
access. However, it is not simple problem because of difficulty in observing
the clear ground surface due to extensive cloud cover during long rainy season.
In this paper, we present a multi-view learning strategy to predict
deforestation status in the Amazon rainforest area with latest deep neural
network models. Multi-modal dataset consists of three types of different
satellites imagery, Sentinel-1, Sentinel-2 and Landsat 8 is utilized to train
and predict deforestation status. MMsegmentation framework is selected to apply
comprehensive data augmentation and diverse networks. The proposed method
effectively and accurately predicts the deforestation status of new queries.
| [
{
"version": "v1",
"created": "Tue, 20 Jun 2023 09:10:06 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Park",
"Seunghan",
""
],
[
"Lee",
"Dongoo",
""
],
[
"Choi",
"Yeonju",
""
],
[
"Moon",
"SungTae",
""
]
] | TITLE: MultiEarth 2023 Deforestation Challenge -- Team FOREVER
ABSTRACT: It is important problem to accurately estimate deforestation of satellite
imagery since this approach can analyse extensive area without direct human
access. However, it is not simple problem because of difficulty in observing
the clear ground surface due to extensive cloud cover during long rainy season.
In this paper, we present a multi-view learning strategy to predict
deforestation status in the Amazon rainforest area with latest deep neural
network models. Multi-modal dataset consists of three types of different
satellites imagery, Sentinel-1, Sentinel-2 and Landsat 8 is utilized to train
and predict deforestation status. MMsegmentation framework is selected to apply
comprehensive data augmentation and diverse networks. The proposed method
effectively and accurately predicts the deforestation status of new queries.
|
2306.11763 | Alexander Van Meekeren | Alexander van Meekeren, Maya Aghaei, Klaas Dijkstra | Exploring the Effectiveness of Dataset Synthesis: An application of
Apple Detection in Orchards | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deep object detection models have achieved notable successes in recent years,
but one major obstacle remains: the requirement for a large amount of training
data. Obtaining such data is a tedious process and is mainly time consuming,
leading to the exploration of new research avenues like synthetic data
generation techniques. In this study, we explore the usability of Stable
Diffusion 2.1-base for generating synthetic datasets of apple trees for object
detection and compare it to a baseline model trained on real-world data. After
creating a dataset of realistic apple trees with prompt engineering and
utilizing a previously trained Stable Diffusion model, the custom dataset was
annotated and evaluated by training a YOLOv5m object detection model to predict
apples in a real-world apple detection dataset. YOLOv5m was chosen for its
rapid inference time and minimal hardware demands. Results demonstrate that the
model trained on generated data is slightly underperforming compared to a
baseline model trained on real-world images when evaluated on a set of
real-world images. However, these findings remain highly promising, as the
average precision difference is only 0.09 and 0.06, respectively. Qualitative
results indicate that the model can accurately predict the location of apples,
except in cases of heavy shading. These findings illustrate the potential of
synthetic data generation techniques as a viable alternative to the collection
of extensive training data for object detection models.
| [
{
"version": "v1",
"created": "Tue, 20 Jun 2023 09:46:01 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"van Meekeren",
"Alexander",
""
],
[
"Aghaei",
"Maya",
""
],
[
"Dijkstra",
"Klaas",
""
]
] | TITLE: Exploring the Effectiveness of Dataset Synthesis: An application of
Apple Detection in Orchards
ABSTRACT: Deep object detection models have achieved notable successes in recent years,
but one major obstacle remains: the requirement for a large amount of training
data. Obtaining such data is a tedious process and is mainly time consuming,
leading to the exploration of new research avenues like synthetic data
generation techniques. In this study, we explore the usability of Stable
Diffusion 2.1-base for generating synthetic datasets of apple trees for object
detection and compare it to a baseline model trained on real-world data. After
creating a dataset of realistic apple trees with prompt engineering and
utilizing a previously trained Stable Diffusion model, the custom dataset was
annotated and evaluated by training a YOLOv5m object detection model to predict
apples in a real-world apple detection dataset. YOLOv5m was chosen for its
rapid inference time and minimal hardware demands. Results demonstrate that the
model trained on generated data is slightly underperforming compared to a
baseline model trained on real-world images when evaluated on a set of
real-world images. However, these findings remain highly promising, as the
average precision difference is only 0.09 and 0.06, respectively. Qualitative
results indicate that the model can accurately predict the location of apples,
except in cases of heavy shading. These findings illustrate the potential of
synthetic data generation techniques as a viable alternative to the collection
of extensive training data for object detection models.
|
2306.11823 | Kamer Ali Yuksel | Kamer Ali Yuksel, Ahmet Gunduz, Mohamed Al-Badrashiny, Shreyas Sharma,
Hassan Sawaf | EvolveMT: an Ensemble MT Engine Improving Itself with Usage Only | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | This paper presents EvolveMT for efficiently combining multiple machine
translation (MT) engines. The proposed system selects the output from a single
engine for each segment by utilizing online learning techniques to predict the
most suitable system for every translation request. A neural quality estimation
metric supervises the method without requiring reference translations. The
online learning capability of this system allows for dynamic adaptation to
alterations in the domain or machine translation engines, thereby obviating the
necessity for additional training. EvolveMT selects a subset of translation
engines to be called based on the source sentence features. The degree of
exploration is configurable according to the desired quality-cost trade-off.
Results from custom datasets demonstrate that EvolveMT achieves similar
translation accuracy at a lower cost than selecting the best translation of
each segment from all translations using an MT quality estimator. To our
knowledge, EvolveMT is the first meta MT system that adapts itself after
deployment to incoming translation requests from the production environment
without needing costly retraining on human feedback.
| [
{
"version": "v1",
"created": "Tue, 20 Jun 2023 18:32:30 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Yuksel",
"Kamer Ali",
""
],
[
"Gunduz",
"Ahmet",
""
],
[
"Al-Badrashiny",
"Mohamed",
""
],
[
"Sharma",
"Shreyas",
""
],
[
"Sawaf",
"Hassan",
""
]
] | TITLE: EvolveMT: an Ensemble MT Engine Improving Itself with Usage Only
ABSTRACT: This paper presents EvolveMT for efficiently combining multiple machine
translation (MT) engines. The proposed system selects the output from a single
engine for each segment by utilizing online learning techniques to predict the
most suitable system for every translation request. A neural quality estimation
metric supervises the method without requiring reference translations. The
online learning capability of this system allows for dynamic adaptation to
alterations in the domain or machine translation engines, thereby obviating the
necessity for additional training. EvolveMT selects a subset of translation
engines to be called based on the source sentence features. The degree of
exploration is configurable according to the desired quality-cost trade-off.
Results from custom datasets demonstrate that EvolveMT achieves similar
translation accuracy at a lower cost than selecting the best translation of
each segment from all translations using an MT quality estimator. To our
knowledge, EvolveMT is the first meta MT system that adapts itself after
deployment to incoming translation requests from the production environment
without needing costly retraining on human feedback.
|
2306.11830 | Jan Sosulski | Jan Sosulski and Michael Tangermann | UMM: Unsupervised Mean-difference Maximization | null | null | null | null | cs.LG stat.AP | http://creativecommons.org/licenses/by/4.0/ | Many brain-computer interfaces make use of brain signals that are elicited in
response to a visual, auditory or tactile stimulus, so-called event-related
potentials (ERPs). In visual ERP speller applications, sets of letters shown on
a screen are flashed randomly, and the participant attends to the target letter
they want to spell. When this letter flashes, the resulting ERP is different
compared to when any other non-target letter flashes. We propose a new
unsupervised approach to detect this attended letter. In each trial, for every
available letter our approach makes the hypothesis that it is in fact the
attended letter, and calculates the ERPs based on each of these hypotheses. We
leverage the fact that only the true hypothesis produces the largest difference
between the class means. Note that this unsupervised method does not require
any changes to the underlying experimental paradigm and therefore can be
employed in almost any ERP-based setup. To deal with limited data, we use a
block-Toeplitz regularized covariance matrix that models the background
activity. We implemented the proposed novel unsupervised mean-difference
maximization (UMM) method and evaluated it in offline replays of brain-computer
interface visual speller datasets. For a dataset that used 16 flashes per
symbol per trial, UMM correctly classifies 3651 out of 3654 letters
($99.92\,\%$) across 25 participants. In another dataset with fewer and shorter
trials, 7344 out of 7383 letters ($99.47\,\%$) are classified correctly across
54 participants with two sessions each. Even in more challenging datasets
obtained from patients with amyotrophic lateral sclerosis ($77.86\,\%$) or when
using auditory ERPs ($82.52\,\%$), the obtained classification rates obtained
by UMM are competitive. In addition, UMM provides stable confidence measures
which can be used to monitor convergence.
| [
{
"version": "v1",
"created": "Tue, 20 Jun 2023 18:39:12 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Sosulski",
"Jan",
""
],
[
"Tangermann",
"Michael",
""
]
] | TITLE: UMM: Unsupervised Mean-difference Maximization
ABSTRACT: Many brain-computer interfaces make use of brain signals that are elicited in
response to a visual, auditory or tactile stimulus, so-called event-related
potentials (ERPs). In visual ERP speller applications, sets of letters shown on
a screen are flashed randomly, and the participant attends to the target letter
they want to spell. When this letter flashes, the resulting ERP is different
compared to when any other non-target letter flashes. We propose a new
unsupervised approach to detect this attended letter. In each trial, for every
available letter our approach makes the hypothesis that it is in fact the
attended letter, and calculates the ERPs based on each of these hypotheses. We
leverage the fact that only the true hypothesis produces the largest difference
between the class means. Note that this unsupervised method does not require
any changes to the underlying experimental paradigm and therefore can be
employed in almost any ERP-based setup. To deal with limited data, we use a
block-Toeplitz regularized covariance matrix that models the background
activity. We implemented the proposed novel unsupervised mean-difference
maximization (UMM) method and evaluated it in offline replays of brain-computer
interface visual speller datasets. For a dataset that used 16 flashes per
symbol per trial, UMM correctly classifies 3651 out of 3654 letters
($99.92\,\%$) across 25 participants. In another dataset with fewer and shorter
trials, 7344 out of 7383 letters ($99.47\,\%$) are classified correctly across
54 participants with two sessions each. Even in more challenging datasets
obtained from patients with amyotrophic lateral sclerosis ($77.86\,\%$) or when
using auditory ERPs ($82.52\,\%$), the obtained classification rates obtained
by UMM are competitive. In addition, UMM provides stable confidence measures
which can be used to monitor convergence.
|
2306.11832 | Juan Ramirez-Orta | Juan Ramirez-Orta and Eduardo Xamena and Ana Maguitman and Axel J.
Soto and Flavia P. Zanoto and Evangelos Milios | QuOTeS: Query-Oriented Technical Summarization | Accepted at ICDAR 2023 | null | null | null | cs.IR cs.CL | http://creativecommons.org/licenses/by/4.0/ | Abstract. When writing an academic paper, researchers often spend
considerable time reviewing and summarizing papers to extract relevant
citations and data to compose the Introduction and Related Work sections. To
address this problem, we propose QuOTeS, an interactive system designed to
retrieve sentences related to a summary of the research from a collection of
potential references and hence assist in the composition of new papers. QuOTeS
integrates techniques from Query-Focused Extractive Summarization and
High-Recall Information Retrieval to provide Interactive Query-Focused
Summarization of scientific documents. To measure the performance of our
system, we carried out a comprehensive user study where participants uploaded
papers related to their research and evaluated the system in terms of its
usability and the quality of the summaries it produces. The results show that
QuOTeS provides a positive user experience and consistently provides
query-focused summaries that are relevant, concise, and complete. We share the
code of our system and the novel Query-Focused Summarization dataset collected
during our experiments at https://github.com/jarobyte91/quotes.
| [
{
"version": "v1",
"created": "Tue, 20 Jun 2023 18:43:24 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Ramirez-Orta",
"Juan",
""
],
[
"Xamena",
"Eduardo",
""
],
[
"Maguitman",
"Ana",
""
],
[
"Soto",
"Axel J.",
""
],
[
"Zanoto",
"Flavia P.",
""
],
[
"Milios",
"Evangelos",
""
]
] | TITLE: QuOTeS: Query-Oriented Technical Summarization
ABSTRACT: Abstract. When writing an academic paper, researchers often spend
considerable time reviewing and summarizing papers to extract relevant
citations and data to compose the Introduction and Related Work sections. To
address this problem, we propose QuOTeS, an interactive system designed to
retrieve sentences related to a summary of the research from a collection of
potential references and hence assist in the composition of new papers. QuOTeS
integrates techniques from Query-Focused Extractive Summarization and
High-Recall Information Retrieval to provide Interactive Query-Focused
Summarization of scientific documents. To measure the performance of our
system, we carried out a comprehensive user study where participants uploaded
papers related to their research and evaluated the system in terms of its
usability and the quality of the summaries it produces. The results show that
QuOTeS provides a positive user experience and consistently provides
query-focused summaries that are relevant, concise, and complete. We share the
code of our system and the novel Query-Focused Summarization dataset collected
during our experiments at https://github.com/jarobyte91/quotes.
|
2306.11843 | Gaetano Rossiello | Michael Glass, Xueqing Wu, Ankita Rajaram Naik, Gaetano Rossiello,
Alfio Gliozzo | Retrieval-Based Transformer for Table Augmentation | Findings of ACL 2023 | null | null | null | cs.CL cs.AI cs.DB cs.IR | http://creativecommons.org/licenses/by/4.0/ | Data preparation, also called data wrangling, is considered one of the most
expensive and time-consuming steps when performing analytics or building
machine learning models. Preparing data typically involves collecting and
merging data from complex heterogeneous, and often large-scale data sources,
such as data lakes. In this paper, we introduce a novel approach toward
automatic data wrangling in an attempt to alleviate the effort of end-users,
e.g. data analysts, in structuring dynamic views from data lakes in the form of
tabular data. We aim to address table augmentation tasks, including row/column
population and data imputation. Given a corpus of tables, we propose a
retrieval augmented self-trained transformer model. Our self-learning strategy
consists in randomly ablating tables from the corpus and training the
retrieval-based model to reconstruct the original values or headers given the
partial tables as input. We adopt this strategy to first train the dense neural
retrieval model encoding table-parts to vectors, and then the end-to-end model
trained to perform table augmentation tasks. We test on EntiTables, the
standard benchmark for table augmentation, as well as introduce a new benchmark
to advance further research: WebTables. Our model consistently and
substantially outperforms both supervised statistical methods and the current
state-of-the-art transformer-based models.
| [
{
"version": "v1",
"created": "Tue, 20 Jun 2023 18:51:21 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Glass",
"Michael",
""
],
[
"Wu",
"Xueqing",
""
],
[
"Naik",
"Ankita Rajaram",
""
],
[
"Rossiello",
"Gaetano",
""
],
[
"Gliozzo",
"Alfio",
""
]
] | TITLE: Retrieval-Based Transformer for Table Augmentation
ABSTRACT: Data preparation, also called data wrangling, is considered one of the most
expensive and time-consuming steps when performing analytics or building
machine learning models. Preparing data typically involves collecting and
merging data from complex heterogeneous, and often large-scale data sources,
such as data lakes. In this paper, we introduce a novel approach toward
automatic data wrangling in an attempt to alleviate the effort of end-users,
e.g. data analysts, in structuring dynamic views from data lakes in the form of
tabular data. We aim to address table augmentation tasks, including row/column
population and data imputation. Given a corpus of tables, we propose a
retrieval augmented self-trained transformer model. Our self-learning strategy
consists in randomly ablating tables from the corpus and training the
retrieval-based model to reconstruct the original values or headers given the
partial tables as input. We adopt this strategy to first train the dense neural
retrieval model encoding table-parts to vectors, and then the end-to-end model
trained to perform table augmentation tasks. We test on EntiTables, the
standard benchmark for table augmentation, as well as introduce a new benchmark
to advance further research: WebTables. Our model consistently and
substantially outperforms both supervised statistical methods and the current
state-of-the-art transformer-based models.
|
2306.11848 | Jakub Caputa | Jakub Caputa, Maciej Wielgosz, Daria {\L}ukasik, Pawe{\l} Russek,
Jakub Grzeszczyk, Micha{\l} Karwatowski, Szymon Mazurek, Rafa{\l}
Fr\k{a}czek, Anna \'Smiech, Ernest Jamro, Sebastian Koryciak, Agnieszka
D\k{a}browska-Boruch, Marcin Pietro\'n, Kazimierz Wiatr | Using super-resolution for enhancing visual perception and segmentation
performance in veterinary cytology | null | null | null | null | cs.CV cs.LG eess.IV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The primary objective of this research was to enhance the quality of semantic
segmentation in cytology images by incorporating super-resolution (SR)
architectures. An additional contribution was the development of a novel
dataset aimed at improving imaging quality in the presence of inaccurate focus.
Our experimental results demonstrate that the integration of SR techniques into
the segmentation pipeline can lead to a significant improvement of up to 25% in
the mean average precision (mAP) segmentation metric. These findings suggest
that leveraging SR architectures holds great promise for advancing the state of
the art in cytology image analysis.
| [
{
"version": "v1",
"created": "Tue, 20 Jun 2023 18:59:27 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Caputa",
"Jakub",
""
],
[
"Wielgosz",
"Maciej",
""
],
[
"Łukasik",
"Daria",
""
],
[
"Russek",
"Paweł",
""
],
[
"Grzeszczyk",
"Jakub",
""
],
[
"Karwatowski",
"Michał",
""
],
[
"Mazurek",
"Szymon",
""
],
[
"Frączek",
"Rafał",
""
],
[
"Śmiech",
"Anna",
""
],
[
"Jamro",
"Ernest",
""
],
[
"Koryciak",
"Sebastian",
""
],
[
"Dąbrowska-Boruch",
"Agnieszka",
""
],
[
"Pietroń",
"Marcin",
""
],
[
"Wiatr",
"Kazimierz",
""
]
] | TITLE: Using super-resolution for enhancing visual perception and segmentation
performance in veterinary cytology
ABSTRACT: The primary objective of this research was to enhance the quality of semantic
segmentation in cytology images by incorporating super-resolution (SR)
architectures. An additional contribution was the development of a novel
dataset aimed at improving imaging quality in the presence of inaccurate focus.
Our experimental results demonstrate that the integration of SR techniques into
the segmentation pipeline can lead to a significant improvement of up to 25% in
the mean average precision (mAP) segmentation metric. These findings suggest
that leveraging SR architectures holds great promise for advancing the state of
the art in cytology image analysis.
|
2306.11853 | Katherine Sytwu | Katherine Sytwu, Luis Rangel DaCosta, Mary C. Scott | Generalization Across Experimental Parameters in Machine Learning
Analysis of High Resolution Transmission Electron Microscopy Datasets | 11 pages, 5 figures | null | null | null | cond-mat.mtrl-sci cs.LG eess.IV | http://creativecommons.org/licenses/by/4.0/ | Neural networks are promising tools for high-throughput and accurate
transmission electron microscopy (TEM) analysis of nanomaterials, but are known
to generalize poorly on data that is "out-of-distribution" from their training
data. Given the limited set of image features typically seen in high-resolution
TEM imaging, it is unclear which images are considered out-of-distribution from
others. Here, we investigate how the choice of metadata features in the
training dataset influences neural network performance, focusing on the example
task of nanoparticle segmentation. We train and validate neural networks across
curated, experimentally-collected high-resolution TEM image datasets of
nanoparticles under controlled imaging and material parameters, including
magnification, dosage, nanoparticle diameter, and nanoparticle material.
Overall, we find that our neural networks are not robust across microscope
parameters, but do generalize across certain sample parameters. Additionally,
data preprocessing heavily influences the generalizability of neural networks
trained on nominally similar datasets. Our results highlight the need to
understand how dataset features affect deployment of data-driven algorithms.
| [
{
"version": "v1",
"created": "Tue, 20 Jun 2023 19:13:49 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Sytwu",
"Katherine",
""
],
[
"DaCosta",
"Luis Rangel",
""
],
[
"Scott",
"Mary C.",
""
]
] | TITLE: Generalization Across Experimental Parameters in Machine Learning
Analysis of High Resolution Transmission Electron Microscopy Datasets
ABSTRACT: Neural networks are promising tools for high-throughput and accurate
transmission electron microscopy (TEM) analysis of nanomaterials, but are known
to generalize poorly on data that is "out-of-distribution" from their training
data. Given the limited set of image features typically seen in high-resolution
TEM imaging, it is unclear which images are considered out-of-distribution from
others. Here, we investigate how the choice of metadata features in the
training dataset influences neural network performance, focusing on the example
task of nanoparticle segmentation. We train and validate neural networks across
curated, experimentally-collected high-resolution TEM image datasets of
nanoparticles under controlled imaging and material parameters, including
magnification, dosage, nanoparticle diameter, and nanoparticle material.
Overall, we find that our neural networks are not robust across microscope
parameters, but do generalize across certain sample parameters. Additionally,
data preprocessing heavily influences the generalizability of neural networks
trained on nominally similar datasets. Our results highlight the need to
understand how dataset features affect deployment of data-driven algorithms.
|
2306.11855 | Mohammad Mehrabi | Mohammad Mehrabi and Ryan A. Rossi | A Model-free Closeness-of-influence Test for Features in Supervised
Learning | null | null | null | null | cs.LG stat.ME | http://creativecommons.org/licenses/by/4.0/ | Understanding the effect of a feature vector $x \in \mathbb{R}^d$ on the
response value (label) $y \in \mathbb{R}$ is the cornerstone of many
statistical learning problems. Ideally, it is desired to understand how a set
of collected features combine together and influence the response value, but
this problem is notoriously difficult, due to the high-dimensionality of data
and limited number of labeled data points, among many others. In this work, we
take a new perspective on this problem, and we study the question of assessing
the difference of influence that the two given features have on the response
value. We first propose a notion of closeness for the influence of features,
and show that our definition recovers the familiar notion of the magnitude of
coefficients in the parametric model. We then propose a novel method to test
for the closeness of influence in general model-free supervised learning
problems. Our proposed test can be used with finite number of samples with
control on type I error rate, no matter the ground truth conditional law
$\mathcal{L}(Y |X)$. We analyze the power of our test for two general learning
problems i) linear regression, and ii) binary classification under mixture of
Gaussian models, and show that under the proper choice of score function, an
internal component of our test, with sufficient number of samples will achieve
full statistical power. We evaluate our findings through extensive numerical
simulations, specifically we adopt the datamodel framework (Ilyas, et al.,
2022) for CIFAR-10 dataset to identify pairs of training samples with different
influence on the trained model via optional black box training mechanisms.
| [
{
"version": "v1",
"created": "Tue, 20 Jun 2023 19:20:18 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Mehrabi",
"Mohammad",
""
],
[
"Rossi",
"Ryan A.",
""
]
] | TITLE: A Model-free Closeness-of-influence Test for Features in Supervised
Learning
ABSTRACT: Understanding the effect of a feature vector $x \in \mathbb{R}^d$ on the
response value (label) $y \in \mathbb{R}$ is the cornerstone of many
statistical learning problems. Ideally, it is desired to understand how a set
of collected features combine together and influence the response value, but
this problem is notoriously difficult, due to the high-dimensionality of data
and limited number of labeled data points, among many others. In this work, we
take a new perspective on this problem, and we study the question of assessing
the difference of influence that the two given features have on the response
value. We first propose a notion of closeness for the influence of features,
and show that our definition recovers the familiar notion of the magnitude of
coefficients in the parametric model. We then propose a novel method to test
for the closeness of influence in general model-free supervised learning
problems. Our proposed test can be used with finite number of samples with
control on type I error rate, no matter the ground truth conditional law
$\mathcal{L}(Y |X)$. We analyze the power of our test for two general learning
problems i) linear regression, and ii) binary classification under mixture of
Gaussian models, and show that under the proper choice of score function, an
internal component of our test, with sufficient number of samples will achieve
full statistical power. We evaluate our findings through extensive numerical
simulations, specifically we adopt the datamodel framework (Ilyas, et al.,
2022) for CIFAR-10 dataset to identify pairs of training samples with different
influence on the trained model via optional black box training mechanisms.
|
2306.11867 | Jian Xu | Jian Xu, Xinyi Tong, Shao-Lun Huang | Personalized Federated Learning with Feature Alignment and Classifier
Collaboration | ICLR 2023, fix some typos and add the code link | null | null | null | cs.LG cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Data heterogeneity is one of the most challenging issues in federated
learning, which motivates a variety of approaches to learn personalized models
for participating clients. One such approach in deep neural networks based
tasks is employing a shared feature representation and learning a customized
classifier head for each client. However, previous works do not utilize the
global knowledge during local representation learning and also neglect the
fine-grained collaboration between local classifier heads, which limit the
model generalization ability. In this work, we conduct explicit local-global
feature alignment by leveraging global semantic knowledge for learning a better
representation. Moreover, we quantify the benefit of classifier combination for
each client as a function of the combining weights and derive an optimization
problem for estimating optimal weights. Finally, extensive evaluation results
on benchmark datasets with various heterogeneous data scenarios demonstrate the
effectiveness of our proposed method. Code is available at
https://github.com/JianXu95/FedPAC
| [
{
"version": "v1",
"created": "Tue, 20 Jun 2023 19:58:58 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Xu",
"Jian",
""
],
[
"Tong",
"Xinyi",
""
],
[
"Huang",
"Shao-Lun",
""
]
] | TITLE: Personalized Federated Learning with Feature Alignment and Classifier
Collaboration
ABSTRACT: Data heterogeneity is one of the most challenging issues in federated
learning, which motivates a variety of approaches to learn personalized models
for participating clients. One such approach in deep neural networks based
tasks is employing a shared feature representation and learning a customized
classifier head for each client. However, previous works do not utilize the
global knowledge during local representation learning and also neglect the
fine-grained collaboration between local classifier heads, which limit the
model generalization ability. In this work, we conduct explicit local-global
feature alignment by leveraging global semantic knowledge for learning a better
representation. Moreover, we quantify the benefit of classifier combination for
each client as a function of the combining weights and derive an optimization
problem for estimating optimal weights. Finally, extensive evaluation results
on benchmark datasets with various heterogeneous data scenarios demonstrate the
effectiveness of our proposed method. Code is available at
https://github.com/JianXu95/FedPAC
|
2306.11882 | Mat\'u\v{s} Sul\'ir | Mat\'u\v{s} Sul\'ir, Sergej Chodarev, Milan Nos\'a\v{l} | Outside the Sandbox: A Study of Input/Output Methods in Java | null | Proceedings of the 27th International Conference on Evaluation and
Assessment in Software Engineering (EASE '23), ACM, 2023, pp. 253-258 | 10.1145/3593434.3593501 | null | cs.SE cs.PL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Programming languages often demarcate the internal sandbox, consisting of
entities such as objects and variables, from the outside world, e.g., files or
network. Although communication with the external world poses fundamental
challenges for live programming, reversible debugging, testing, and program
analysis in general, studies about this phenomenon are rare. In this paper, we
present a preliminary empirical study about the prevalence of input/output
(I/O) method usage in Java. We manually categorized 1435 native methods in a
Java Standard Edition distribution into non-I/O and I/O-related methods, which
were further classified into areas such as desktop or file-related ones.
According to the static analysis of a call graph for 798 projects, about 57% of
methods potentially call I/O natives. The results of dynamic analysis on 16
benchmarks showed that 21% of the executed methods directly or indirectly
called an I/O native. We conclude that neglecting I/O is not a viable option
for tool designers and suggest the integration of I/O-related metadata with
source code to facilitate their querying.
| [
{
"version": "v1",
"created": "Tue, 20 Jun 2023 20:54:02 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Sulír",
"Matúš",
""
],
[
"Chodarev",
"Sergej",
""
],
[
"Nosáľ",
"Milan",
""
]
] | TITLE: Outside the Sandbox: A Study of Input/Output Methods in Java
ABSTRACT: Programming languages often demarcate the internal sandbox, consisting of
entities such as objects and variables, from the outside world, e.g., files or
network. Although communication with the external world poses fundamental
challenges for live programming, reversible debugging, testing, and program
analysis in general, studies about this phenomenon are rare. In this paper, we
present a preliminary empirical study about the prevalence of input/output
(I/O) method usage in Java. We manually categorized 1435 native methods in a
Java Standard Edition distribution into non-I/O and I/O-related methods, which
were further classified into areas such as desktop or file-related ones.
According to the static analysis of a call graph for 798 projects, about 57% of
methods potentially call I/O natives. The results of dynamic analysis on 16
benchmarks showed that 21% of the executed methods directly or indirectly
called an I/O native. We conclude that neglecting I/O is not a viable option
for tool designers and suggest the integration of I/O-related metadata with
source code to facilitate their querying.
|
2306.11890 | Wolfgang Pernice | Wolfgang M. Pernice, Michael Doron, Alex Quach, Aditya Pratapa, Sultan
Kenjeyev, Nicholas De Veaux, Michio Hirano, Juan C. Caicedo | Out of Distribution Generalization via Interventional Style Transfer in
Single-Cell Microscopy | Accepted at CVPR 2023 CVMI | null | null | null | cs.CV cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Real-world deployment of computer vision systems, including in the discovery
processes of biomedical research, requires causal representations that are
invariant to contextual nuisances and generalize to new data. Leveraging the
internal replicate structure of two novel single-cell fluorescent microscopy
datasets, we propose generally applicable tests to assess the extent to which
models learn causal representations across increasingly challenging levels of
OOD-generalization. We show that despite seemingly strong performance, as
assessed by other established metrics, both naive and contemporary baselines
designed to ward against confounding, collapse on these tests. We introduce a
new method, Interventional Style Transfer (IST), that substantially improves
OOD generalization by generating interventional training distributions in which
spurious correlations between biological causes and nuisances are mitigated. We
publish our code and datasets.
| [
{
"version": "v1",
"created": "Thu, 15 Jun 2023 20:08:16 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Pernice",
"Wolfgang M.",
""
],
[
"Doron",
"Michael",
""
],
[
"Quach",
"Alex",
""
],
[
"Pratapa",
"Aditya",
""
],
[
"Kenjeyev",
"Sultan",
""
],
[
"De Veaux",
"Nicholas",
""
],
[
"Hirano",
"Michio",
""
],
[
"Caicedo",
"Juan C.",
""
]
] | TITLE: Out of Distribution Generalization via Interventional Style Transfer in
Single-Cell Microscopy
ABSTRACT: Real-world deployment of computer vision systems, including in the discovery
processes of biomedical research, requires causal representations that are
invariant to contextual nuisances and generalize to new data. Leveraging the
internal replicate structure of two novel single-cell fluorescent microscopy
datasets, we propose generally applicable tests to assess the extent to which
models learn causal representations across increasingly challenging levels of
OOD-generalization. We show that despite seemingly strong performance, as
assessed by other established metrics, both naive and contemporary baselines
designed to ward against confounding, collapse on these tests. We introduce a
new method, Interventional Style Transfer (IST), that substantially improves
OOD generalization by generating interventional training distributions in which
spurious correlations between biological causes and nuisances are mitigated. We
publish our code and datasets.
|
2306.11899 | Linus Pithan | Linus Pithan (1), Vladimir Starostin (1), David Mare\v{c}ek (2), Lukas
Petersdorf (3), Constantin V\"olter (1), Valentin Munteanu (1), Maciej
Jankowski (4), Oleg Konovalov (4), Alexander Gerlach (1), Alexander
Hinderhofer (1), Bridget Murphy (3), Stefan Kowarik (2), Frank Schreiber (1)
((1) Universit\"at T\"ubingen Germany, (2) Universit\"at Graz Austria, (3)
Universit\"at Kiel Germany, (4) ESRF France) | Closing the loop: Autonomous experiments enabled by
machine-learning-based online data analysis in synchrotron beamline
environments | null | null | null | null | physics.data-an cond-mat.mtrl-sci cs.LG physics.app-ph | http://creativecommons.org/licenses/by/4.0/ | Recently, there has been significant interest in applying machine learning
(ML) techniques to X-ray scattering experiments, which proves to be a valuable
tool for enhancing research that involves large or rapidly generated datasets.
ML allows for the automated interpretation of experimental results,
particularly those obtained from synchrotron or neutron facilities. The speed
at which ML models can process data presents an important opportunity to
establish a closed-loop feedback system, enabling real-time decision-making
based on online data analysis. In this study, we describe the incorporation of
ML into a closed-loop workflow for X-ray reflectometry (XRR), using the growth
of organic thin films as an example. Our focus lies on the beamline integration
of ML-based online data analysis and closed-loop feedback. We present solutions
that provide an elementary data analysis in real time during the experiment
without introducing the additional software dependencies in the beamline
control software environment. Our data demonstrates the accuracy and robustness
of ML methods for analyzing XRR curves and Bragg reflections and its autonomous
control over a vacuum deposition setup.
| [
{
"version": "v1",
"created": "Tue, 20 Jun 2023 21:21:19 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Pithan",
"Linus",
""
],
[
"Starostin",
"Vladimir",
""
],
[
"Mareček",
"David",
""
],
[
"Petersdorf",
"Lukas",
""
],
[
"Völter",
"Constantin",
""
],
[
"Munteanu",
"Valentin",
""
],
[
"Jankowski",
"Maciej",
""
],
[
"Konovalov",
"Oleg",
""
],
[
"Gerlach",
"Alexander",
""
],
[
"Hinderhofer",
"Alexander",
""
],
[
"Murphy",
"Bridget",
""
],
[
"Kowarik",
"Stefan",
""
],
[
"Schreiber",
"Frank",
""
]
] | TITLE: Closing the loop: Autonomous experiments enabled by
machine-learning-based online data analysis in synchrotron beamline
environments
ABSTRACT: Recently, there has been significant interest in applying machine learning
(ML) techniques to X-ray scattering experiments, which proves to be a valuable
tool for enhancing research that involves large or rapidly generated datasets.
ML allows for the automated interpretation of experimental results,
particularly those obtained from synchrotron or neutron facilities. The speed
at which ML models can process data presents an important opportunity to
establish a closed-loop feedback system, enabling real-time decision-making
based on online data analysis. In this study, we describe the incorporation of
ML into a closed-loop workflow for X-ray reflectometry (XRR), using the growth
of organic thin films as an example. Our focus lies on the beamline integration
of ML-based online data analysis and closed-loop feedback. We present solutions
that provide an elementary data analysis in real time during the experiment
without introducing the additional software dependencies in the beamline
control software environment. Our data demonstrates the accuracy and robustness
of ML methods for analyzing XRR curves and Bragg reflections and its autonomous
control over a vacuum deposition setup.
|
2306.11912 | Michael Cooper | Ali Hossein Gharari Foomani, Michael Cooper, Russell Greiner, Rahul G.
Krishnan | Copula-Based Deep Survival Models for Dependent Censoring | 23 pages, 7 figures | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A survival dataset describes a set of instances (e.g. patients) and provides,
for each, either the time until an event (e.g. death), or the censoring time
(e.g. when lost to follow-up - which is a lower bound on the time until the
event). We consider the challenge of survival prediction: learning, from such
data, a predictive model that can produce an individual survival distribution
for a novel instance. Many contemporary methods of survival prediction
implicitly assume that the event and censoring distributions are independent
conditional on the instance's covariates - a strong assumption that is
difficult to verify (as we observe only one outcome for each instance) and
which can induce significant bias when it does not hold. This paper presents a
parametric model of survival that extends modern non-linear survival analysis
by relaxing the assumption of conditional independence. On synthetic and
semi-synthetic data, our approach significantly improves estimates of survival
distributions compared to the standard that assumes conditional independence in
the data.
| [
{
"version": "v1",
"created": "Tue, 20 Jun 2023 21:51:13 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Foomani",
"Ali Hossein Gharari",
""
],
[
"Cooper",
"Michael",
""
],
[
"Greiner",
"Russell",
""
],
[
"Krishnan",
"Rahul G.",
""
]
] | TITLE: Copula-Based Deep Survival Models for Dependent Censoring
ABSTRACT: A survival dataset describes a set of instances (e.g. patients) and provides,
for each, either the time until an event (e.g. death), or the censoring time
(e.g. when lost to follow-up - which is a lower bound on the time until the
event). We consider the challenge of survival prediction: learning, from such
data, a predictive model that can produce an individual survival distribution
for a novel instance. Many contemporary methods of survival prediction
implicitly assume that the event and censoring distributions are independent
conditional on the instance's covariates - a strong assumption that is
difficult to verify (as we observe only one outcome for each instance) and
which can induce significant bias when it does not hold. This paper presents a
parametric model of survival that extends modern non-linear survival analysis
by relaxing the assumption of conditional independence. On synthetic and
semi-synthetic data, our approach significantly improves estimates of survival
distributions compared to the standard that assumes conditional independence in
the data.
|
2306.11918 | Hang Wang | Hang Wang, Sen Lin, Junshan Zhang | Adaptive Ensemble Q-learning: Minimizing Estimation Bias via Error
Feedback | NeurIPS 2021 | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The ensemble method is a promising way to mitigate the overestimation issue
in Q-learning, where multiple function approximators are used to estimate the
action values. It is known that the estimation bias hinges heavily on the
ensemble size (i.e., the number of Q-function approximators used in the
target), and that determining the `right' ensemble size is highly nontrivial,
because of the time-varying nature of the function approximation errors during
the learning process. To tackle this challenge, we first derive an upper bound
and a lower bound on the estimation bias, based on which the ensemble size is
adapted to drive the bias to be nearly zero, thereby coping with the impact of
the time-varying approximation errors accordingly. Motivated by the theoretic
findings, we advocate that the ensemble method can be combined with Model
Identification Adaptive Control (MIAC) for effective ensemble size adaptation.
Specifically, we devise Adaptive Ensemble Q-learning (AdaEQ), a generalized
ensemble method with two key steps: (a) approximation error characterization
which serves as the feedback for flexibly controlling the ensemble size, and
(b) ensemble size adaptation tailored towards minimizing the estimation bias.
Extensive experiments are carried out to show that AdaEQ can improve the
learning performance than the existing methods for the MuJoCo benchmark.
| [
{
"version": "v1",
"created": "Tue, 20 Jun 2023 22:06:14 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Wang",
"Hang",
""
],
[
"Lin",
"Sen",
""
],
[
"Zhang",
"Junshan",
""
]
] | TITLE: Adaptive Ensemble Q-learning: Minimizing Estimation Bias via Error
Feedback
ABSTRACT: The ensemble method is a promising way to mitigate the overestimation issue
in Q-learning, where multiple function approximators are used to estimate the
action values. It is known that the estimation bias hinges heavily on the
ensemble size (i.e., the number of Q-function approximators used in the
target), and that determining the `right' ensemble size is highly nontrivial,
because of the time-varying nature of the function approximation errors during
the learning process. To tackle this challenge, we first derive an upper bound
and a lower bound on the estimation bias, based on which the ensemble size is
adapted to drive the bias to be nearly zero, thereby coping with the impact of
the time-varying approximation errors accordingly. Motivated by the theoretic
findings, we advocate that the ensemble method can be combined with Model
Identification Adaptive Control (MIAC) for effective ensemble size adaptation.
Specifically, we devise Adaptive Ensemble Q-learning (AdaEQ), a generalized
ensemble method with two key steps: (a) approximation error characterization
which serves as the feedback for flexibly controlling the ensemble size, and
(b) ensemble size adaptation tailored towards minimizing the estimation bias.
Extensive experiments are carried out to show that AdaEQ can improve the
learning performance than the existing methods for the MuJoCo benchmark.
|
2306.11922 | Charles Guille-Escuret | Charles Guille-Escuret, Hiroki Naganuma, Kilian Fatras, Ioannis
Mitliagkas | No Wrong Turns: The Simple Geometry Of Neural Networks Optimization
Paths | null | null | null | null | cs.LG math.OC | http://creativecommons.org/licenses/by/4.0/ | Understanding the optimization dynamics of neural networks is necessary for
closing the gap between theory and practice. Stochastic first-order
optimization algorithms are known to efficiently locate favorable minima in
deep neural networks. This efficiency, however, contrasts with the non-convex
and seemingly complex structure of neural loss landscapes. In this study, we
delve into the fundamental geometric properties of sampled gradients along
optimization paths. We focus on two key quantities, which appear in the
restricted secant inequality and error bound. Both hold high significance for
first-order optimization. Our analysis reveals that these quantities exhibit
predictable, consistent behavior throughout training, despite the stochasticity
induced by sampling minibatches. Our findings suggest that not only do
optimization trajectories never encounter significant obstacles, but they also
maintain stable dynamics during the majority of training. These observed
properties are sufficiently expressive to theoretically guarantee linear
convergence and prescribe learning rate schedules mirroring empirical
practices. We conduct our experiments on image classification, semantic
segmentation and language modeling across different batch sizes, network
architectures, datasets, optimizers, and initialization seeds. We discuss the
impact of each factor. Our work provides novel insights into the properties of
neural network loss functions, and opens the door to theoretical frameworks
more relevant to prevalent practice.
| [
{
"version": "v1",
"created": "Tue, 20 Jun 2023 22:10:40 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Guille-Escuret",
"Charles",
""
],
[
"Naganuma",
"Hiroki",
""
],
[
"Fatras",
"Kilian",
""
],
[
"Mitliagkas",
"Ioannis",
""
]
] | TITLE: No Wrong Turns: The Simple Geometry Of Neural Networks Optimization
Paths
ABSTRACT: Understanding the optimization dynamics of neural networks is necessary for
closing the gap between theory and practice. Stochastic first-order
optimization algorithms are known to efficiently locate favorable minima in
deep neural networks. This efficiency, however, contrasts with the non-convex
and seemingly complex structure of neural loss landscapes. In this study, we
delve into the fundamental geometric properties of sampled gradients along
optimization paths. We focus on two key quantities, which appear in the
restricted secant inequality and error bound. Both hold high significance for
first-order optimization. Our analysis reveals that these quantities exhibit
predictable, consistent behavior throughout training, despite the stochasticity
induced by sampling minibatches. Our findings suggest that not only do
optimization trajectories never encounter significant obstacles, but they also
maintain stable dynamics during the majority of training. These observed
properties are sufficiently expressive to theoretically guarantee linear
convergence and prescribe learning rate schedules mirroring empirical
practices. We conduct our experiments on image classification, semantic
segmentation and language modeling across different batch sizes, network
architectures, datasets, optimizers, and initialization seeds. We discuss the
impact of each factor. Our work provides novel insights into the properties of
neural network loss functions, and opens the door to theoretical frameworks
more relevant to prevalent practice.
|
2306.11955 | Gusseppe Bravo-Rocca | Gusseppe Bravo-Rocca, Peini Liu, Jordi Guitart, Ajay Dholakia, David
Ellison | TADIL: Task-Agnostic Domain-Incremental Learning through Task-ID
Inference using Transformer Nearest-Centroid Embeddings | An early version of this work was presented at CVPR 2023, LXAI
Workshop | null | null | null | cs.LG cs.CV | http://creativecommons.org/licenses/by/4.0/ | Machine Learning (ML) models struggle with data that changes over time or
across domains due to factors such as noise, occlusion, illumination, or
frequency, unlike humans who can learn from such non independent and
identically distributed data. Consequently, a Continual Learning (CL) approach
is indispensable, particularly, Domain-Incremental Learning. In this paper, we
propose a novel pipeline for identifying tasks in domain-incremental learning
scenarios without supervision. The pipeline comprises four steps. First, we
obtain base embeddings from the raw data using an existing transformer-based
model. Second, we group the embedding densities based on their similarity to
obtain the nearest points to each cluster centroid. Third, we train an
incremental task classifier using only these few points. Finally, we leverage
the lightweight computational requirements of the pipeline to devise an
algorithm that decides in an online fashion when to learn a new task using the
task classifier and a drift detector. We conduct experiments using the SODA10M
real-world driving dataset and several CL strategies. We demonstrate that the
performance of these CL strategies with our pipeline can match the ground-truth
approach, both in classical experiments assuming task boundaries, and also in
more realistic task-agnostic scenarios that require detecting new tasks
on-the-fly
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 00:55:02 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Bravo-Rocca",
"Gusseppe",
""
],
[
"Liu",
"Peini",
""
],
[
"Guitart",
"Jordi",
""
],
[
"Dholakia",
"Ajay",
""
],
[
"Ellison",
"David",
""
]
] | TITLE: TADIL: Task-Agnostic Domain-Incremental Learning through Task-ID
Inference using Transformer Nearest-Centroid Embeddings
ABSTRACT: Machine Learning (ML) models struggle with data that changes over time or
across domains due to factors such as noise, occlusion, illumination, or
frequency, unlike humans who can learn from such non independent and
identically distributed data. Consequently, a Continual Learning (CL) approach
is indispensable, particularly, Domain-Incremental Learning. In this paper, we
propose a novel pipeline for identifying tasks in domain-incremental learning
scenarios without supervision. The pipeline comprises four steps. First, we
obtain base embeddings from the raw data using an existing transformer-based
model. Second, we group the embedding densities based on their similarity to
obtain the nearest points to each cluster centroid. Third, we train an
incremental task classifier using only these few points. Finally, we leverage
the lightweight computational requirements of the pipeline to devise an
algorithm that decides in an online fashion when to learn a new task using the
task classifier and a drift detector. We conduct experiments using the SODA10M
real-world driving dataset and several CL strategies. We demonstrate that the
performance of these CL strategies with our pipeline can match the ground-truth
approach, both in classical experiments assuming task boundaries, and also in
more realistic task-agnostic scenarios that require detecting new tasks
on-the-fly
|
2306.11970 | Xiangjun Tang | Xiangjun Tang, Linjun Wu, He Wang, Bo Hu, Xu Gong, Yuchen Liao,
Songnan Li, Qilong Kou, Xiaogang Jin | RSMT: Real-time Stylized Motion Transition for Characters | null | SIGGRAPH 2023 Conference Proceedings | 10.1145/3588432.3591514 | null | cs.CV cs.GR cs.GT | http://creativecommons.org/licenses/by/4.0/ | Styled online in-between motion generation has important application
scenarios in computer animation and games. Its core challenge lies in the need
to satisfy four critical requirements simultaneously: generation speed, motion
quality, style diversity, and synthesis controllability. While the first two
challenges demand a delicate balance between simple fast models and learning
capacity for generation quality, the latter two are rarely investigated
together in existing methods, which largely focus on either control without
style or uncontrolled stylized motions. To this end, we propose a Real-time
Stylized Motion Transition method (RSMT) to achieve all aforementioned goals.
Our method consists of two critical, independent components: a general motion
manifold model and a style motion sampler. The former acts as a high-quality
motion source and the latter synthesizes styled motions on the fly under
control signals. Since both components can be trained separately on different
datasets, our method provides great flexibility, requires less data, and
generalizes well when no/few samples are available for unseen styles. Through
exhaustive evaluation, our method proves to be fast, high-quality, versatile,
and controllable. The code and data are available at
{https://github.com/yuyujunjun/RSMT-Realtime-Stylized-Motion-Transition.}
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 01:50:04 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Tang",
"Xiangjun",
""
],
[
"Wu",
"Linjun",
""
],
[
"Wang",
"He",
""
],
[
"Hu",
"Bo",
""
],
[
"Gong",
"Xu",
""
],
[
"Liao",
"Yuchen",
""
],
[
"Li",
"Songnan",
""
],
[
"Kou",
"Qilong",
""
],
[
"Jin",
"Xiaogang",
""
]
] | TITLE: RSMT: Real-time Stylized Motion Transition for Characters
ABSTRACT: Styled online in-between motion generation has important application
scenarios in computer animation and games. Its core challenge lies in the need
to satisfy four critical requirements simultaneously: generation speed, motion
quality, style diversity, and synthesis controllability. While the first two
challenges demand a delicate balance between simple fast models and learning
capacity for generation quality, the latter two are rarely investigated
together in existing methods, which largely focus on either control without
style or uncontrolled stylized motions. To this end, we propose a Real-time
Stylized Motion Transition method (RSMT) to achieve all aforementioned goals.
Our method consists of two critical, independent components: a general motion
manifold model and a style motion sampler. The former acts as a high-quality
motion source and the latter synthesizes styled motions on the fly under
control signals. Since both components can be trained separately on different
datasets, our method provides great flexibility, requires less data, and
generalizes well when no/few samples are available for unseen styles. Through
exhaustive evaluation, our method proves to be fast, high-quality, versatile,
and controllable. The code and data are available at
{https://github.com/yuyujunjun/RSMT-Realtime-Stylized-Motion-Transition.}
|
2306.11984 | Se-In Jang | Se-In Jang, Cristina Lois, Emma Thibault, J. Alex Becker, Yafei Dong,
Marc D. Normandin, Julie C. Price, Keith A. Johnson, Georges El Fakhri, Kuang
Gong | TauPETGen: Text-Conditional Tau PET Image Synthesis Based on Latent
Diffusion Models | null | null | null | null | eess.IV cs.AI cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In this work, we developed a novel text-guided image synthesis technique
which could generate realistic tau PET images from textual descriptions and the
subject's MR image. The generated tau PET images have the potential to be used
in examining relations between different measures and also increasing the
public availability of tau PET datasets. The method was based on latent
diffusion models. Both textual descriptions and the subject's MR prior image
were utilized as conditions during image generation. The subject's MR image can
provide anatomical details, while the text descriptions, such as gender, scan
time, cognitive test scores, and amyloid status, can provide further guidance
regarding where the tau neurofibrillary tangles might be deposited. Preliminary
experimental results based on clinical [18F]MK-6240 datasets demonstrate the
feasibility of the proposed method in generating realistic tau PET images at
different clinical stages.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 02:27:07 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Jang",
"Se-In",
""
],
[
"Lois",
"Cristina",
""
],
[
"Thibault",
"Emma",
""
],
[
"Becker",
"J. Alex",
""
],
[
"Dong",
"Yafei",
""
],
[
"Normandin",
"Marc D.",
""
],
[
"Price",
"Julie C.",
""
],
[
"Johnson",
"Keith A.",
""
],
[
"Fakhri",
"Georges El",
""
],
[
"Gong",
"Kuang",
""
]
] | TITLE: TauPETGen: Text-Conditional Tau PET Image Synthesis Based on Latent
Diffusion Models
ABSTRACT: In this work, we developed a novel text-guided image synthesis technique
which could generate realistic tau PET images from textual descriptions and the
subject's MR image. The generated tau PET images have the potential to be used
in examining relations between different measures and also increasing the
public availability of tau PET datasets. The method was based on latent
diffusion models. Both textual descriptions and the subject's MR prior image
were utilized as conditions during image generation. The subject's MR image can
provide anatomical details, while the text descriptions, such as gender, scan
time, cognitive test scores, and amyloid status, can provide further guidance
regarding where the tau neurofibrillary tangles might be deposited. Preliminary
experimental results based on clinical [18F]MK-6240 datasets demonstrate the
feasibility of the proposed method in generating realistic tau PET images at
different clinical stages.
|
2306.11985 | Aida Brankovic Dr | Aida Brankovic, David Cook, Jessica Rahman, Wenjie Huang, Sankalp
Khanna | Evaluation of Popular XAI Applied to Clinical Prediction Models: Can
They be Trusted? | null | null | null | null | cs.LG cs.CY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The absence of transparency and explainability hinders the clinical adoption
of Machine learning (ML) algorithms. Although various methods of explainable
artificial intelligence (XAI) have been suggested, there is a lack of
literature that delves into their practicality and assesses them based on
criteria that could foster trust in clinical environments. To address this gap
this study evaluates two popular XAI methods used for explaining predictive
models in the healthcare context in terms of whether they (i) generate
domain-appropriate representation, i.e. coherent with respect to the
application task, (ii) impact clinical workflow and (iii) are consistent. To
that end, explanations generated at the cohort and patient levels were
analysed. The paper reports the first benchmarking of the XAI methods applied
to risk prediction models obtained by evaluating the concordance between
generated explanations and the trigger of a future clinical deterioration
episode recorded by the data collection system. We carried out an analysis
using two Electronic Medical Records (EMR) datasets sourced from Australian
major hospitals. The findings underscore the limitations of state-of-the-art
XAI methods in the clinical context and their potential benefits. We discuss
these limitations and contribute to the theoretical development of trustworthy
XAI solutions where clinical decision support guides the choice of intervention
by suggesting the pattern or drivers for clinical deterioration in the future.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 02:29:30 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Brankovic",
"Aida",
""
],
[
"Cook",
"David",
""
],
[
"Rahman",
"Jessica",
""
],
[
"Huang",
"Wenjie",
""
],
[
"Khanna",
"Sankalp",
""
]
] | TITLE: Evaluation of Popular XAI Applied to Clinical Prediction Models: Can
They be Trusted?
ABSTRACT: The absence of transparency and explainability hinders the clinical adoption
of Machine learning (ML) algorithms. Although various methods of explainable
artificial intelligence (XAI) have been suggested, there is a lack of
literature that delves into their practicality and assesses them based on
criteria that could foster trust in clinical environments. To address this gap
this study evaluates two popular XAI methods used for explaining predictive
models in the healthcare context in terms of whether they (i) generate
domain-appropriate representation, i.e. coherent with respect to the
application task, (ii) impact clinical workflow and (iii) are consistent. To
that end, explanations generated at the cohort and patient levels were
analysed. The paper reports the first benchmarking of the XAI methods applied
to risk prediction models obtained by evaluating the concordance between
generated explanations and the trigger of a future clinical deterioration
episode recorded by the data collection system. We carried out an analysis
using two Electronic Medical Records (EMR) datasets sourced from Australian
major hospitals. The findings underscore the limitations of state-of-the-art
XAI methods in the clinical context and their potential benefits. We discuss
these limitations and contribute to the theoretical development of trustworthy
XAI solutions where clinical decision support guides the choice of intervention
by suggesting the pattern or drivers for clinical deterioration in the future.
|
2306.12015 | Milind Rao | Milind Rao, Gopinath Chennupati, Gautam Tiwari, Anit Kumar Sahu,
Anirudh Raju, Ariya Rastrow, Jasha Droppo | Federated Self-Learning with Weak Supervision for Speech Recognition | Proceedings of ICASSP 2023 | null | null | null | eess.AS cs.SD | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Automatic speech recognition (ASR) models with low-footprint are increasingly
being deployed on edge devices for conversational agents, which enhances
privacy. We study the problem of federated continual incremental learning for
recurrent neural network-transducer (RNN-T) ASR models in the privacy-enhancing
scheme of learning on-device, without access to ground truth human transcripts
or machine transcriptions from a stronger ASR model. In particular, we study
the performance of a self-learning based scheme, with a paired teacher model
updated through an exponential moving average of ASR models. Further, we
propose using possibly noisy weak-supervision signals such as feedback scores
and natural language understanding semantics determined from user behavior
across multiple turns in a session of interactions with the conversational
agent. These signals are leveraged in a multi-task policy-gradient training
approach to improve the performance of self-learning for ASR. Finally, we show
how catastrophic forgetting can be mitigated by combining on-device learning
with a memory-replay approach using selected historical datasets. These
innovations allow for 10% relative improvement in WER on new use cases with
minimal degradation on other test sets in the absence of strong-supervision
signals such as ground-truth transcriptions.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 04:41:42 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Rao",
"Milind",
""
],
[
"Chennupati",
"Gopinath",
""
],
[
"Tiwari",
"Gautam",
""
],
[
"Sahu",
"Anit Kumar",
""
],
[
"Raju",
"Anirudh",
""
],
[
"Rastrow",
"Ariya",
""
],
[
"Droppo",
"Jasha",
""
]
] | TITLE: Federated Self-Learning with Weak Supervision for Speech Recognition
ABSTRACT: Automatic speech recognition (ASR) models with low-footprint are increasingly
being deployed on edge devices for conversational agents, which enhances
privacy. We study the problem of federated continual incremental learning for
recurrent neural network-transducer (RNN-T) ASR models in the privacy-enhancing
scheme of learning on-device, without access to ground truth human transcripts
or machine transcriptions from a stronger ASR model. In particular, we study
the performance of a self-learning based scheme, with a paired teacher model
updated through an exponential moving average of ASR models. Further, we
propose using possibly noisy weak-supervision signals such as feedback scores
and natural language understanding semantics determined from user behavior
across multiple turns in a session of interactions with the conversational
agent. These signals are leveraged in a multi-task policy-gradient training
approach to improve the performance of self-learning for ASR. Finally, we show
how catastrophic forgetting can be mitigated by combining on-device learning
with a memory-replay approach using selected historical datasets. These
innovations allow for 10% relative improvement in WER on new use cases with
minimal degradation on other test sets in the absence of strong-supervision
signals such as ground-truth transcriptions.
|
2306.12020 | Mohan Zhou | Mohan Zhou, Yalong Bai, Wei Zhang, Ting Yao, Tiejun Zhao, Tao Mei | Visual-Aware Text-to-Speech | accepted as oral and top 3% paper by ICASSP 2023 | ICASSP 2023 - 2023 IEEE International Conference on Acoustics,
Speech and Signal Processing (ICASSP) 2023, 1-5 | 10.1109/ICASSP49357.2023.10095084 | null | eess.AS cs.CL cs.SD | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Dynamically synthesizing talking speech that actively responds to a listening
head is critical during the face-to-face interaction. For example, the speaker
could take advantage of the listener's facial expression to adjust the tones,
stressed syllables, or pauses. In this work, we present a new visual-aware
text-to-speech (VA-TTS) task to synthesize speech conditioned on both textual
inputs and sequential visual feedback (e.g., nod, smile) of the listener in
face-to-face communication. Different from traditional text-to-speech, VA-TTS
highlights the impact of visual modality. On this newly-minted task, we devise
a baseline model to fuse phoneme linguistic information and listener visual
signals for speech synthesis. Extensive experiments on multimodal conversation
dataset ViCo-X verify our proposal for generating more natural audio with
scenario-appropriate rhythm and prosody.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 05:11:39 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Zhou",
"Mohan",
""
],
[
"Bai",
"Yalong",
""
],
[
"Zhang",
"Wei",
""
],
[
"Yao",
"Ting",
""
],
[
"Zhao",
"Tiejun",
""
],
[
"Mei",
"Tao",
""
]
] | TITLE: Visual-Aware Text-to-Speech
ABSTRACT: Dynamically synthesizing talking speech that actively responds to a listening
head is critical during the face-to-face interaction. For example, the speaker
could take advantage of the listener's facial expression to adjust the tones,
stressed syllables, or pauses. In this work, we present a new visual-aware
text-to-speech (VA-TTS) task to synthesize speech conditioned on both textual
inputs and sequential visual feedback (e.g., nod, smile) of the listener in
face-to-face communication. Different from traditional text-to-speech, VA-TTS
highlights the impact of visual modality. On this newly-minted task, we devise
a baseline model to fuse phoneme linguistic information and listener visual
signals for speech synthesis. Extensive experiments on multimodal conversation
dataset ViCo-X verify our proposal for generating more natural audio with
scenario-appropriate rhythm and prosody.
|
2306.12050 | Daichi Haraguchi | Naoya Yasukochi, Hideaki Hayashi, Daichi Haraguchi, Seiichi Uchida | Analyzing Font Style Usage and Contextual Factors in Real Images | Accepted at ICDAR 2023 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | There are various font styles in the world. Different styles give different
impressions and readability. This paper analyzes the relationship between font
styles and contextual factors that might affect font style selection with
large-scale datasets. For example, we will analyze the relationship between
font style and its surrounding object (such as ``bus'') by using about 800,000
words in the Open Images dataset. We also use a book cover dataset to analyze
the relationship between font styles with book genres. Moreover, the meaning of
the word is assumed as another contextual factor. For these numeric analyses,
we utilize our own font-style feature extraction model and word2vec. As a
result of co-occurrence-based relationship analysis, we found several instances
of specific font styles being used for specific contextual factors.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 06:43:22 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Yasukochi",
"Naoya",
""
],
[
"Hayashi",
"Hideaki",
""
],
[
"Haraguchi",
"Daichi",
""
],
[
"Uchida",
"Seiichi",
""
]
] | TITLE: Analyzing Font Style Usage and Contextual Factors in Real Images
ABSTRACT: There are various font styles in the world. Different styles give different
impressions and readability. This paper analyzes the relationship between font
styles and contextual factors that might affect font style selection with
large-scale datasets. For example, we will analyze the relationship between
font style and its surrounding object (such as ``bus'') by using about 800,000
words in the Open Images dataset. We also use a book cover dataset to analyze
the relationship between font styles with book genres. Moreover, the meaning of
the word is assumed as another contextual factor. For these numeric analyses,
we utilize our own font-style feature extraction model and word2vec. As a
result of co-occurrence-based relationship analysis, we found several instances
of specific font styles being used for specific contextual factors.
|
2306.12069 | Jialin Chen | Jialin Chen, Zhuosheng Zhang, Hai Zhao | Modeling Hierarchical Reasoning Chains by Linking Discourse Units and
Key Phrases for Reading Comprehension | Accepted at COLING 2022, 9 pages | null | null | null | cs.CL cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Machine reading comprehension (MRC) poses new challenges over logical
reasoning, which aims to understand the implicit logical relations entailed in
the given contexts and perform inference over them. Due to the complexity of
logic, logical relations exist at different granularity levels. However, most
existing methods of logical reasoning individually focus on either entity-aware
or discourse-based information but ignore the hierarchical relations that may
even have mutual effects. In this paper, we propose a holistic graph network
(HGN) which deals with context at both discourse level and word level, as the
basis for logical reasoning, to provide a more fine-grained relation
extraction. Specifically, node-level and type-level relations, which can be
interpreted as bridges in the reasoning process, are modeled by a hierarchical
interaction mechanism to improve the interpretation of MRC systems.
Experimental results on logical reasoning QA datasets (ReClor and LogiQA) and
natural language inference datasets (SNLI and ANLI) show the effectiveness and
generalization of our method, and in-depth analysis verifies its capability to
understand complex logical relations.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 07:34:27 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Chen",
"Jialin",
""
],
[
"Zhang",
"Zhuosheng",
""
],
[
"Zhao",
"Hai",
""
]
] | TITLE: Modeling Hierarchical Reasoning Chains by Linking Discourse Units and
Key Phrases for Reading Comprehension
ABSTRACT: Machine reading comprehension (MRC) poses new challenges over logical
reasoning, which aims to understand the implicit logical relations entailed in
the given contexts and perform inference over them. Due to the complexity of
logic, logical relations exist at different granularity levels. However, most
existing methods of logical reasoning individually focus on either entity-aware
or discourse-based information but ignore the hierarchical relations that may
even have mutual effects. In this paper, we propose a holistic graph network
(HGN) which deals with context at both discourse level and word level, as the
basis for logical reasoning, to provide a more fine-grained relation
extraction. Specifically, node-level and type-level relations, which can be
interpreted as bridges in the reasoning process, are modeled by a hierarchical
interaction mechanism to improve the interpretation of MRC systems.
Experimental results on logical reasoning QA datasets (ReClor and LogiQA) and
natural language inference datasets (SNLI and ANLI) show the effectiveness and
generalization of our method, and in-depth analysis verifies its capability to
understand complex logical relations.
|
2306.12077 | Kai Lagemann | Kai Lagemann, Christian Lagemann, Sach Mukherjee | Learning Latent Dynamics via Invariant Decomposition and
(Spatio-)Temporal Transformers | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | We propose a method for learning dynamical systems from high-dimensional
empirical data that combines variational autoencoders and (spatio-)temporal
attention within a framework designed to enforce certain
scientifically-motivated invariances. We focus on the setting in which data are
available from multiple different instances of a system whose underlying
dynamical model is entirely unknown at the outset. The approach rests on a
separation into an instance-specific encoding (capturing initial conditions,
constants etc.) and a latent dynamics model that is itself universal across all
instances/realizations of the system. The separation is achieved in an
automated, data-driven manner and only empirical data are required as inputs to
the model. The approach allows effective inference of system behaviour at any
continuous time but does not require an explicit neural ODE formulation, which
makes it efficient and highly scalable. We study behaviour through simple
theoretical analyses and extensive experiments on synthetic and real-world
datasets. The latter investigate learning the dynamics of complex systems based
on finite data and show that the proposed approach can outperform
state-of-the-art neural-dynamical models. We study also more general inductive
bias in the context of transfer to data obtained under entirely novel system
interventions. Overall, our results provide a promising new framework for
efficiently learning dynamical models from heterogeneous data with potential
applications in a wide range of fields including physics, medicine, biology and
engineering.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 07:52:07 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Lagemann",
"Kai",
""
],
[
"Lagemann",
"Christian",
""
],
[
"Mukherjee",
"Sach",
""
]
] | TITLE: Learning Latent Dynamics via Invariant Decomposition and
(Spatio-)Temporal Transformers
ABSTRACT: We propose a method for learning dynamical systems from high-dimensional
empirical data that combines variational autoencoders and (spatio-)temporal
attention within a framework designed to enforce certain
scientifically-motivated invariances. We focus on the setting in which data are
available from multiple different instances of a system whose underlying
dynamical model is entirely unknown at the outset. The approach rests on a
separation into an instance-specific encoding (capturing initial conditions,
constants etc.) and a latent dynamics model that is itself universal across all
instances/realizations of the system. The separation is achieved in an
automated, data-driven manner and only empirical data are required as inputs to
the model. The approach allows effective inference of system behaviour at any
continuous time but does not require an explicit neural ODE formulation, which
makes it efficient and highly scalable. We study behaviour through simple
theoretical analyses and extensive experiments on synthetic and real-world
datasets. The latter investigate learning the dynamics of complex systems based
on finite data and show that the proposed approach can outperform
state-of-the-art neural-dynamical models. We study also more general inductive
bias in the context of transfer to data obtained under entirely novel system
interventions. Overall, our results provide a promising new framework for
efficiently learning dynamical models from heterogeneous data with potential
applications in a wide range of fields including physics, medicine, biology and
engineering.
|
2306.12079 | Zheng Wang | Zheng Wang, Xiaoliang Fan, Zhaopeng Peng, Xueheng Li, Ziqi Yang,
Mingkuan Feng, Zhicheng Yang, Xiao Liu, and Cheng Wang | FLGo: A Fully Customizable Federated Learning Platform | null | null | null | null | cs.LG cs.DC | http://creativecommons.org/licenses/by/4.0/ | Federated learning (FL) has found numerous applications in healthcare,
finance, and IoT scenarios. Many existing FL frameworks offer a range of
benchmarks to evaluate the performance of FL under realistic conditions.
However, the process of customizing simulations to accommodate
application-specific settings, data heterogeneity, and system heterogeneity
typically remains unnecessarily complicated. This creates significant hurdles
for traditional ML researchers in exploring the usage of FL, while also
compromising the shareability of codes across FL frameworks. To address this
issue, we propose a novel lightweight FL platform called FLGo, to facilitate
cross-application FL studies with a high degree of shareability. Our platform
offers 40+ benchmarks, 20+ algorithms, and 2 system simulators as
out-of-the-box plugins. We also provide user-friendly APIs for quickly
customizing new plugins that can be readily shared and reused for improved
reproducibility. Finally, we develop a range of experimental tools, including
parallel acceleration, experiment tracker and analyzer, and parameters
auto-tuning. FLGo is maintained at \url{flgo-xmu.github.io}.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 07:55:29 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Wang",
"Zheng",
""
],
[
"Fan",
"Xiaoliang",
""
],
[
"Peng",
"Zhaopeng",
""
],
[
"Li",
"Xueheng",
""
],
[
"Yang",
"Ziqi",
""
],
[
"Feng",
"Mingkuan",
""
],
[
"Yang",
"Zhicheng",
""
],
[
"Liu",
"Xiao",
""
],
[
"Wang",
"Cheng",
""
]
] | TITLE: FLGo: A Fully Customizable Federated Learning Platform
ABSTRACT: Federated learning (FL) has found numerous applications in healthcare,
finance, and IoT scenarios. Many existing FL frameworks offer a range of
benchmarks to evaluate the performance of FL under realistic conditions.
However, the process of customizing simulations to accommodate
application-specific settings, data heterogeneity, and system heterogeneity
typically remains unnecessarily complicated. This creates significant hurdles
for traditional ML researchers in exploring the usage of FL, while also
compromising the shareability of codes across FL frameworks. To address this
issue, we propose a novel lightweight FL platform called FLGo, to facilitate
cross-application FL studies with a high degree of shareability. Our platform
offers 40+ benchmarks, 20+ algorithms, and 2 system simulators as
out-of-the-box plugins. We also provide user-friendly APIs for quickly
customizing new plugins that can be readily shared and reused for improved
reproducibility. Finally, we develop a range of experimental tools, including
parallel acceleration, experiment tracker and analyzer, and parameters
auto-tuning. FLGo is maintained at \url{flgo-xmu.github.io}.
|
2306.12085 | Hanyu Mao | Chanyue Wu, Dong Wang, Hanyu Mao, Ying Li | HSR-Diff:Hyperspectral Image Super-Resolution via Conditional Diffusion
Models | null | null | null | null | cs.CV eess.IV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Despite the proven significance of hyperspectral images (HSIs) in performing
various computer vision tasks, its potential is adversely affected by the
low-resolution (LR) property in the spatial domain, resulting from multiple
physical factors. Inspired by recent advancements in deep generative models, we
propose an HSI Super-resolution (SR) approach with Conditional Diffusion Models
(HSR-Diff) that merges a high-resolution (HR) multispectral image (MSI) with
the corresponding LR-HSI. HSR-Diff generates an HR-HSI via repeated refinement,
in which the HR-HSI is initialized with pure Gaussian noise and iteratively
refined. At each iteration, the noise is removed with a Conditional Denoising
Transformer (CDF ormer) that is trained on denoising at different noise levels,
conditioned on the hierarchical feature maps of HR-MSI and LR-HSI. In addition,
a progressive learning strategy is employed to exploit the global information
of full-resolution images. Systematic experiments have been conducted on four
public datasets, demonstrating that HSR-Diff outperforms state-of-the-art
methods.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 08:04:30 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Wu",
"Chanyue",
""
],
[
"Wang",
"Dong",
""
],
[
"Mao",
"Hanyu",
""
],
[
"Li",
"Ying",
""
]
] | TITLE: HSR-Diff:Hyperspectral Image Super-Resolution via Conditional Diffusion
Models
ABSTRACT: Despite the proven significance of hyperspectral images (HSIs) in performing
various computer vision tasks, its potential is adversely affected by the
low-resolution (LR) property in the spatial domain, resulting from multiple
physical factors. Inspired by recent advancements in deep generative models, we
propose an HSI Super-resolution (SR) approach with Conditional Diffusion Models
(HSR-Diff) that merges a high-resolution (HR) multispectral image (MSI) with
the corresponding LR-HSI. HSR-Diff generates an HR-HSI via repeated refinement,
in which the HR-HSI is initialized with pure Gaussian noise and iteratively
refined. At each iteration, the noise is removed with a Conditional Denoising
Transformer (CDF ormer) that is trained on denoising at different noise levels,
conditioned on the hierarchical feature maps of HR-MSI and LR-HSI. In addition,
a progressive learning strategy is employed to exploit the global information
of full-resolution images. Systematic experiments have been conducted on four
public datasets, demonstrating that HSR-Diff outperforms state-of-the-art
methods.
|
2306.12089 | Yujin Baek | Yujin Baek (1), Koanho Lee (1), Dayeon Ki (2), Hyoung-Gyu Lee (3),
Cheonbok Park (3) and Jaegul Choo (1) ((1) KAIST, (2) Korea University, (3)
Papago, Naver Corp.) | Towards Accurate Translation via Semantically Appropriate Application of
Lexical Constraints | Findings of ACL2023. 15 pages | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Lexically-constrained NMT (LNMT) aims to incorporate user-provided
terminology into translations. Despite its practical advantages, existing work
has not evaluated LNMT models under challenging real-world conditions. In this
paper, we focus on two important but under-studied issues that lie in the
current evaluation process of LNMT studies. The model needs to cope with
challenging lexical constraints that are "homographs" or "unseen" during
training. To this end, we first design a homograph disambiguation module to
differentiate the meanings of homographs. Moreover, we propose PLUMCOT, which
integrates contextually rich information about unseen lexical constraints from
pre-trained language models and strengthens a copy mechanism of the pointer
network via direct supervision of a copying score. We also release HOLLY, an
evaluation benchmark for assessing the ability of a model to cope with
"homographic" and "unseen" lexical constraints. Experiments on HOLLY and the
previous test setup show the effectiveness of our method. The effects of
PLUMCOT are shown to be remarkable in "unseen" constraints. Our dataset is
available at https://github.com/papago-lab/HOLLY-benchmark
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 08:08:15 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Baek",
"Yujin",
""
],
[
"Lee",
"Koanho",
""
],
[
"Ki",
"Dayeon",
""
],
[
"Lee",
"Hyoung-Gyu",
""
],
[
"Park",
"Cheonbok",
""
],
[
"Choo",
"Jaegul",
""
]
] | TITLE: Towards Accurate Translation via Semantically Appropriate Application of
Lexical Constraints
ABSTRACT: Lexically-constrained NMT (LNMT) aims to incorporate user-provided
terminology into translations. Despite its practical advantages, existing work
has not evaluated LNMT models under challenging real-world conditions. In this
paper, we focus on two important but under-studied issues that lie in the
current evaluation process of LNMT studies. The model needs to cope with
challenging lexical constraints that are "homographs" or "unseen" during
training. To this end, we first design a homograph disambiguation module to
differentiate the meanings of homographs. Moreover, we propose PLUMCOT, which
integrates contextually rich information about unseen lexical constraints from
pre-trained language models and strengthens a copy mechanism of the pointer
network via direct supervision of a copying score. We also release HOLLY, an
evaluation benchmark for assessing the ability of a model to cope with
"homographic" and "unseen" lexical constraints. Experiments on HOLLY and the
previous test setup show the effectiveness of our method. The effects of
PLUMCOT are shown to be remarkable in "unseen" constraints. Our dataset is
available at https://github.com/papago-lab/HOLLY-benchmark
|
2306.12091 | Jiaqi Han | Jiaqi Han, Wenbing Huang, Yu Rong, Tingyang Xu, Fuchun Sun, Junzhou
Huang | Structure-Aware DropEdge Towards Deep Graph Convolutional Networks | IEEE Transactions on Neural Networks and Learning Systems, 2023 | null | 10.1109/TNNLS.2023.3288484 | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | It has been discovered that Graph Convolutional Networks (GCNs) encounter a
remarkable drop in performance when multiple layers are piled up. The main
factor that accounts for why deep GCNs fail lies in over-smoothing, which
isolates the network output from the input with the increase of network depth,
weakening expressivity and trainability. In this paper, we start by
investigating refined measures upon DropEdge -- an existing simple yet
effective technique to relieve over-smoothing. We term our method as DropEdge++
for its two structure-aware samplers in contrast to DropEdge: layer-dependent
sampler and feature-dependent sampler. Regarding the layer-dependent sampler,
we interestingly find that increasingly sampling edges from the bottom layer
yields superior performance than the decreasing counterpart as well as
DropEdge. We theoretically reveal this phenomenon with Mean-Edge-Number (MEN),
a metric closely related to over-smoothing. For the feature-dependent sampler,
we associate the edge sampling probability with the feature similarity of node
pairs, and prove that it further correlates the convergence subspace of the
output layer with the input features. Extensive experiments on several node
classification benchmarks, including both full- and semi- supervised tasks,
illustrate the efficacy of DropEdge++ and its compatibility with a variety of
backbones by achieving generally better performance over DropEdge and the
no-drop version.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 08:11:40 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Han",
"Jiaqi",
""
],
[
"Huang",
"Wenbing",
""
],
[
"Rong",
"Yu",
""
],
[
"Xu",
"Tingyang",
""
],
[
"Sun",
"Fuchun",
""
],
[
"Huang",
"Junzhou",
""
]
] | TITLE: Structure-Aware DropEdge Towards Deep Graph Convolutional Networks
ABSTRACT: It has been discovered that Graph Convolutional Networks (GCNs) encounter a
remarkable drop in performance when multiple layers are piled up. The main
factor that accounts for why deep GCNs fail lies in over-smoothing, which
isolates the network output from the input with the increase of network depth,
weakening expressivity and trainability. In this paper, we start by
investigating refined measures upon DropEdge -- an existing simple yet
effective technique to relieve over-smoothing. We term our method as DropEdge++
for its two structure-aware samplers in contrast to DropEdge: layer-dependent
sampler and feature-dependent sampler. Regarding the layer-dependent sampler,
we interestingly find that increasingly sampling edges from the bottom layer
yields superior performance than the decreasing counterpart as well as
DropEdge. We theoretically reveal this phenomenon with Mean-Edge-Number (MEN),
a metric closely related to over-smoothing. For the feature-dependent sampler,
we associate the edge sampling probability with the feature similarity of node
pairs, and prove that it further correlates the convergence subspace of the
output layer with the input features. Extensive experiments on several node
classification benchmarks, including both full- and semi- supervised tasks,
illustrate the efficacy of DropEdge++ and its compatibility with a variety of
backbones by achieving generally better performance over DropEdge and the
no-drop version.
|
2306.12098 | Renjie Cheng | Renjie Cheng, Zhemin Zhuang, Shuxin Zhuang, Lei Xie and Jingfeng Guo | MSW-Transformer: Multi-Scale Shifted Windows Transformer Networks for
12-Lead ECG Classification | null | null | null | null | eess.SP cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Automatic classification of electrocardiogram (ECG) signals plays a crucial
role in the early prevention and diagnosis of cardiovascular diseases. While
ECG signals can be used for the diagnosis of various diseases, their
pathological characteristics exhibit minimal variations, posing a challenge to
automatic classification models. Existing methods primarily utilize
convolutional neural networks to extract ECG signal features for
classification, which may not fully capture the pathological feature
differences of different diseases. Transformer networks have advantages in
feature extraction for sequence data, but the complete network is complex and
relies on large-scale datasets. To address these challenges, we propose a
single-layer Transformer network called Multi-Scale Shifted Windows Transformer
Networks (MSW-Transformer), which uses a multi-window sliding attention
mechanism at different scales to capture features in different dimensions. The
self-attention is restricted to non-overlapping local windows via shifted
windows, and different window scales have different receptive fields. A
learnable feature fusion method is then proposed to integrate features from
different windows to further enhance model performance. Furthermore, we
visualize the attention mechanism of the multi-window shifted mechanism to
achieve better clinical interpretation in the ECG classification task. The
proposed model achieves state-of-the-art performance on five classification
tasks of the PTBXL-2020 12-lead ECG dataset, which includes 5 diagnostic
superclasses, 23 diagnostic subclasses, 12 rhythm classes, 17 morphology
classes, and 44 diagnosis classes, with average macro-F1 scores of 77.85%,
47.57%, 66.13%, 34.60%, and 34.29%, and average sample-F1 scores of 81.26%,
68.27%, 91.32%, 50.07%, and 63.19%, respectively.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 08:27:26 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Cheng",
"Renjie",
""
],
[
"Zhuang",
"Zhemin",
""
],
[
"Zhuang",
"Shuxin",
""
],
[
"Xie",
"Lei",
""
],
[
"Guo",
"Jingfeng",
""
]
] | TITLE: MSW-Transformer: Multi-Scale Shifted Windows Transformer Networks for
12-Lead ECG Classification
ABSTRACT: Automatic classification of electrocardiogram (ECG) signals plays a crucial
role in the early prevention and diagnosis of cardiovascular diseases. While
ECG signals can be used for the diagnosis of various diseases, their
pathological characteristics exhibit minimal variations, posing a challenge to
automatic classification models. Existing methods primarily utilize
convolutional neural networks to extract ECG signal features for
classification, which may not fully capture the pathological feature
differences of different diseases. Transformer networks have advantages in
feature extraction for sequence data, but the complete network is complex and
relies on large-scale datasets. To address these challenges, we propose a
single-layer Transformer network called Multi-Scale Shifted Windows Transformer
Networks (MSW-Transformer), which uses a multi-window sliding attention
mechanism at different scales to capture features in different dimensions. The
self-attention is restricted to non-overlapping local windows via shifted
windows, and different window scales have different receptive fields. A
learnable feature fusion method is then proposed to integrate features from
different windows to further enhance model performance. Furthermore, we
visualize the attention mechanism of the multi-window shifted mechanism to
achieve better clinical interpretation in the ECG classification task. The
proposed model achieves state-of-the-art performance on five classification
tasks of the PTBXL-2020 12-lead ECG dataset, which includes 5 diagnostic
superclasses, 23 diagnostic subclasses, 12 rhythm classes, 17 morphology
classes, and 44 diagnosis classes, with average macro-F1 scores of 77.85%,
47.57%, 66.13%, 34.60%, and 34.29%, and average sample-F1 scores of 81.26%,
68.27%, 91.32%, 50.07%, and 63.19%, respectively.
|
2306.12100 | Nikunj Gupta | Aditya Thakur, Harish Chauhan, Nikunj Gupta | Efficient ResNets: Residual Network Design | null | null | null | null | cs.CV cs.LG | http://creativecommons.org/publicdomain/zero/1.0/ | ResNets (or Residual Networks) are one of the most commonly used models for
image classification tasks. In this project, we design and train a modified
ResNet model for CIFAR-10 image classification. In particular, we aimed at
maximizing the test accuracy on the CIFAR-10 benchmark while keeping the size
of our ResNet model under the specified fixed budget of 5 million trainable
parameters. Model size, typically measured as the number of trainable
parameters, is important when models need to be stored on devices with limited
storage capacity (e.g. IoT/edge devices). In this article, we present our
residual network design which has less than 5 million parameters. We show that
our ResNet achieves a test accuracy of 96.04% on CIFAR-10 which is much higher
than ResNet18 (which has greater than 11 million trainable parameters) when
equipped with a number of training strategies and suitable ResNet
hyperparameters. Models and code are available at
https://github.com/Nikunj-Gupta/Efficient_ResNets.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 08:28:51 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Thakur",
"Aditya",
""
],
[
"Chauhan",
"Harish",
""
],
[
"Gupta",
"Nikunj",
""
]
] | TITLE: Efficient ResNets: Residual Network Design
ABSTRACT: ResNets (or Residual Networks) are one of the most commonly used models for
image classification tasks. In this project, we design and train a modified
ResNet model for CIFAR-10 image classification. In particular, we aimed at
maximizing the test accuracy on the CIFAR-10 benchmark while keeping the size
of our ResNet model under the specified fixed budget of 5 million trainable
parameters. Model size, typically measured as the number of trainable
parameters, is important when models need to be stored on devices with limited
storage capacity (e.g. IoT/edge devices). In this article, we present our
residual network design which has less than 5 million parameters. We show that
our ResNet achieves a test accuracy of 96.04% on CIFAR-10 which is much higher
than ResNet18 (which has greater than 11 million trainable parameters) when
equipped with a number of training strategies and suitable ResNet
hyperparameters. Models and code are available at
https://github.com/Nikunj-Gupta/Efficient_ResNets.
|
2306.12113 | Yongxin Cao | Yongxin Cao, Fanghua Liu, Lai Jiang, Cheng Bao, You Miao and Yang Chen | Lightweight wood panel defect detection method incorporating attention
mechanism and feature fusion network | null | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In recent years, deep learning has made significant progress in wood panel
defect detection. However, there are still challenges such as low detection ,
slow detection speed, and difficulties in deploying embedded devices on wood
panel surfaces. To overcome these issues, we propose a lightweight wood panel
defect detection method called YOLOv5-LW, which incorporates attention
mechanisms and a feature fusion network.Firstly, to enhance the detection
capability of acceptable defects, we introduce the Multi-scale Bi-directional
Feature Pyramid Network (MBiFPN) as a feature fusion network. The MBiFPN
reduces feature loss, enriches local and detailed features, and improves the
model's detection capability for acceptable defects.Secondly, to achieve a
lightweight design, we reconstruct the ShuffleNetv2 network model as the
backbone network. This reconstruction reduces the number of parameters and
computational requirements while maintaining performance. We also introduce the
Stem Block and Spatial Pyramid Pooling Fast (SPPF) models to compensate for any
accuracy loss resulting from the lightweight design, ensuring the model's
detection capabilities remain intact while being computationally
efficient.Thirdly, we enhance the backbone network by incorporating Efficient
Channel Attention (ECA), which improves the network's focus on key information
relevant to defect detection. By attending to essential features, the model
becomes more proficient in accurately identifying and localizing defects.We
validate the proposed method using a self-developed wood panel defect
dataset.The experimental results demonstrate the effectiveness of the improved
YOLOv5-LW method. Compared to the original model, our approach achieves a
92.8\% accuracy rate, reduces the number of parameters by 27.78\%, compresses
computational volume by 41.25\%, improves detection inference speed by 10.16\%
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 08:55:45 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Cao",
"Yongxin",
""
],
[
"Liu",
"Fanghua",
""
],
[
"Jiang",
"Lai",
""
],
[
"Bao",
"Cheng",
""
],
[
"Miao",
"You",
""
],
[
"Chen",
"Yang",
""
]
] | TITLE: Lightweight wood panel defect detection method incorporating attention
mechanism and feature fusion network
ABSTRACT: In recent years, deep learning has made significant progress in wood panel
defect detection. However, there are still challenges such as low detection ,
slow detection speed, and difficulties in deploying embedded devices on wood
panel surfaces. To overcome these issues, we propose a lightweight wood panel
defect detection method called YOLOv5-LW, which incorporates attention
mechanisms and a feature fusion network.Firstly, to enhance the detection
capability of acceptable defects, we introduce the Multi-scale Bi-directional
Feature Pyramid Network (MBiFPN) as a feature fusion network. The MBiFPN
reduces feature loss, enriches local and detailed features, and improves the
model's detection capability for acceptable defects.Secondly, to achieve a
lightweight design, we reconstruct the ShuffleNetv2 network model as the
backbone network. This reconstruction reduces the number of parameters and
computational requirements while maintaining performance. We also introduce the
Stem Block and Spatial Pyramid Pooling Fast (SPPF) models to compensate for any
accuracy loss resulting from the lightweight design, ensuring the model's
detection capabilities remain intact while being computationally
efficient.Thirdly, we enhance the backbone network by incorporating Efficient
Channel Attention (ECA), which improves the network's focus on key information
relevant to defect detection. By attending to essential features, the model
becomes more proficient in accurately identifying and localizing defects.We
validate the proposed method using a self-developed wood panel defect
dataset.The experimental results demonstrate the effectiveness of the improved
YOLOv5-LW method. Compared to the original model, our approach achieves a
92.8\% accuracy rate, reduces the number of parameters by 27.78\%, compresses
computational volume by 41.25\%, improves detection inference speed by 10.16\%
|
2306.12132 | Krishna Ronanki | Krishna Ronanki, Beatriz Cabrero-Daniel, and Christian Berger | ChatGPT as a tool for User Story Quality Evaluation: Trustworthy Out of
the Box? | 9 Pages, 2 Tables, 1 Figure. Accepted at AI-Assisted Agile Software
Development Workshop (Co-located with XP 2023) | null | null | null | cs.SE | http://creativecommons.org/licenses/by/4.0/ | In Agile software development, user stories play a vital role in capturing
and conveying end-user needs, prioritizing features, and facilitating
communication and collaboration within development teams. However, automated
methods for evaluating user stories require training in NLP tools and can be
time-consuming to develop and integrate. This study explores using ChatGPT for
user story quality evaluation and compares its performance with an existing
benchmark. Our study shows that ChatGPT's evaluation aligns well with human
evaluation, and we propose a ``best of three'' strategy to improve its output
stability. We also discuss the concept of trustworthiness in AI and its
implications for non-experts using ChatGPT's unprocessed outputs. Our research
contributes to understanding the reliability and applicability of AI in user
story evaluation and offers recommendations for future research.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 09:26:27 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Ronanki",
"Krishna",
""
],
[
"Cabrero-Daniel",
"Beatriz",
""
],
[
"Berger",
"Christian",
""
]
] | TITLE: ChatGPT as a tool for User Story Quality Evaluation: Trustworthy Out of
the Box?
ABSTRACT: In Agile software development, user stories play a vital role in capturing
and conveying end-user needs, prioritizing features, and facilitating
communication and collaboration within development teams. However, automated
methods for evaluating user stories require training in NLP tools and can be
time-consuming to develop and integrate. This study explores using ChatGPT for
user story quality evaluation and compares its performance with an existing
benchmark. Our study shows that ChatGPT's evaluation aligns well with human
evaluation, and we propose a ``best of three'' strategy to improve its output
stability. We also discuss the concept of trustworthiness in AI and its
implications for non-experts using ChatGPT's unprocessed outputs. Our research
contributes to understanding the reliability and applicability of AI in user
story evaluation and offers recommendations for future research.
|
2306.12139 | Congxi Xiao | Congxi Xiao, Jingbo Zhou, Jizhou Huang, Tong Xu, Hui Xiong | Spatial Heterophily Aware Graph Neural Networks | Accepted by KDD 2023 | null | 10.1145/3580305.3599510 | null | cs.LG cs.SI | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Graph Neural Networks (GNNs) have been broadly applied in many urban
applications upon formulating a city as an urban graph whose nodes are urban
objects like regions or points of interest. Recently, a few enhanced GNN
architectures have been developed to tackle heterophily graphs where connected
nodes are dissimilar. However, urban graphs usually can be observed to possess
a unique spatial heterophily property; that is, the dissimilarity of neighbors
at different spatial distances can exhibit great diversity. This property has
not been explored, while it often exists. To this end, in this paper, we
propose a metric, named Spatial Diversity Score, to quantitatively measure the
spatial heterophily and show how it can influence the performance of GNNs.
Indeed, our experimental investigation clearly shows that existing heterophilic
GNNs are still deficient in handling the urban graph with high spatial
diversity score. This, in turn, may degrade their effectiveness in urban
applications. Along this line, we propose a Spatial Heterophily Aware Graph
Neural Network (SHGNN), to tackle the spatial diversity of heterophily of urban
graphs. Based on the key observation that spatially close neighbors on the
urban graph present a more similar mode of difference to the central node, we
first design a rotation-scaling spatial aggregation module, whose core idea is
to properly group the spatially close neighbors and separately process each
group with less diversity inside. Then, a heterophily-sensitive spatial
interaction module is designed to adaptively capture the commonality and
diverse dissimilarity in different spatial groups. Extensive experiments on
three real-world urban datasets demonstrate the superiority of our SHGNN over
several its competitors.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 09:35:50 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Xiao",
"Congxi",
""
],
[
"Zhou",
"Jingbo",
""
],
[
"Huang",
"Jizhou",
""
],
[
"Xu",
"Tong",
""
],
[
"Xiong",
"Hui",
""
]
] | TITLE: Spatial Heterophily Aware Graph Neural Networks
ABSTRACT: Graph Neural Networks (GNNs) have been broadly applied in many urban
applications upon formulating a city as an urban graph whose nodes are urban
objects like regions or points of interest. Recently, a few enhanced GNN
architectures have been developed to tackle heterophily graphs where connected
nodes are dissimilar. However, urban graphs usually can be observed to possess
a unique spatial heterophily property; that is, the dissimilarity of neighbors
at different spatial distances can exhibit great diversity. This property has
not been explored, while it often exists. To this end, in this paper, we
propose a metric, named Spatial Diversity Score, to quantitatively measure the
spatial heterophily and show how it can influence the performance of GNNs.
Indeed, our experimental investigation clearly shows that existing heterophilic
GNNs are still deficient in handling the urban graph with high spatial
diversity score. This, in turn, may degrade their effectiveness in urban
applications. Along this line, we propose a Spatial Heterophily Aware Graph
Neural Network (SHGNN), to tackle the spatial diversity of heterophily of urban
graphs. Based on the key observation that spatially close neighbors on the
urban graph present a more similar mode of difference to the central node, we
first design a rotation-scaling spatial aggregation module, whose core idea is
to properly group the spatially close neighbors and separately process each
group with less diversity inside. Then, a heterophily-sensitive spatial
interaction module is designed to adaptively capture the commonality and
diverse dissimilarity in different spatial groups. Extensive experiments on
three real-world urban datasets demonstrate the superiority of our SHGNN over
several its competitors.
|
2306.12144 | Ying Li | Ying Li, Xiaodong Lee, Botao Peng, Themis Palpanas, Jingan Xue | PrivSketch: A Private Sketch-based Frequency Estimation Protocol for
Data Streams | null | null | null | null | cs.CR | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Local differential privacy (LDP) has recently become a popular
privacy-preserving data collection technique protecting users' privacy. The
main problem of data stream collection under LDP is the poor utility due to
multi-item collection from a very large domain. This paper proposes PrivSketch,
a high-utility frequency estimation protocol taking advantage of sketches,
suitable for private data stream collection. Combining the proposed background
information and a decode-first collection-side workflow, PrivSketch improves
the utility by reducing the errors introduced by the sketching algorithm and
the privacy budget utilization when collecting multiple items. We analytically
prove the superior accuracy and privacy characteristics of PrivSketch, and also
evaluate them experimentally. Our evaluation, with several diverse synthetic
and real datasets, demonstrates that PrivSketch is 1-3 orders of magnitude
better than the competitors in terms of utility in both frequency estimation
and frequent item estimation, while being up to ~100x faster.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 09:42:13 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Li",
"Ying",
""
],
[
"Lee",
"Xiaodong",
""
],
[
"Peng",
"Botao",
""
],
[
"Palpanas",
"Themis",
""
],
[
"Xue",
"Jingan",
""
]
] | TITLE: PrivSketch: A Private Sketch-based Frequency Estimation Protocol for
Data Streams
ABSTRACT: Local differential privacy (LDP) has recently become a popular
privacy-preserving data collection technique protecting users' privacy. The
main problem of data stream collection under LDP is the poor utility due to
multi-item collection from a very large domain. This paper proposes PrivSketch,
a high-utility frequency estimation protocol taking advantage of sketches,
suitable for private data stream collection. Combining the proposed background
information and a decode-first collection-side workflow, PrivSketch improves
the utility by reducing the errors introduced by the sketching algorithm and
the privacy budget utilization when collecting multiple items. We analytically
prove the superior accuracy and privacy characteristics of PrivSketch, and also
evaluate them experimentally. Our evaluation, with several diverse synthetic
and real datasets, demonstrates that PrivSketch is 1-3 orders of magnitude
better than the competitors in terms of utility in both frequency estimation
and frequent item estimation, while being up to ~100x faster.
|
2306.12150 | Rick Wilming | Marta Oliveira, Rick Wilming, Benedict Clark, C\'eline Budding, Fabian
Eitel, Kerstin Ritter, Stefan Haufe | Benchmark data to study the influence of pre-training on explanation
performance in MR image classification | Under review | null | null | null | cs.CV cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Convolutional Neural Networks (CNNs) are frequently and successfully used in
medical prediction tasks. They are often used in combination with transfer
learning, leading to improved performance when training data for the task are
scarce. The resulting models are highly complex and typically do not provide
any insight into their predictive mechanisms, motivating the field of
'explainable' artificial intelligence (XAI). However, previous studies have
rarely quantitatively evaluated the 'explanation performance' of XAI methods
against ground-truth data, and transfer learning and its influence on objective
measures of explanation performance has not been investigated. Here, we propose
a benchmark dataset that allows for quantifying explanation performance in a
realistic magnetic resonance imaging (MRI) classification task. We employ this
benchmark to understand the influence of transfer learning on the quality of
explanations. Experimental results show that popular XAI methods applied to the
same underlying model differ vastly in performance, even when considering only
correctly classified examples. We further observe that explanation performance
strongly depends on the task used for pre-training and the number of CNN layers
pre-trained. These results hold after correcting for a substantial correlation
between explanation and classification performance.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 09:53:37 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Oliveira",
"Marta",
""
],
[
"Wilming",
"Rick",
""
],
[
"Clark",
"Benedict",
""
],
[
"Budding",
"Céline",
""
],
[
"Eitel",
"Fabian",
""
],
[
"Ritter",
"Kerstin",
""
],
[
"Haufe",
"Stefan",
""
]
] | TITLE: Benchmark data to study the influence of pre-training on explanation
performance in MR image classification
ABSTRACT: Convolutional Neural Networks (CNNs) are frequently and successfully used in
medical prediction tasks. They are often used in combination with transfer
learning, leading to improved performance when training data for the task are
scarce. The resulting models are highly complex and typically do not provide
any insight into their predictive mechanisms, motivating the field of
'explainable' artificial intelligence (XAI). However, previous studies have
rarely quantitatively evaluated the 'explanation performance' of XAI methods
against ground-truth data, and transfer learning and its influence on objective
measures of explanation performance has not been investigated. Here, we propose
a benchmark dataset that allows for quantifying explanation performance in a
realistic magnetic resonance imaging (MRI) classification task. We employ this
benchmark to understand the influence of transfer learning on the quality of
explanations. Experimental results show that popular XAI methods applied to the
same underlying model differ vastly in performance, even when considering only
correctly classified examples. We further observe that explanation performance
strongly depends on the task used for pre-training and the number of CNN layers
pre-trained. These results hold after correcting for a substantial correlation
between explanation and classification performance.
|
2306.12156 | Yongqi An | Xu Zhao, Wenchao Ding, Yongqi An, Yinglong Du, Tao Yu, Min Li, Ming
Tang, Jinqiao Wang | Fast Segment Anything | Technical Report. The code is released at
https://github.com/CASIA-IVA-Lab/FastSAM | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The recently proposed segment anything model (SAM) has made a significant
influence in many computer vision tasks. It is becoming a foundation step for
many high-level tasks, like image segmentation, image caption, and image
editing. However, its huge computation costs prevent it from wider applications
in industry scenarios. The computation mainly comes from the Transformer
architecture at high-resolution inputs. In this paper, we propose a speed-up
alternative method for this fundamental task with comparable performance. By
reformulating the task as segments-generation and prompting, we find that a
regular CNN detector with an instance segmentation branch can also accomplish
this task well. Specifically, we convert this task to the well-studied instance
segmentation task and directly train the existing instance segmentation method
using only 1/50 of the SA-1B dataset published by SAM authors. With our method,
we achieve a comparable performance with the SAM method at 50 times higher
run-time speed. We give sufficient experimental results to demonstrate its
effectiveness. The codes and demos will be released at
https://github.com/CASIA-IVA-Lab/FastSAM.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 10:08:29 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Zhao",
"Xu",
""
],
[
"Ding",
"Wenchao",
""
],
[
"An",
"Yongqi",
""
],
[
"Du",
"Yinglong",
""
],
[
"Yu",
"Tao",
""
],
[
"Li",
"Min",
""
],
[
"Tang",
"Ming",
""
],
[
"Wang",
"Jinqiao",
""
]
] | TITLE: Fast Segment Anything
ABSTRACT: The recently proposed segment anything model (SAM) has made a significant
influence in many computer vision tasks. It is becoming a foundation step for
many high-level tasks, like image segmentation, image caption, and image
editing. However, its huge computation costs prevent it from wider applications
in industry scenarios. The computation mainly comes from the Transformer
architecture at high-resolution inputs. In this paper, we propose a speed-up
alternative method for this fundamental task with comparable performance. By
reformulating the task as segments-generation and prompting, we find that a
regular CNN detector with an instance segmentation branch can also accomplish
this task well. Specifically, we convert this task to the well-studied instance
segmentation task and directly train the existing instance segmentation method
using only 1/50 of the SA-1B dataset published by SAM authors. With our method,
we achieve a comparable performance with the SAM method at 50 times higher
run-time speed. We give sufficient experimental results to demonstrate its
effectiveness. The codes and demos will be released at
https://github.com/CASIA-IVA-Lab/FastSAM.
|
2306.12161 | Mouna Rabhi | Mouna Rabhi and Roberto Di Pietro | Adversarial Attacks Neutralization via Data Set Randomization | null | null | null | null | cs.LG cs.AI cs.CR | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Adversarial attacks on deep-learning models pose a serious threat to their
reliability and security. Existing defense mechanisms are narrow addressing a
specific type of attack or being vulnerable to sophisticated attacks. We
propose a new defense mechanism that, while being focused on image-based
classifiers, is general with respect to the cited category. It is rooted on
hyperspace projection. In particular, our solution provides a pseudo-random
projection of the original dataset into a new dataset. The proposed defense
mechanism creates a set of diverse projected datasets, where each projected
dataset is used to train a specific classifier, resulting in different trained
classifiers with different decision boundaries. During testing, it randomly
selects a classifier to test the input. Our approach does not sacrifice
accuracy over legitimate input. Other than detailing and providing a thorough
characterization of our defense mechanism, we also provide a proof of concept
of using four optimization-based adversarial attacks (PGD, FGSM, IGSM, and
C\&W) and a generative adversarial attack testing them on the MNIST dataset.
Our experimental results show that our solution increases the robustness of
deep learning models against adversarial attacks and significantly reduces the
attack success rate by at least 89% for optimization attacks and 78% for
generative attacks. We also analyze the relationship between the number of used
hyperspaces and the efficacy of the defense mechanism. As expected, the two are
positively correlated, offering an easy-to-tune parameter to enforce the
desired level of security. The generality and scalability of our solution and
adaptability to different attack scenarios, combined with the excellent
achieved results, other than providing a robust defense against adversarial
attacks on deep learning networks, also lay the groundwork for future research
in the field.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 10:17:55 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Rabhi",
"Mouna",
""
],
[
"Di Pietro",
"Roberto",
""
]
] | TITLE: Adversarial Attacks Neutralization via Data Set Randomization
ABSTRACT: Adversarial attacks on deep-learning models pose a serious threat to their
reliability and security. Existing defense mechanisms are narrow addressing a
specific type of attack or being vulnerable to sophisticated attacks. We
propose a new defense mechanism that, while being focused on image-based
classifiers, is general with respect to the cited category. It is rooted on
hyperspace projection. In particular, our solution provides a pseudo-random
projection of the original dataset into a new dataset. The proposed defense
mechanism creates a set of diverse projected datasets, where each projected
dataset is used to train a specific classifier, resulting in different trained
classifiers with different decision boundaries. During testing, it randomly
selects a classifier to test the input. Our approach does not sacrifice
accuracy over legitimate input. Other than detailing and providing a thorough
characterization of our defense mechanism, we also provide a proof of concept
of using four optimization-based adversarial attacks (PGD, FGSM, IGSM, and
C\&W) and a generative adversarial attack testing them on the MNIST dataset.
Our experimental results show that our solution increases the robustness of
deep learning models against adversarial attacks and significantly reduces the
attack success rate by at least 89% for optimization attacks and 78% for
generative attacks. We also analyze the relationship between the number of used
hyperspaces and the efficacy of the defense mechanism. As expected, the two are
positively correlated, offering an easy-to-tune parameter to enforce the
desired level of security. The generality and scalability of our solution and
adaptability to different attack scenarios, combined with the excellent
achieved results, other than providing a robust defense against adversarial
attacks on deep learning networks, also lay the groundwork for future research
in the field.
|
2306.12188 | Nati Daniel | Ariel Larey, Omri Asraf, Adam Kelder, Itzik Wilf, Ofer Kruzel, Nati
Daniel | Facial Expression Re-targeting from a Single Character | null | null | null | null | cs.GR cs.AI cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Video retargeting for digital face animation is used in virtual reality,
social media, gaming, movies, and video conference, aiming to animate avatars'
facial expressions based on videos of human faces. The standard method to
represent facial expressions for 3D characters is by blendshapes, a vector of
weights representing the avatar's neutral shape and its variations under facial
expressions, e.g., smile, puff, blinking. Datasets of paired frames with
blendshape vectors are rare, and labeling can be laborious, time-consuming, and
subjective. In this work, we developed an approach that handles the lack of
appropriate datasets. Instead, we used a synthetic dataset of only one
character. To generalize various characters, we re-represented each frame to
face landmarks. We developed a unique deep-learning architecture that groups
landmarks for each facial organ and connects them to relevant blendshape
weights. Additionally, we incorporated complementary methods for facial
expressions that landmarks did not represent well and gave special attention to
eye expressions. We have demonstrated the superiority of our approach to
previous research in qualitative and quantitative metrics. Our approach
achieved a higher MOS of 68% and a lower MSE of 44.2% when tested on videos
with various users and expressions.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 11:35:22 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Larey",
"Ariel",
""
],
[
"Asraf",
"Omri",
""
],
[
"Kelder",
"Adam",
""
],
[
"Wilf",
"Itzik",
""
],
[
"Kruzel",
"Ofer",
""
],
[
"Daniel",
"Nati",
""
]
] | TITLE: Facial Expression Re-targeting from a Single Character
ABSTRACT: Video retargeting for digital face animation is used in virtual reality,
social media, gaming, movies, and video conference, aiming to animate avatars'
facial expressions based on videos of human faces. The standard method to
represent facial expressions for 3D characters is by blendshapes, a vector of
weights representing the avatar's neutral shape and its variations under facial
expressions, e.g., smile, puff, blinking. Datasets of paired frames with
blendshape vectors are rare, and labeling can be laborious, time-consuming, and
subjective. In this work, we developed an approach that handles the lack of
appropriate datasets. Instead, we used a synthetic dataset of only one
character. To generalize various characters, we re-represented each frame to
face landmarks. We developed a unique deep-learning architecture that groups
landmarks for each facial organ and connects them to relevant blendshape
weights. Additionally, we incorporated complementary methods for facial
expressions that landmarks did not represent well and gave special attention to
eye expressions. We have demonstrated the superiority of our approach to
previous research in qualitative and quantitative metrics. Our approach
achieved a higher MOS of 68% and a lower MSE of 44.2% when tested on videos
with various users and expressions.
|
2306.12198 | Mohamad Ballout | Mohamad Ballout and Ulf Krumnack and Gunther Heidemann and Kai-Uwe
K\"uhnberger | Opening the Black Box: Analyzing Attention Weights and Hidden States in
Pre-trained Language Models for Non-language Tasks | null | null | null | null | cs.CL cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Investigating deep learning language models has always been a significant
research area due to the ``black box" nature of most advanced models. With the
recent advancements in pre-trained language models based on transformers and
their increasing integration into daily life, addressing this issue has become
more pressing. In order to achieve an explainable AI model, it is essential to
comprehend the procedural steps involved and compare them with human thought
processes. Thus, in this paper, we use simple, well-understood non-language
tasks to explore these models' inner workings. Specifically, we apply a
pre-trained language model to constrained arithmetic problems with hierarchical
structure, to analyze their attention weight scores and hidden states. The
investigation reveals promising results, with the model addressing hierarchical
problems in a moderately structured manner, similar to human problem-solving
strategies. Additionally, by inspecting the attention weights layer by layer,
we uncover an unconventional finding that layer 10, rather than the model's
final layer, is the optimal layer to unfreeze for the least parameter-intensive
approach to fine-tune the model. We support these findings with entropy
analysis and token embeddings similarity analysis. The attention analysis
allows us to hypothesize that the model can generalize to longer sequences in
ListOps dataset, a conclusion later confirmed through testing on sequences
longer than those in the training set. Lastly, by utilizing a straightforward
task in which the model predicts the winner of a Tic Tac Toe game, we identify
limitations in attention analysis, particularly its inability to capture 2D
patterns.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 11:48:07 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Ballout",
"Mohamad",
""
],
[
"Krumnack",
"Ulf",
""
],
[
"Heidemann",
"Gunther",
""
],
[
"Kühnberger",
"Kai-Uwe",
""
]
] | TITLE: Opening the Black Box: Analyzing Attention Weights and Hidden States in
Pre-trained Language Models for Non-language Tasks
ABSTRACT: Investigating deep learning language models has always been a significant
research area due to the ``black box" nature of most advanced models. With the
recent advancements in pre-trained language models based on transformers and
their increasing integration into daily life, addressing this issue has become
more pressing. In order to achieve an explainable AI model, it is essential to
comprehend the procedural steps involved and compare them with human thought
processes. Thus, in this paper, we use simple, well-understood non-language
tasks to explore these models' inner workings. Specifically, we apply a
pre-trained language model to constrained arithmetic problems with hierarchical
structure, to analyze their attention weight scores and hidden states. The
investigation reveals promising results, with the model addressing hierarchical
problems in a moderately structured manner, similar to human problem-solving
strategies. Additionally, by inspecting the attention weights layer by layer,
we uncover an unconventional finding that layer 10, rather than the model's
final layer, is the optimal layer to unfreeze for the least parameter-intensive
approach to fine-tune the model. We support these findings with entropy
analysis and token embeddings similarity analysis. The attention analysis
allows us to hypothesize that the model can generalize to longer sequences in
ListOps dataset, a conclusion later confirmed through testing on sequences
longer than those in the training set. Lastly, by utilizing a straightforward
task in which the model predicts the winner of a Tic Tac Toe game, we identify
limitations in attention analysis, particularly its inability to capture 2D
patterns.
|
2306.12203 | David Gillsj\"o | David Gillsj\"o, Gabrielle Flood, Kalle {\AA}str\"om | Polygon Detection for Room Layout Estimation using Heterogeneous Graphs
and Wireframes | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents a neural network based semantic plane detection method
utilizing polygon representations. The method can for example be used to solve
room layout estimations tasks. The method is built on, combines and further
develops several different modules from previous research. The network takes an
RGB image and estimates a wireframe as well as a feature space using an
hourglass backbone. From these, line and junction features are sampled. The
lines and junctions are then represented as an undirected graph, from which
polygon representations of the sought planes are obtained. Two different
methods for this last step are investigated, where the most promising method is
built on a heterogeneous graph transformer. The final output is in all cases a
projection of the semantic planes in 2D. The methods are evaluated on the
Structured 3D dataset and we investigate the performance both using sampled and
estimated wireframes. The experiments show the potential of the graph-based
method by outperforming state of the art methods in Room Layout estimation in
the 2D metrics using synthetic wireframe detections.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 11:55:15 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Gillsjö",
"David",
""
],
[
"Flood",
"Gabrielle",
""
],
[
"Åström",
"Kalle",
""
]
] | TITLE: Polygon Detection for Room Layout Estimation using Heterogeneous Graphs
and Wireframes
ABSTRACT: This paper presents a neural network based semantic plane detection method
utilizing polygon representations. The method can for example be used to solve
room layout estimations tasks. The method is built on, combines and further
develops several different modules from previous research. The network takes an
RGB image and estimates a wireframe as well as a feature space using an
hourglass backbone. From these, line and junction features are sampled. The
lines and junctions are then represented as an undirected graph, from which
polygon representations of the sought planes are obtained. Two different
methods for this last step are investigated, where the most promising method is
built on a heterogeneous graph transformer. The final output is in all cases a
projection of the semantic planes in 2D. The methods are evaluated on the
Structured 3D dataset and we investigate the performance both using sampled and
estimated wireframes. The experiments show the potential of the graph-based
method by outperforming state of the art methods in Room Layout estimation in
the 2D metrics using synthetic wireframe detections.
|
2306.12205 | Mohamad Ballout | Mohamad Ballout, Ulf Krumnack, Gunther Heidemann and Kai-Uwe
K\"uhnberger | Investigating Pre-trained Language Models on Cross-Domain Datasets, a
Step Closer to General AI | null | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Pre-trained language models have recently emerged as a powerful tool for
fine-tuning a variety of language tasks. Ideally, when models are pre-trained
on large amount of data, they are expected to gain implicit knowledge. In this
paper, we investigate the ability of pre-trained language models to generalize
to different non-language tasks. In particular, we test them on tasks from
different domains such as computer vision, reasoning on hierarchical data, and
protein fold prediction. The four pre-trained models that we used, T5, BART,
BERT, and GPT-2 achieve outstanding results. They all have similar performance
and they outperform transformers that are trained from scratch by a large
margin. For instance, pre-trained language models perform better on the Listops
dataset, with an average accuracy of 58.7\%, compared to transformers trained
from scratch, which have an average accuracy of 29.0\%. The significant
improvement demonstrated across three types of datasets suggests that
pre-training on language helps the models to acquire general knowledge,
bringing us a step closer to general AI. We also showed that reducing the
number of parameters in pre-trained language models does not have a great
impact as the performance drops slightly when using T5-Small instead of
T5-Base. In fact, when using only 2\% of the parameters, we achieved a great
improvement compared to training from scratch. Finally, in contrast to prior
work, we find out that using pre-trained embeddings for the input layer is
necessary to achieve the desired results.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 11:55:17 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Ballout",
"Mohamad",
""
],
[
"Krumnack",
"Ulf",
""
],
[
"Heidemann",
"Gunther",
""
],
[
"Kühnberger",
"Kai-Uwe",
""
]
] | TITLE: Investigating Pre-trained Language Models on Cross-Domain Datasets, a
Step Closer to General AI
ABSTRACT: Pre-trained language models have recently emerged as a powerful tool for
fine-tuning a variety of language tasks. Ideally, when models are pre-trained
on large amount of data, they are expected to gain implicit knowledge. In this
paper, we investigate the ability of pre-trained language models to generalize
to different non-language tasks. In particular, we test them on tasks from
different domains such as computer vision, reasoning on hierarchical data, and
protein fold prediction. The four pre-trained models that we used, T5, BART,
BERT, and GPT-2 achieve outstanding results. They all have similar performance
and they outperform transformers that are trained from scratch by a large
margin. For instance, pre-trained language models perform better on the Listops
dataset, with an average accuracy of 58.7\%, compared to transformers trained
from scratch, which have an average accuracy of 29.0\%. The significant
improvement demonstrated across three types of datasets suggests that
pre-training on language helps the models to acquire general knowledge,
bringing us a step closer to general AI. We also showed that reducing the
number of parameters in pre-trained language models does not have a great
impact as the performance drops slightly when using T5-Small instead of
T5-Base. In fact, when using only 2\% of the parameters, we achieved a great
improvement compared to training from scratch. Finally, in contrast to prior
work, we find out that using pre-trained embeddings for the input layer is
necessary to achieve the desired results.
|
2306.12232 | Xuanji Xiao | Wanda Li, Wenhao Zheng, Xuanji Xiao, Suhang Wang | STAN: Stage-Adaptive Network for Multi-Task Recommendation by Learning
User Lifecycle-Based Representation | null | RecSys 2023 | null | null | cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recommendation systems play a vital role in many online platforms, with their
primary objective being to satisfy and retain users. As directly optimizing
user retention is challenging, multiple evaluation metrics are often employed.
Existing methods generally formulate the optimization of these evaluation
metrics as a multitask learning problem, but often overlook the fact that user
preferences for different tasks are personalized and change over time.
Identifying and tracking the evolution of user preferences can lead to better
user retention. To address this issue, we introduce the concept of "user
lifecycle", consisting of multiple stages characterized by users' varying
preferences for different tasks. We propose a novel Stage-Adaptive Network
(STAN) framework for modeling user lifecycle stages. STAN first identifies
latent user lifecycle stages based on learned user preferences, and then
employs the stage representation to enhance multi-task learning performance.
Our experimental results using both public and industrial datasets demonstrate
that the proposed model significantly improves multi-task prediction
performance compared to state-of-the-art methods, highlighting the importance
of considering user lifecycle stages in recommendation systems. Furthermore,
online A/B testing reveals that our model outperforms the existing model,
achieving a significant improvement of 3.05% in staytime per user and 0.88% in
CVR. These results indicate that our approach effectively improves the overall
efficiency of the multi-task recommendation system.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 12:45:17 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Li",
"Wanda",
""
],
[
"Zheng",
"Wenhao",
""
],
[
"Xiao",
"Xuanji",
""
],
[
"Wang",
"Suhang",
""
]
] | TITLE: STAN: Stage-Adaptive Network for Multi-Task Recommendation by Learning
User Lifecycle-Based Representation
ABSTRACT: Recommendation systems play a vital role in many online platforms, with their
primary objective being to satisfy and retain users. As directly optimizing
user retention is challenging, multiple evaluation metrics are often employed.
Existing methods generally formulate the optimization of these evaluation
metrics as a multitask learning problem, but often overlook the fact that user
preferences for different tasks are personalized and change over time.
Identifying and tracking the evolution of user preferences can lead to better
user retention. To address this issue, we introduce the concept of "user
lifecycle", consisting of multiple stages characterized by users' varying
preferences for different tasks. We propose a novel Stage-Adaptive Network
(STAN) framework for modeling user lifecycle stages. STAN first identifies
latent user lifecycle stages based on learned user preferences, and then
employs the stage representation to enhance multi-task learning performance.
Our experimental results using both public and industrial datasets demonstrate
that the proposed model significantly improves multi-task prediction
performance compared to state-of-the-art methods, highlighting the importance
of considering user lifecycle stages in recommendation systems. Furthermore,
online A/B testing reveals that our model outperforms the existing model,
achieving a significant improvement of 3.05% in staytime per user and 0.88% in
CVR. These results indicate that our approach effectively improves the overall
efficiency of the multi-task recommendation system.
|
2306.12242 | Adam Marcus | Adam Marcus, Paul Bentley, Daniel Rueckert | Concurrent ischemic lesion age estimation and segmentation of CT brain
using a Transformer-based network | null | null | 10.1109/TMI.2023.3287361 | null | eess.IV cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | The cornerstone of stroke care is expedient management that varies depending
on the time since stroke onset. Consequently, clinical decision making is
centered on accurate knowledge of timing and often requires a radiologist to
interpret Computed Tomography (CT) of the brain to confirm the occurrence and
age of an event. These tasks are particularly challenging due to the subtle
expression of acute ischemic lesions and the dynamic nature of their
appearance. Automation efforts have not yet applied deep learning to estimate
lesion age and treated these two tasks independently, so, have overlooked their
inherent complementary relationship. To leverage this, we propose a novel
end-to-end multi-task transformer-based network optimized for concurrent
segmentation and age estimation of cerebral ischemic lesions. By utilizing
gated positional self-attention and CT-specific data augmentation, the proposed
method can capture long-range spatial dependencies while maintaining its
ability to be trained from scratch under low-data regimes commonly found in
medical imaging. Furthermore, to better combine multiple predictions, we
incorporate uncertainty by utilizing quantile loss to facilitate estimating a
probability density function of lesion age. The effectiveness of our model is
then extensively evaluated on a clinical dataset consisting of 776 CT images
from two medical centers. Experimental results demonstrate that our method
obtains promising performance, with an area under the curve (AUC) of 0.933 for
classifying lesion ages <=4.5 hours compared to 0.858 using a conventional
approach, and outperforms task-specific state-of-the-art algorithms.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 13:00:49 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Marcus",
"Adam",
""
],
[
"Bentley",
"Paul",
""
],
[
"Rueckert",
"Daniel",
""
]
] | TITLE: Concurrent ischemic lesion age estimation and segmentation of CT brain
using a Transformer-based network
ABSTRACT: The cornerstone of stroke care is expedient management that varies depending
on the time since stroke onset. Consequently, clinical decision making is
centered on accurate knowledge of timing and often requires a radiologist to
interpret Computed Tomography (CT) of the brain to confirm the occurrence and
age of an event. These tasks are particularly challenging due to the subtle
expression of acute ischemic lesions and the dynamic nature of their
appearance. Automation efforts have not yet applied deep learning to estimate
lesion age and treated these two tasks independently, so, have overlooked their
inherent complementary relationship. To leverage this, we propose a novel
end-to-end multi-task transformer-based network optimized for concurrent
segmentation and age estimation of cerebral ischemic lesions. By utilizing
gated positional self-attention and CT-specific data augmentation, the proposed
method can capture long-range spatial dependencies while maintaining its
ability to be trained from scratch under low-data regimes commonly found in
medical imaging. Furthermore, to better combine multiple predictions, we
incorporate uncertainty by utilizing quantile loss to facilitate estimating a
probability density function of lesion age. The effectiveness of our model is
then extensively evaluated on a clinical dataset consisting of 776 CT images
from two medical centers. Experimental results demonstrate that our method
obtains promising performance, with an area under the curve (AUC) of 0.933 for
classifying lesion ages <=4.5 hours compared to 0.858 using a conventional
approach, and outperforms task-specific state-of-the-art algorithms.
|
2306.12255 | Carolyn Anderson | Jingmiao Zhao and Carolyn Jane Anderson | Solving and Generating NPR Sunday Puzzles with Large Language Models | To appear in the Proceedings of the 14th International Conference on
Computational Creativity (ICCC) | null | null | null | cs.CL | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We explore the ability of large language models to solve and generate puzzles
from the NPR Sunday Puzzle game show using PUZZLEQA, a dataset comprising 15
years of on-air puzzles. We evaluate four large language models using PUZZLEQA,
in both multiple choice and free response formats, and explore two prompt
engineering techniques to improve free response performance: chain-of-thought
reasoning and prompt summarization. We find that state-of-the-art large
language models can solve many PUZZLEQA puzzles: the best model, GPT-3.5,
achieves 50.2% loose accuracy. However, in our few-shot puzzle generation
experiment, we find no evidence that models can generate puzzles: GPT-3.5
generates puzzles with answers that do not conform to the generated rules.
Puzzle generation remains a challenging task for future work.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 13:23:48 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Zhao",
"Jingmiao",
""
],
[
"Anderson",
"Carolyn Jane",
""
]
] | TITLE: Solving and Generating NPR Sunday Puzzles with Large Language Models
ABSTRACT: We explore the ability of large language models to solve and generate puzzles
from the NPR Sunday Puzzle game show using PUZZLEQA, a dataset comprising 15
years of on-air puzzles. We evaluate four large language models using PUZZLEQA,
in both multiple choice and free response formats, and explore two prompt
engineering techniques to improve free response performance: chain-of-thought
reasoning and prompt summarization. We find that state-of-the-art large
language models can solve many PUZZLEQA puzzles: the best model, GPT-3.5,
achieves 50.2% loose accuracy. However, in our few-shot puzzle generation
experiment, we find no evidence that models can generate puzzles: GPT-3.5
generates puzzles with answers that do not conform to the generated rules.
Puzzle generation remains a challenging task for future work.
|
2306.12272 | Rui Patrick Xian | R. Patrick Xian, Ryan J. Morelock, Ido Hadar, Charles B. Musgrave,
Christopher Sutton | From structure mining to unsupervised exploration of atomic octahedral
networks | 56 pages | null | null | null | cond-mat.mtrl-sci cs.CE cs.LG math.CO | http://creativecommons.org/licenses/by-sa/4.0/ | Networks of atom-centered coordination octahedra commonly occur in inorganic
and hybrid solid-state materials. Characterizing their spatial arrangements and
characteristics is crucial for relating structures to properties for many
materials families. The traditional method using case-by-case inspection
becomes prohibitive for discovering trends and similarities in large datasets.
Here, we operationalize chemical intuition to automate the geometric parsing,
quantification, and classification of coordination octahedral networks. We find
axis-resolved tilting trends in ABO$_{3}$ perovskite polymorphs, which assist
in detecting oxidation state changes. Moreover, we develop a scale-invariant
encoding scheme to represent these networks, which, combined with
human-assisted unsupervised machine learning, allows us to taxonomize the
inorganic framework polytypes in hybrid iodoplumbates (A$_x$Pb$_y$I$_z$).
Consequently, we uncover a violation of Pauling's third rule and the design
principles underpinning their topological diversity. Our results offer a
glimpse into the vast design space of atomic octahedral networks and inform
high-throughput, targeted screening of specific structure types.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 13:49:35 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Xian",
"R. Patrick",
""
],
[
"Morelock",
"Ryan J.",
""
],
[
"Hadar",
"Ido",
""
],
[
"Musgrave",
"Charles B.",
""
],
[
"Sutton",
"Christopher",
""
]
] | TITLE: From structure mining to unsupervised exploration of atomic octahedral
networks
ABSTRACT: Networks of atom-centered coordination octahedra commonly occur in inorganic
and hybrid solid-state materials. Characterizing their spatial arrangements and
characteristics is crucial for relating structures to properties for many
materials families. The traditional method using case-by-case inspection
becomes prohibitive for discovering trends and similarities in large datasets.
Here, we operationalize chemical intuition to automate the geometric parsing,
quantification, and classification of coordination octahedral networks. We find
axis-resolved tilting trends in ABO$_{3}$ perovskite polymorphs, which assist
in detecting oxidation state changes. Moreover, we develop a scale-invariant
encoding scheme to represent these networks, which, combined with
human-assisted unsupervised machine learning, allows us to taxonomize the
inorganic framework polytypes in hybrid iodoplumbates (A$_x$Pb$_y$I$_z$).
Consequently, we uncover a violation of Pauling's third rule and the design
principles underpinning their topological diversity. Our results offer a
glimpse into the vast design space of atomic octahedral networks and inform
high-throughput, targeted screening of specific structure types.
|
2306.12276 | Emadeldeen Hamdan | Ziliang Hong, Emadeldeen Hamdan, Yifei Zhao, Tianxiao Ye, Hongyi Pan,
A. Enis Cetin | Wildfire Detection Via Transfer Learning: A Survey | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | This paper surveys different publicly available neural network models used
for detecting wildfires using regular visible-range cameras which are placed on
hilltops or forest lookout towers. The neural network models are pre-trained on
ImageNet-1K and fine-tuned on a custom wildfire dataset. The performance of
these models is evaluated on a diverse set of wildfire images, and the survey
provides useful information for those interested in using transfer learning for
wildfire detection. Swin Transformer-tiny has the highest AUC value but
ConvNext-tiny detects all the wildfire events and has the lowest false alarm
rate in our dataset.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 13:57:04 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Hong",
"Ziliang",
""
],
[
"Hamdan",
"Emadeldeen",
""
],
[
"Zhao",
"Yifei",
""
],
[
"Ye",
"Tianxiao",
""
],
[
"Pan",
"Hongyi",
""
],
[
"Cetin",
"A. Enis",
""
]
] | TITLE: Wildfire Detection Via Transfer Learning: A Survey
ABSTRACT: This paper surveys different publicly available neural network models used
for detecting wildfires using regular visible-range cameras which are placed on
hilltops or forest lookout towers. The neural network models are pre-trained on
ImageNet-1K and fine-tuned on a custom wildfire dataset. The performance of
these models is evaluated on a diverse set of wildfire images, and the survey
provides useful information for those interested in using transfer learning for
wildfire detection. Swin Transformer-tiny has the highest AUC value but
ConvNext-tiny detects all the wildfire events and has the lowest false alarm
rate in our dataset.
|
2306.12282 | Negin Golrezaei | Negin Golrezaei and Patrick Jaillet and Zijie Zhou | Online Resource Allocation with Convex-set Machine-Learned Advice | 74 pages, 5 figures | null | null | null | cs.DS cs.LG math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Decision-makers often have access to a machine-learned prediction about
demand, referred to as advice, which can potentially be utilized in online
decision-making processes for resource allocation. However, exploiting such
advice poses challenges due to its potential inaccuracy. To address this issue,
we propose a framework that enhances online resource allocation decisions with
potentially unreliable machine-learned (ML) advice. We assume here that this
advice is represented by a general convex uncertainty set for the demand
vector.
We introduce a parameterized class of Pareto optimal online resource
allocation algorithms that strike a balance between consistent and robust
ratios. The consistent ratio measures the algorithm's performance (compared to
the optimal hindsight solution) when the ML advice is accurate, while the
robust ratio captures performance under an adversarial demand process when the
advice is inaccurate. Specifically, in a C-Pareto optimal setting, we maximize
the robust ratio while ensuring that the consistent ratio is at least C. Our
proposed C-Pareto optimal algorithm is an adaptive protection level algorithm,
which extends the classical fixed protection level algorithm introduced in
Littlewood (2005) and Ball and Queyranne (2009). Solving a complex non-convex
continuous optimization problem characterizes the adaptive protection level
algorithm. To complement our algorithms, we present a simple method for
computing the maximum achievable consistent ratio, which serves as an estimate
for the maximum value of the ML advice. Additionally, we present numerical
studies to evaluate the performance of our algorithm in comparison to benchmark
algorithms. The results demonstrate that by adjusting the parameter C, our
algorithms effectively strike a balance between worst-case and average
performance, outperforming the benchmark algorithms.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 14:09:33 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Golrezaei",
"Negin",
""
],
[
"Jaillet",
"Patrick",
""
],
[
"Zhou",
"Zijie",
""
]
] | TITLE: Online Resource Allocation with Convex-set Machine-Learned Advice
ABSTRACT: Decision-makers often have access to a machine-learned prediction about
demand, referred to as advice, which can potentially be utilized in online
decision-making processes for resource allocation. However, exploiting such
advice poses challenges due to its potential inaccuracy. To address this issue,
we propose a framework that enhances online resource allocation decisions with
potentially unreliable machine-learned (ML) advice. We assume here that this
advice is represented by a general convex uncertainty set for the demand
vector.
We introduce a parameterized class of Pareto optimal online resource
allocation algorithms that strike a balance between consistent and robust
ratios. The consistent ratio measures the algorithm's performance (compared to
the optimal hindsight solution) when the ML advice is accurate, while the
robust ratio captures performance under an adversarial demand process when the
advice is inaccurate. Specifically, in a C-Pareto optimal setting, we maximize
the robust ratio while ensuring that the consistent ratio is at least C. Our
proposed C-Pareto optimal algorithm is an adaptive protection level algorithm,
which extends the classical fixed protection level algorithm introduced in
Littlewood (2005) and Ball and Queyranne (2009). Solving a complex non-convex
continuous optimization problem characterizes the adaptive protection level
algorithm. To complement our algorithms, we present a simple method for
computing the maximum achievable consistent ratio, which serves as an estimate
for the maximum value of the ML advice. Additionally, we present numerical
studies to evaluate the performance of our algorithm in comparison to benchmark
algorithms. The results demonstrate that by adjusting the parameter C, our
algorithms effectively strike a balance between worst-case and average
performance, outperforming the benchmark algorithms.
|
2306.12298 | Yuan-Gen Wang | Fengchuang Xing, Yuan-Gen Wang, Weixuan Tang, Guopu Zhu, Sam Kwong | StarVQA+: Co-training Space-Time Attention for Video Quality Assessment | null | null | null | null | cs.CV cs.LG eess.IV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Self-attention based Transformer has achieved great success in many computer
vision tasks. However, its application to video quality assessment (VQA) has
not been satisfactory so far. Evaluating the quality of in-the-wild videos is
challenging due to the unknown of pristine reference and shooting distortion.
This paper presents a co-trained Space-Time Attention network for the VQA
problem, termed StarVQA+. Specifically, we first build StarVQA+ by alternately
concatenating the divided space-time attention. Then, to facilitate the
training of StarVQA+, we design a vectorized regression loss by encoding the
mean opinion score (MOS) to the probability vector and embedding a special
token as the learnable variable of MOS, leading to better fitting of human's
rating process. Finally, to solve the data hungry problem with Transformer, we
propose to co-train the spatial and temporal attention weights using both
images and videos. Various experiments are conducted on the de-facto
in-the-wild video datasets, including LIVE-Qualcomm, LIVE-VQC, KoNViD-1k,
YouTube-UGC, LSVQ, LSVQ-1080p, and DVL2021. Experimental results demonstrate
the superiority of the proposed StarVQA+ over the state-of-the-art.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 14:27:31 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Xing",
"Fengchuang",
""
],
[
"Wang",
"Yuan-Gen",
""
],
[
"Tang",
"Weixuan",
""
],
[
"Zhu",
"Guopu",
""
],
[
"Kwong",
"Sam",
""
]
] | TITLE: StarVQA+: Co-training Space-Time Attention for Video Quality Assessment
ABSTRACT: Self-attention based Transformer has achieved great success in many computer
vision tasks. However, its application to video quality assessment (VQA) has
not been satisfactory so far. Evaluating the quality of in-the-wild videos is
challenging due to the unknown of pristine reference and shooting distortion.
This paper presents a co-trained Space-Time Attention network for the VQA
problem, termed StarVQA+. Specifically, we first build StarVQA+ by alternately
concatenating the divided space-time attention. Then, to facilitate the
training of StarVQA+, we design a vectorized regression loss by encoding the
mean opinion score (MOS) to the probability vector and embedding a special
token as the learnable variable of MOS, leading to better fitting of human's
rating process. Finally, to solve the data hungry problem with Transformer, we
propose to co-train the spatial and temporal attention weights using both
images and videos. Various experiments are conducted on the de-facto
in-the-wild video datasets, including LIVE-Qualcomm, LIVE-VQC, KoNViD-1k,
YouTube-UGC, LSVQ, LSVQ-1080p, and DVL2021. Experimental results demonstrate
the superiority of the proposed StarVQA+ over the state-of-the-art.
|
2306.12365 | Guoyao Shen | Guoyao Shen, Boran Hao, Mengyu Li, Chad W. Farris, Ioannis Ch.
Paschalidis, Stephan W. Anderson, Xin Zhang | Attention Hybrid Variational Net for Accelerated MRI Reconstruction | 22 pages, 4 figures, 3 tables | null | null | null | eess.IV cs.CV cs.LG physics.med-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The application of compressed sensing (CS)-enabled data reconstruction for
accelerating magnetic resonance imaging (MRI) remains a challenging problem.
This is due to the fact that the information lost in k-space from the
acceleration mask makes it difficult to reconstruct an image similar to the
quality of a fully sampled image. Multiple deep learning-based structures have
been proposed for MRI reconstruction using CS, both in the k-space and image
domains as well as using unrolled optimization methods. However, the drawback
of these structures is that they are not fully utilizing the information from
both domains (k-space and image). Herein, we propose a deep learning-based
attention hybrid variational network that performs learning in both the k-space
and image domain. We evaluate our method on a well-known open-source MRI
dataset and a clinical MRI dataset of patients diagnosed with strokes from our
institution to demonstrate the performance of our network. In addition to
quantitative evaluation, we undertook a blinded comparison of image quality
across networks performed by a subspecialty trained radiologist. Overall, we
demonstrate that our network achieves a superior performance among others under
multiple reconstruction tasks.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 16:19:07 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Shen",
"Guoyao",
""
],
[
"Hao",
"Boran",
""
],
[
"Li",
"Mengyu",
""
],
[
"Farris",
"Chad W.",
""
],
[
"Paschalidis",
"Ioannis Ch.",
""
],
[
"Anderson",
"Stephan W.",
""
],
[
"Zhang",
"Xin",
""
]
] | TITLE: Attention Hybrid Variational Net for Accelerated MRI Reconstruction
ABSTRACT: The application of compressed sensing (CS)-enabled data reconstruction for
accelerating magnetic resonance imaging (MRI) remains a challenging problem.
This is due to the fact that the information lost in k-space from the
acceleration mask makes it difficult to reconstruct an image similar to the
quality of a fully sampled image. Multiple deep learning-based structures have
been proposed for MRI reconstruction using CS, both in the k-space and image
domains as well as using unrolled optimization methods. However, the drawback
of these structures is that they are not fully utilizing the information from
both domains (k-space and image). Herein, we propose a deep learning-based
attention hybrid variational network that performs learning in both the k-space
and image domain. We evaluate our method on a well-known open-source MRI
dataset and a clinical MRI dataset of patients diagnosed with strokes from our
institution to demonstrate the performance of our network. In addition to
quantitative evaluation, we undertook a blinded comparison of image quality
across networks performed by a subspecialty trained radiologist. Overall, we
demonstrate that our network achieves a superior performance among others under
multiple reconstruction tasks.
|
2306.12376 | Bidur Khanal | Bidur Khanal, Binod Bhattarai, Bishesh Khanal, Danail Stoyanov,
Cristian A. Linte | M-VAAL: Multimodal Variational Adversarial Active Learning for
Downstream Medical Image Analysis Tasks | null | null | null | null | eess.IV cs.CV | http://creativecommons.org/licenses/by/4.0/ | Acquiring properly annotated data is expensive in the medical field as it
requires experts, time-consuming protocols, and rigorous validation. Active
learning attempts to minimize the need for large annotated samples by actively
sampling the most informative examples for annotation. These examples
contribute significantly to improving the performance of supervised machine
learning models, and thus, active learning can play an essential role in
selecting the most appropriate information in deep learning-based diagnosis,
clinical assessments, and treatment planning. Although some existing works have
proposed methods for sampling the best examples for annotation in medical image
analysis, they are not task-agnostic and do not use multimodal auxiliary
information in the sampler, which has the potential to increase robustness.
Therefore, in this work, we propose a Multimodal Variational Adversarial Active
Learning (M-VAAL) method that uses auxiliary information from additional
modalities to enhance the active sampling. We applied our method to two
datasets: i) brain tumor segmentation and multi-label classification using the
BraTS2018 dataset, and ii) chest X-ray image classification using the
COVID-QU-Ex dataset. Our results show a promising direction toward
data-efficient learning under limited annotations.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 16:40:37 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Khanal",
"Bidur",
""
],
[
"Bhattarai",
"Binod",
""
],
[
"Khanal",
"Bishesh",
""
],
[
"Stoyanov",
"Danail",
""
],
[
"Linte",
"Cristian A.",
""
]
] | TITLE: M-VAAL: Multimodal Variational Adversarial Active Learning for
Downstream Medical Image Analysis Tasks
ABSTRACT: Acquiring properly annotated data is expensive in the medical field as it
requires experts, time-consuming protocols, and rigorous validation. Active
learning attempts to minimize the need for large annotated samples by actively
sampling the most informative examples for annotation. These examples
contribute significantly to improving the performance of supervised machine
learning models, and thus, active learning can play an essential role in
selecting the most appropriate information in deep learning-based diagnosis,
clinical assessments, and treatment planning. Although some existing works have
proposed methods for sampling the best examples for annotation in medical image
analysis, they are not task-agnostic and do not use multimodal auxiliary
information in the sampler, which has the potential to increase robustness.
Therefore, in this work, we propose a Multimodal Variational Adversarial Active
Learning (M-VAAL) method that uses auxiliary information from additional
modalities to enhance the active sampling. We applied our method to two
datasets: i) brain tumor segmentation and multi-label classification using the
BraTS2018 dataset, and ii) chest X-ray image classification using the
COVID-QU-Ex dataset. Our results show a promising direction toward
data-efficient learning under limited annotations.
|
2306.12380 | Samir M. Perlaza | Samir M. Perlaza, I\~naki Esnaola, Gaetan Bisson, H. Vincent Poor | On the Validation of Gibbs Algorithms: Training Datasets, Test Datasets
and their Aggregation | In Proc. IEEE International Symposium on Information Theory (ISIT),
Taipei, Taiwan, Jun., 2023. arXiv admin note: text overlap with
arXiv:2211.06617 | null | null | null | cs.LG cs.IT math.IT math.PR math.ST stat.TH | http://creativecommons.org/licenses/by/4.0/ | The dependence on training data of the Gibbs algorithm (GA) is analytically
characterized. By adopting the expected empirical risk as the performance
metric, the sensitivity of the GA is obtained in closed form. In this case,
sensitivity is the performance difference with respect to an arbitrary
alternative algorithm. This description enables the development of explicit
expressions involving the training errors and test errors of GAs trained with
different datasets. Using these tools, dataset aggregation is studied and
different figures of merit to evaluate the generalization capabilities of GAs
are introduced. For particular sizes of such datasets and parameters of the
GAs, a connection between Jeffrey's divergence, training and test errors is
established.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 16:51:50 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Perlaza",
"Samir M.",
""
],
[
"Esnaola",
"Iñaki",
""
],
[
"Bisson",
"Gaetan",
""
],
[
"Poor",
"H. Vincent",
""
]
] | TITLE: On the Validation of Gibbs Algorithms: Training Datasets, Test Datasets
and their Aggregation
ABSTRACT: The dependence on training data of the Gibbs algorithm (GA) is analytically
characterized. By adopting the expected empirical risk as the performance
metric, the sensitivity of the GA is obtained in closed form. In this case,
sensitivity is the performance difference with respect to an arbitrary
alternative algorithm. This description enables the development of explicit
expressions involving the training errors and test errors of GAs trained with
different datasets. Using these tools, dataset aggregation is studied and
different figures of merit to evaluate the generalization capabilities of GAs
are introduced. For particular sizes of such datasets and parameters of the
GAs, a connection between Jeffrey's divergence, training and test errors is
established.
|
2306.12384 | Chaopeng Shen | Jiangtao Liu, Yuchen Bian and Chaopeng Shen | Probing the limit of hydrologic predictability with the Transformer
network | null | null | null | null | cs.LG cs.AI physics.geo-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | For a number of years since its introduction to hydrology, recurrent neural
networks like long short-term memory (LSTM) have proven remarkably difficult to
surpass in terms of daily hydrograph metrics on known, comparable benchmarks.
Outside of hydrology, Transformers have now become the model of choice for
sequential prediction tasks, making it a curious architecture to investigate.
Here, we first show that a vanilla Transformer architecture is not competitive
against LSTM on the widely benchmarked CAMELS dataset, and lagged especially
for the high-flow metrics due to short-term processes. However, a
recurrence-free variant of Transformer can obtain mixed comparisons with LSTM,
producing the same Kling-Gupta efficiency coefficient (KGE), along with other
metrics. The lack of advantages for the Transformer is linked to the Markovian
nature of the hydrologic prediction problem. Similar to LSTM, the Transformer
can also merge multiple forcing dataset to improve model performance. While the
Transformer results are not higher than current state-of-the-art, we still
learned some valuable lessons: (1) the vanilla Transformer architecture is not
suitable for hydrologic modeling; (2) the proposed recurrence-free modification
can improve Transformer performance so future work can continue to test more of
such modifications; and (3) the prediction limits on the dataset should be
close to the current state-of-the-art model. As a non-recurrent model, the
Transformer may bear scale advantages for learning from bigger datasets and
storing knowledge. This work serves as a reference point for future
modifications of the model.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 17:06:54 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Liu",
"Jiangtao",
""
],
[
"Bian",
"Yuchen",
""
],
[
"Shen",
"Chaopeng",
""
]
] | TITLE: Probing the limit of hydrologic predictability with the Transformer
network
ABSTRACT: For a number of years since its introduction to hydrology, recurrent neural
networks like long short-term memory (LSTM) have proven remarkably difficult to
surpass in terms of daily hydrograph metrics on known, comparable benchmarks.
Outside of hydrology, Transformers have now become the model of choice for
sequential prediction tasks, making it a curious architecture to investigate.
Here, we first show that a vanilla Transformer architecture is not competitive
against LSTM on the widely benchmarked CAMELS dataset, and lagged especially
for the high-flow metrics due to short-term processes. However, a
recurrence-free variant of Transformer can obtain mixed comparisons with LSTM,
producing the same Kling-Gupta efficiency coefficient (KGE), along with other
metrics. The lack of advantages for the Transformer is linked to the Markovian
nature of the hydrologic prediction problem. Similar to LSTM, the Transformer
can also merge multiple forcing dataset to improve model performance. While the
Transformer results are not higher than current state-of-the-art, we still
learned some valuable lessons: (1) the vanilla Transformer architecture is not
suitable for hydrologic modeling; (2) the proposed recurrence-free modification
can improve Transformer performance so future work can continue to test more of
such modifications; and (3) the prediction limits on the dataset should be
close to the current state-of-the-art model. As a non-recurrent model, the
Transformer may bear scale advantages for learning from bigger datasets and
storing knowledge. This work serves as a reference point for future
modifications of the model.
|
2306.12387 | Weijie Zhang | Weijie Jack Zhang | Solving Dialogue Grounding Embodied Task in a Simulated Environment
using Further Masked Language Modeling | Work in Progress | null | null | null | cs.CL cs.HC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Enhancing AI systems with efficient communication skills that align with
human understanding is crucial for their effective assistance to human users.
Proactive initiatives from the system side are needed to discern specific
circumstances and interact aptly with users to solve these scenarios. In this
research, we opt for a collective building assignment taken from the Minecraft
dataset. Our proposed method employs language modeling to enhance task
understanding through state-of-the-art (SOTA) methods using language models.
These models focus on grounding multi-modal understandinging and task-oriented
dialogue comprehension tasks. This focus aids in gaining insights into how well
these models interpret and respond to a variety of inputs and tasks. Our
experimental results provide compelling evidence of the superiority of our
proposed method. This showcases a substantial improvement and points towards a
promising direction for future research in this domain.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 17:17:09 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Zhang",
"Weijie Jack",
""
]
] | TITLE: Solving Dialogue Grounding Embodied Task in a Simulated Environment
using Further Masked Language Modeling
ABSTRACT: Enhancing AI systems with efficient communication skills that align with
human understanding is crucial for their effective assistance to human users.
Proactive initiatives from the system side are needed to discern specific
circumstances and interact aptly with users to solve these scenarios. In this
research, we opt for a collective building assignment taken from the Minecraft
dataset. Our proposed method employs language modeling to enhance task
understanding through state-of-the-art (SOTA) methods using language models.
These models focus on grounding multi-modal understandinging and task-oriented
dialogue comprehension tasks. This focus aids in gaining insights into how well
these models interpret and respond to a variety of inputs and tasks. Our
experimental results provide compelling evidence of the superiority of our
proposed method. This showcases a substantial improvement and points towards a
promising direction for future research in this domain.
|
2306.12398 | Walter Zimmer | Aral Hekimoglu, Philipp Friedrich, Walter Zimmer, Michael Schmidt,
Alvaro Marcos-Ramiro, Alois C. Knoll | Multi-Task Consistency for Active Learning | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-sa/4.0/ | Learning-based solutions for vision tasks require a large amount of labeled
training data to ensure their performance and reliability. In single-task
vision-based settings, inconsistency-based active learning has proven to be
effective in selecting informative samples for annotation. However, there is a
lack of research exploiting the inconsistency between multiple tasks in
multi-task networks. To address this gap, we propose a novel multi-task active
learning strategy for two coupled vision tasks: object detection and semantic
segmentation. Our approach leverages the inconsistency between them to identify
informative samples across both tasks. We propose three constraints that
specify how the tasks are coupled and introduce a method for determining the
pixels belonging to the object detected by a bounding box, to later quantify
the constraints as inconsistency scores. To evaluate the effectiveness of our
approach, we establish multiple baselines for multi-task active learning and
introduce a new metric, mean Detection Segmentation Quality (mDSQ), tailored
for the multi-task active learning comparison that addresses the performance of
both tasks. We conduct extensive experiments on the nuImages and A9 datasets,
demonstrating that our approach outperforms existing state-of-the-art methods
by up to 3.4% mDSQ on nuImages. Our approach achieves 95% of the fully-trained
performance using only 67% of the available data, corresponding to 20% fewer
labels compared to random selection and 5% fewer labels compared to
state-of-the-art selection strategy. Our code will be made publicly available
after the review process.
| [
{
"version": "v1",
"created": "Wed, 21 Jun 2023 17:34:31 GMT"
}
] | 2023-06-22T00:00:00 | [
[
"Hekimoglu",
"Aral",
""
],
[
"Friedrich",
"Philipp",
""
],
[
"Zimmer",
"Walter",
""
],
[
"Schmidt",
"Michael",
""
],
[
"Marcos-Ramiro",
"Alvaro",
""
],
[
"Knoll",
"Alois C.",
""
]
] | TITLE: Multi-Task Consistency for Active Learning
ABSTRACT: Learning-based solutions for vision tasks require a large amount of labeled
training data to ensure their performance and reliability. In single-task
vision-based settings, inconsistency-based active learning has proven to be
effective in selecting informative samples for annotation. However, there is a
lack of research exploiting the inconsistency between multiple tasks in
multi-task networks. To address this gap, we propose a novel multi-task active
learning strategy for two coupled vision tasks: object detection and semantic
segmentation. Our approach leverages the inconsistency between them to identify
informative samples across both tasks. We propose three constraints that
specify how the tasks are coupled and introduce a method for determining the
pixels belonging to the object detected by a bounding box, to later quantify
the constraints as inconsistency scores. To evaluate the effectiveness of our
approach, we establish multiple baselines for multi-task active learning and
introduce a new metric, mean Detection Segmentation Quality (mDSQ), tailored
for the multi-task active learning comparison that addresses the performance of
both tasks. We conduct extensive experiments on the nuImages and A9 datasets,
demonstrating that our approach outperforms existing state-of-the-art methods
by up to 3.4% mDSQ on nuImages. Our approach achieves 95% of the fully-trained
performance using only 67% of the available data, corresponding to 20% fewer
labels compared to random selection and 5% fewer labels compared to
state-of-the-art selection strategy. Our code will be made publicly available
after the review process.
|
1806.00910 | Abeed Sarker | Abeed Sarker and Graciela Gonzalez-Hernandez | An unsupervised and customizable misspelling generator for mining noisy
health-related text sources | null | J Biomed Inform. 2018 Dec;88:98-107. Epub 2018 Nov 13 | 10.1016/j.jbi.2018.11.007 | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | In this paper, we present a customizable datacentric system that
automatically generates common misspellings for complex health-related terms.
The spelling variant generator relies on a dense vector model learned from
large unlabeled text, which is used to find semantically close terms to the
original/seed keyword, followed by the filtering of terms that are lexically
dissimilar beyond a given threshold. The process is executed recursively,
converging when no new terms similar (lexically and semantically) to the seed
keyword are found. Weighting of intra-word character sequence similarities
allows further problem-specific customization of the system. On a dataset
prepared for this study, our system outperforms the current state-of-the-art
for medication name variant generation with best F1-score of 0.69 and
F1/4-score of 0.78. Extrinsic evaluation of the system on a set of
cancer-related terms showed an increase of over 67% in retrieval rate from
Twitter posts when the generated variants are included. Our proposed spelling
variant generator has several advantages over the current state-of-the-art and
other types of variant generators-(i) it is capable of filtering out lexically
similar but semantically dissimilar terms, (ii) the number of variants
generated is low as many low-frequency and ambiguous misspellings are filtered
out, and (iii) the system is fully automatic, customizable and easily
executable. While the base system is fully unsupervised, we show how
supervision maybe employed to adjust weights for task-specific customization.
The performance and significant relative simplicity of our proposed approach
makes it a much needed misspelling generation resource for health-related text
mining from noisy sources. The source code for the system has been made
publicly available for research purposes.
| [
{
"version": "v1",
"created": "Mon, 4 Jun 2018 01:07:37 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Sarker",
"Abeed",
""
],
[
"Gonzalez-Hernandez",
"Graciela",
""
]
] | TITLE: An unsupervised and customizable misspelling generator for mining noisy
health-related text sources
ABSTRACT: In this paper, we present a customizable datacentric system that
automatically generates common misspellings for complex health-related terms.
The spelling variant generator relies on a dense vector model learned from
large unlabeled text, which is used to find semantically close terms to the
original/seed keyword, followed by the filtering of terms that are lexically
dissimilar beyond a given threshold. The process is executed recursively,
converging when no new terms similar (lexically and semantically) to the seed
keyword are found. Weighting of intra-word character sequence similarities
allows further problem-specific customization of the system. On a dataset
prepared for this study, our system outperforms the current state-of-the-art
for medication name variant generation with best F1-score of 0.69 and
F1/4-score of 0.78. Extrinsic evaluation of the system on a set of
cancer-related terms showed an increase of over 67% in retrieval rate from
Twitter posts when the generated variants are included. Our proposed spelling
variant generator has several advantages over the current state-of-the-art and
other types of variant generators-(i) it is capable of filtering out lexically
similar but semantically dissimilar terms, (ii) the number of variants
generated is low as many low-frequency and ambiguous misspellings are filtered
out, and (iii) the system is fully automatic, customizable and easily
executable. While the base system is fully unsupervised, we show how
supervision maybe employed to adjust weights for task-specific customization.
The performance and significant relative simplicity of our proposed approach
makes it a much needed misspelling generation resource for health-related text
mining from noisy sources. The source code for the system has been made
publicly available for research purposes.
|
1904.03800 | Shuhao Zhang | Shuhao Zhang, Yingjun Wu, Feng Zhang, and Bingsheng He | Towards Concurrent Stateful Stream Processing on Multicore Processors
(Technical Report) | null | null | 10.1109/ICDE48307.2020.00136 | null | cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent data stream processing systems (DSPSs) can achieve excellent
performance when processing large volumes of data under tight latency
constraints. However, they sacrifice support for concurrent state access that
eases the burden of developing stateful stream applications. Recently, some
have proposed managing concurrent state access during stream processing by
modeling state accesses as transactions. However, these are realized with locks
involving serious contention overhead. Their coarse-grained processing paradigm
further magnifies contention issues and tends to poorly utilize modern
multicore architectures. This paper introduces TStream , a novel DSPS
supporting efficient concurrent state access on multicore processors.
Transactional semantics is employed like previous work, but scalability is
greatly improved due to two novel designs: 1) dual-mode scheduling, which
exposes more parallelism opportunities, 2) dynamic restructuring execution,
which aggressively exploits the parallelism opportunities from dual-mode
scheduling without centralized lock contentions. To validate our proposal, we
evaluate TStream with a benchmark of four applications on a modern multicore
machine. The experimental results show that 1) TStream achieves up to 4.8 times
higher throughput with similar processing latency compared to the
state-of-the-art and 2) unlike prior solutions, TStream is highly tolerant of
varying application workloads such as key skewness and multi-partition state
accesses.
| [
{
"version": "v1",
"created": "Mon, 8 Apr 2019 02:10:03 GMT"
},
{
"version": "v2",
"created": "Wed, 10 Apr 2019 04:00:55 GMT"
},
{
"version": "v3",
"created": "Mon, 2 Sep 2019 02:40:31 GMT"
},
{
"version": "v4",
"created": "Thu, 16 Jan 2020 06:17:26 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Zhang",
"Shuhao",
""
],
[
"Wu",
"Yingjun",
""
],
[
"Zhang",
"Feng",
""
],
[
"He",
"Bingsheng",
""
]
] | TITLE: Towards Concurrent Stateful Stream Processing on Multicore Processors
(Technical Report)
ABSTRACT: Recent data stream processing systems (DSPSs) can achieve excellent
performance when processing large volumes of data under tight latency
constraints. However, they sacrifice support for concurrent state access that
eases the burden of developing stateful stream applications. Recently, some
have proposed managing concurrent state access during stream processing by
modeling state accesses as transactions. However, these are realized with locks
involving serious contention overhead. Their coarse-grained processing paradigm
further magnifies contention issues and tends to poorly utilize modern
multicore architectures. This paper introduces TStream , a novel DSPS
supporting efficient concurrent state access on multicore processors.
Transactional semantics is employed like previous work, but scalability is
greatly improved due to two novel designs: 1) dual-mode scheduling, which
exposes more parallelism opportunities, 2) dynamic restructuring execution,
which aggressively exploits the parallelism opportunities from dual-mode
scheduling without centralized lock contentions. To validate our proposal, we
evaluate TStream with a benchmark of four applications on a modern multicore
machine. The experimental results show that 1) TStream achieves up to 4.8 times
higher throughput with similar processing latency compared to the
state-of-the-art and 2) unlike prior solutions, TStream is highly tolerant of
varying application workloads such as key skewness and multi-partition state
accesses.
|
2004.04243 | Stefan Constantin | Stefan Constantin and Alex Waibel | Error correction and extraction in request dialogs | 10 pages, 8 figures, 3 tables, presented at ICNLSP 2022 | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | We propose a dialog system utility component that gets the last two
utterances of a user and can detect whether the last utterance is an error
correction of the second last utterance. If yes, it corrects the second last
utterance according to the error correction in the last utterance and outputs
the extracted pairs of reparandum and repair entity. This component offers two
advantages, learning the concept of corrections to avoid collecting corrections
for every new domain and extracting reparandum and repair pairs, which offers
the possibility to learn out of it.
For the error correction one sequence labeling and two sequence to sequence
approaches are presented. For the error correction detection these three error
correction approaches can also be used and in addition, we present a sequence
classification approach. One error correction detection and one error
correction approach can be combined to a pipeline or the error correction
approaches can be trained and used end-to-end to avoid two components. We
modified the EPIC-KITCHENS-100 dataset to evaluate the approaches for
correcting entity phrases in request dialogs. For error correction detection
and correction, we got an accuracy of 96.40 % on synthetic validation data and
an accuracy of 77.81 % on human-created real-world test data.
| [
{
"version": "v1",
"created": "Wed, 8 Apr 2020 20:49:10 GMT"
},
{
"version": "v2",
"created": "Mon, 21 Nov 2022 16:26:36 GMT"
},
{
"version": "v3",
"created": "Mon, 1 May 2023 15:42:40 GMT"
},
{
"version": "v4",
"created": "Tue, 20 Jun 2023 17:58:11 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Constantin",
"Stefan",
""
],
[
"Waibel",
"Alex",
""
]
] | TITLE: Error correction and extraction in request dialogs
ABSTRACT: We propose a dialog system utility component that gets the last two
utterances of a user and can detect whether the last utterance is an error
correction of the second last utterance. If yes, it corrects the second last
utterance according to the error correction in the last utterance and outputs
the extracted pairs of reparandum and repair entity. This component offers two
advantages, learning the concept of corrections to avoid collecting corrections
for every new domain and extracting reparandum and repair pairs, which offers
the possibility to learn out of it.
For the error correction one sequence labeling and two sequence to sequence
approaches are presented. For the error correction detection these three error
correction approaches can also be used and in addition, we present a sequence
classification approach. One error correction detection and one error
correction approach can be combined to a pipeline or the error correction
approaches can be trained and used end-to-end to avoid two components. We
modified the EPIC-KITCHENS-100 dataset to evaluate the approaches for
correcting entity phrases in request dialogs. For error correction detection
and correction, we got an accuracy of 96.40 % on synthetic validation data and
an accuracy of 77.81 % on human-created real-world test data.
|
2010.03902 | Mahesh Pal Dr. | Mahesh Pal, Akshay, B. Charan Teja | IRX-1D: A Simple Deep Learning Architecture for Remote Sensing
Classifications | Want to improve this manuscript as it is not accepted by journal in
present form | null | null | null | cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We proposes a simple deep learning architecture combining elements of
Inception, ResNet and Xception networks. Four new datasets were used for
classification with both small and large training samples. Results in terms of
classification accuracy suggests improved performance by proposed architecture
in comparison to Bayesian optimised 2D-CNN with small training samples.
Comparison of results using small training sample with Indiana Pines
hyperspectral dataset suggests comparable or better performance by proposed
architecture than nine reported works using different deep learning
architectures. In spite of achieving high classification accuracy with limited
training samples, comparison of classified image suggests different land cover
classes are assigned to same area when compared with the classified image
provided by the model trained using large training samples with all datasets.
| [
{
"version": "v1",
"created": "Thu, 8 Oct 2020 11:07:02 GMT"
},
{
"version": "v2",
"created": "Mon, 19 Jun 2023 05:51:05 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Pal",
"Mahesh",
""
],
[
"Akshay",
"",
""
],
[
"Teja",
"B. Charan",
""
]
] | TITLE: IRX-1D: A Simple Deep Learning Architecture for Remote Sensing
Classifications
ABSTRACT: We proposes a simple deep learning architecture combining elements of
Inception, ResNet and Xception networks. Four new datasets were used for
classification with both small and large training samples. Results in terms of
classification accuracy suggests improved performance by proposed architecture
in comparison to Bayesian optimised 2D-CNN with small training samples.
Comparison of results using small training sample with Indiana Pines
hyperspectral dataset suggests comparable or better performance by proposed
architecture than nine reported works using different deep learning
architectures. In spite of achieving high classification accuracy with limited
training samples, comparison of classified image suggests different land cover
classes are assigned to same area when compared with the classified image
provided by the model trained using large training samples with all datasets.
|
2012.03385 | Daniel Seita | Daniel Seita, Pete Florence, Jonathan Tompson, Erwin Coumans, Vikas
Sindhwani, Ken Goldberg, Andy Zeng | Learning to Rearrange Deformable Cables, Fabrics, and Bags with
Goal-Conditioned Transporter Networks | See https://berkeleyautomation.github.io/bags/ for project website
and code; v3 is ICRA 2021 version and v4 adds physical experiments and
improves simulation results | null | null | null | cs.RO cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Rearranging and manipulating deformable objects such as cables, fabrics, and
bags is a long-standing challenge in robotic manipulation. The complex dynamics
and high-dimensional configuration spaces of deformables, compared to rigid
objects, make manipulation difficult not only for multi-step planning, but even
for goal specification. Goals cannot be as easily specified as rigid object
poses, and may involve complex relative spatial relations such as "place the
item inside the bag". In this work, we develop a suite of simulated benchmarks
with 1D, 2D, and 3D deformable structures, including tasks that involve
image-based goal-conditioning and multi-step deformable manipulation. We
propose embedding goal-conditioning into Transporter Networks, a recently
proposed model architecture for learning robotic manipulation that rearranges
deep features to infer displacements that can represent pick and place actions.
In simulation and in physical experiments, we demonstrate that goal-conditioned
Transporter Networks enable agents to manipulate deformable structures into
flexibly specified configurations without test-time visual anchors for target
locations. We also significantly extend prior results using Transporter
Networks for manipulating deformable objects by testing on tasks with 2D and 3D
deformables. Supplementary material is available at
https://berkeleyautomation.github.io/bags/.
| [
{
"version": "v1",
"created": "Sun, 6 Dec 2020 22:21:54 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Dec 2020 23:00:29 GMT"
},
{
"version": "v3",
"created": "Fri, 26 Mar 2021 14:40:25 GMT"
},
{
"version": "v4",
"created": "Sun, 18 Jun 2023 18:54:13 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Seita",
"Daniel",
""
],
[
"Florence",
"Pete",
""
],
[
"Tompson",
"Jonathan",
""
],
[
"Coumans",
"Erwin",
""
],
[
"Sindhwani",
"Vikas",
""
],
[
"Goldberg",
"Ken",
""
],
[
"Zeng",
"Andy",
""
]
] | TITLE: Learning to Rearrange Deformable Cables, Fabrics, and Bags with
Goal-Conditioned Transporter Networks
ABSTRACT: Rearranging and manipulating deformable objects such as cables, fabrics, and
bags is a long-standing challenge in robotic manipulation. The complex dynamics
and high-dimensional configuration spaces of deformables, compared to rigid
objects, make manipulation difficult not only for multi-step planning, but even
for goal specification. Goals cannot be as easily specified as rigid object
poses, and may involve complex relative spatial relations such as "place the
item inside the bag". In this work, we develop a suite of simulated benchmarks
with 1D, 2D, and 3D deformable structures, including tasks that involve
image-based goal-conditioning and multi-step deformable manipulation. We
propose embedding goal-conditioning into Transporter Networks, a recently
proposed model architecture for learning robotic manipulation that rearranges
deep features to infer displacements that can represent pick and place actions.
In simulation and in physical experiments, we demonstrate that goal-conditioned
Transporter Networks enable agents to manipulate deformable structures into
flexibly specified configurations without test-time visual anchors for target
locations. We also significantly extend prior results using Transporter
Networks for manipulating deformable objects by testing on tasks with 2D and 3D
deformables. Supplementary material is available at
https://berkeleyautomation.github.io/bags/.
|
2012.08835 | Sergio Maffeis | Rishi Rabheru, Hazim Hanif, Sergio Maffeis | A Hybrid Graph Neural Network Approach for Detecting PHP Vulnerabilities | A poster version of this paper appeared as
https://doi.org/10.1145/3412841.3442132 | IEEE Conference on Dependable and Secure Computing (DSC), 2022 | 10.1109/DSC54232.2022.9888816 | null | cs.CR cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents DeepTective, a deep learning approach to detect
vulnerabilities in PHP source code. Our approach implements a novel hybrid
technique that combines Gated Recurrent Units and Graph Convolutional Networks
to detect SQLi, XSS and OSCI vulnerabilities leveraging both syntactic and
semantic information. We evaluate DeepTective and compare it to the state of
the art on an established synthetic dataset and on a novel real-world dataset
collected from GitHub. Experimental results show that DeepTective achieves near
perfect classification on the synthetic dataset, and an F1 score of 88.12% on
the realistic dataset, outperforming related approaches. We validate
DeepTective in the wild by discovering 4 novel vulnerabilities in established
WordPress plugins.
| [
{
"version": "v1",
"created": "Wed, 16 Dec 2020 10:17:53 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Rabheru",
"Rishi",
""
],
[
"Hanif",
"Hazim",
""
],
[
"Maffeis",
"Sergio",
""
]
] | TITLE: A Hybrid Graph Neural Network Approach for Detecting PHP Vulnerabilities
ABSTRACT: This paper presents DeepTective, a deep learning approach to detect
vulnerabilities in PHP source code. Our approach implements a novel hybrid
technique that combines Gated Recurrent Units and Graph Convolutional Networks
to detect SQLi, XSS and OSCI vulnerabilities leveraging both syntactic and
semantic information. We evaluate DeepTective and compare it to the state of
the art on an established synthetic dataset and on a novel real-world dataset
collected from GitHub. Experimental results show that DeepTective achieves near
perfect classification on the synthetic dataset, and an F1 score of 88.12% on
the realistic dataset, outperforming related approaches. We validate
DeepTective in the wild by discovering 4 novel vulnerabilities in established
WordPress plugins.
|
2101.02830 | Stephen MacDonell | Osayande P. Omondiagbe, Sherlock A. Licorish and Stephen G. MacDonell | Features that Predict the Acceptability of Java and JavaScript Answers
on Stack Overflow | Conference, 11 pages, 3 figures, 2 tables | Proceedings of the 23rd International Conference on Evaluation and
Assessment in Software Engineering (EASE2019). Copenhagen, Denmark, ACM
Press, pp.101-110 | 10.1145/3319008.3319024 | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Context: Stack Overflow is a popular community question and answer portal
used by practitioners to solve problems during software development. Developers
can focus their attention on answers that have been accepted or where members
have recorded high votes in judging good answers when searching for help.
However, the latter mechanism (votes) can be unreliable, and there is currently
no way to differentiate between an answer that is likely to be accepted and
those that will not be accepted by looking at the answer's characteristics.
Objective: In potentially providing a mechanism to identify acceptable answers,
this study examines the features that distinguish an accepted answer from an
unaccepted answer. Methods: We studied the Stack Overflow dataset by analyzing
questions and answers for the two most popular tags (Java and JavaScript). Our
dataset comprised 249,588 posts drawn from 2014-2016. We use random forest and
neural network models to predict accepted answers, and study the features with
the highest predictive power in those two models. Results: Our findings reveal
that the length of code in answers, reputation of users, similarity of the text
between questions and answers, and the time lag between questions and answers
have the highest predictive power for differentiating accepted and unaccepted
answers. Conclusion: Tools may leverage these findings in supporting developers
and reducing the effort they must dedicate to searching for suitable answers on
Stack Overflow.
| [
{
"version": "v1",
"created": "Fri, 8 Jan 2021 03:09:38 GMT"
},
{
"version": "v2",
"created": "Mon, 19 Jun 2023 09:18:04 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Omondiagbe",
"Osayande P.",
""
],
[
"Licorish",
"Sherlock A.",
""
],
[
"MacDonell",
"Stephen G.",
""
]
] | TITLE: Features that Predict the Acceptability of Java and JavaScript Answers
on Stack Overflow
ABSTRACT: Context: Stack Overflow is a popular community question and answer portal
used by practitioners to solve problems during software development. Developers
can focus their attention on answers that have been accepted or where members
have recorded high votes in judging good answers when searching for help.
However, the latter mechanism (votes) can be unreliable, and there is currently
no way to differentiate between an answer that is likely to be accepted and
those that will not be accepted by looking at the answer's characteristics.
Objective: In potentially providing a mechanism to identify acceptable answers,
this study examines the features that distinguish an accepted answer from an
unaccepted answer. Methods: We studied the Stack Overflow dataset by analyzing
questions and answers for the two most popular tags (Java and JavaScript). Our
dataset comprised 249,588 posts drawn from 2014-2016. We use random forest and
neural network models to predict accepted answers, and study the features with
the highest predictive power in those two models. Results: Our findings reveal
that the length of code in answers, reputation of users, similarity of the text
between questions and answers, and the time lag between questions and answers
have the highest predictive power for differentiating accepted and unaccepted
answers. Conclusion: Tools may leverage these findings in supporting developers
and reducing the effort they must dedicate to searching for suitable answers on
Stack Overflow.
|
2105.12769 | Alexander Jung | Yasmin SarcheshmehPour, Yu Tian, Linli Zhang, Alexander Jung | Clustered Federated Learning via Generalized Total Variation
Minimization | null | null | null | null | cs.LG cs.DC stat.ML | http://creativecommons.org/licenses/by/4.0/ | We study optimization methods to train local (or personalized) models for
decentralized collections of local datasets with an intrinsic network
structure. This network structure arises from domain-specific notions of
similarity between local datasets. Examples for such notions include
spatio-temporal proximity, statistical dependencies or functional relations.
Our main conceptual contribution is to formulate federated learning as
generalized total variation (GTV) minimization. This formulation unifies and
considerably extends existing federated learning methods. It is highly flexible
and can be combined with a broad range of parametric models, including
generalized linear models or deep neural networks. Our main algorithmic
contribution is a fully decentralized federated learning algorithm. This
algorithm is obtained by applying an established primal-dual method to solve
GTV minimization. It can be implemented as message passing and is robust
against inexact computations that arise from limited computational resources
including processing time or bandwidth. Our main analytic contribution is an
upper bound on the deviation between the local model parameters learnt by our
algorithm and an oracle-based clustered federated learning method. This upper
bound reveals conditions on the local models and the network structure of local
datasets such that GTV minimization is able to pool (nearly) homogeneous local
datasets.
| [
{
"version": "v1",
"created": "Wed, 26 May 2021 18:07:19 GMT"
},
{
"version": "v2",
"created": "Mon, 7 Nov 2022 15:21:01 GMT"
},
{
"version": "v3",
"created": "Fri, 23 Dec 2022 17:07:13 GMT"
},
{
"version": "v4",
"created": "Sun, 18 Jun 2023 17:14:37 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"SarcheshmehPour",
"Yasmin",
""
],
[
"Tian",
"Yu",
""
],
[
"Zhang",
"Linli",
""
],
[
"Jung",
"Alexander",
""
]
] | TITLE: Clustered Federated Learning via Generalized Total Variation
Minimization
ABSTRACT: We study optimization methods to train local (or personalized) models for
decentralized collections of local datasets with an intrinsic network
structure. This network structure arises from domain-specific notions of
similarity between local datasets. Examples for such notions include
spatio-temporal proximity, statistical dependencies or functional relations.
Our main conceptual contribution is to formulate federated learning as
generalized total variation (GTV) minimization. This formulation unifies and
considerably extends existing federated learning methods. It is highly flexible
and can be combined with a broad range of parametric models, including
generalized linear models or deep neural networks. Our main algorithmic
contribution is a fully decentralized federated learning algorithm. This
algorithm is obtained by applying an established primal-dual method to solve
GTV minimization. It can be implemented as message passing and is robust
against inexact computations that arise from limited computational resources
including processing time or bandwidth. Our main analytic contribution is an
upper bound on the deviation between the local model parameters learnt by our
algorithm and an oracle-based clustered federated learning method. This upper
bound reveals conditions on the local models and the network structure of local
datasets such that GTV minimization is able to pool (nearly) homogeneous local
datasets.
|
2110.06848 | Chun-Hsiao Yeh | Chun-Hsiao Yeh, Cheng-Yao Hong, Yen-Chi Hsu, Tyng-Luh Liu, Yubei Chen
and Yann LeCun | Decoupled Contrastive Learning | Accepted by ECCV2022 | null | null | null | cs.LG cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Contrastive learning (CL) is one of the most successful paradigms for
self-supervised learning (SSL). In a principled way, it considers two augmented
"views" of the same image as positive to be pulled closer, and all other images
as negative to be pushed further apart. However, behind the impressive success
of CL-based techniques, their formulation often relies on heavy-computation
settings, including large sample batches, extensive training epochs, etc. We
are thus motivated to tackle these issues and establish a simple, efficient,
yet competitive baseline of contrastive learning. Specifically, we identify,
from theoretical and empirical studies, a noticeable negative-positive-coupling
(NPC) effect in the widely used InfoNCE loss, leading to unsuitable learning
efficiency concerning the batch size. By removing the NPC effect, we propose
decoupled contrastive learning (DCL) loss, which removes the positive term from
the denominator and significantly improves the learning efficiency. DCL
achieves competitive performance with less sensitivity to sub-optimal
hyperparameters, requiring neither large batches in SimCLR, momentum encoding
in MoCo, or large epochs. We demonstrate with various benchmarks while
manifesting robustness as much less sensitive to suboptimal hyperparameters.
Notably, SimCLR with DCL achieves 68.2% ImageNet-1K top-1 accuracy using batch
size 256 within 200 epochs pre-training, outperforming its SimCLR baseline by
6.4%. Further, DCL can be combined with the SOTA contrastive learning method,
NNCLR, to achieve 72.3% ImageNet-1K top-1 accuracy with 512 batch size in 400
epochs, which represents a new SOTA in contrastive learning. We believe DCL
provides a valuable baseline for future contrastive SSL studies.
| [
{
"version": "v1",
"created": "Wed, 13 Oct 2021 16:38:43 GMT"
},
{
"version": "v2",
"created": "Sat, 23 Oct 2021 07:03:26 GMT"
},
{
"version": "v3",
"created": "Sat, 30 Jul 2022 02:37:33 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Yeh",
"Chun-Hsiao",
""
],
[
"Hong",
"Cheng-Yao",
""
],
[
"Hsu",
"Yen-Chi",
""
],
[
"Liu",
"Tyng-Luh",
""
],
[
"Chen",
"Yubei",
""
],
[
"LeCun",
"Yann",
""
]
] | TITLE: Decoupled Contrastive Learning
ABSTRACT: Contrastive learning (CL) is one of the most successful paradigms for
self-supervised learning (SSL). In a principled way, it considers two augmented
"views" of the same image as positive to be pulled closer, and all other images
as negative to be pushed further apart. However, behind the impressive success
of CL-based techniques, their formulation often relies on heavy-computation
settings, including large sample batches, extensive training epochs, etc. We
are thus motivated to tackle these issues and establish a simple, efficient,
yet competitive baseline of contrastive learning. Specifically, we identify,
from theoretical and empirical studies, a noticeable negative-positive-coupling
(NPC) effect in the widely used InfoNCE loss, leading to unsuitable learning
efficiency concerning the batch size. By removing the NPC effect, we propose
decoupled contrastive learning (DCL) loss, which removes the positive term from
the denominator and significantly improves the learning efficiency. DCL
achieves competitive performance with less sensitivity to sub-optimal
hyperparameters, requiring neither large batches in SimCLR, momentum encoding
in MoCo, or large epochs. We demonstrate with various benchmarks while
manifesting robustness as much less sensitive to suboptimal hyperparameters.
Notably, SimCLR with DCL achieves 68.2% ImageNet-1K top-1 accuracy using batch
size 256 within 200 epochs pre-training, outperforming its SimCLR baseline by
6.4%. Further, DCL can be combined with the SOTA contrastive learning method,
NNCLR, to achieve 72.3% ImageNet-1K top-1 accuracy with 512 batch size in 400
epochs, which represents a new SOTA in contrastive learning. We believe DCL
provides a valuable baseline for future contrastive SSL studies.
|
2201.03521 | Karolina Seweryn | Daniel Ziembicki, Anna Wr\'oblewska, Karolina Seweryn | Polish Natural Language Inference and Factivity -- an Expert-based
Dataset and Benchmarks | null | null | 10.1017/S1351324923000220 | null | cs.CL cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Despite recent breakthroughs in Machine Learning for Natural Language
Processing, the Natural Language Inference (NLI) problems still constitute a
challenge. To this purpose we contribute a new dataset that focuses exclusively
on the factivity phenomenon; however, our task remains the same as other NLI
tasks, i.e. prediction of entailment, contradiction or neutral (ECN). The
dataset contains entirely natural language utterances in Polish and gathers
2,432 verb-complement pairs and 309 unique verbs. The dataset is based on the
National Corpus of Polish (NKJP) and is a representative sample in regards to
frequency of main verbs and other linguistic features (e.g. occurrence of
internal negation). We found that transformer BERT-based models working on
sentences obtained relatively good results ($\approx89\%$ F1 score). Even
though better results were achieved using linguistic features ($\approx91\%$ F1
score), this model requires more human labour (humans in the loop) because
features were prepared manually by expert linguists. BERT-based models
consuming only the input sentences show that they capture most of the
complexity of NLI/factivity. Complex cases in the phenomenon - e.g. cases with
entitlement (E) and non-factive verbs - remain an open issue for further
research.
| [
{
"version": "v1",
"created": "Mon, 10 Jan 2022 18:32:55 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Ziembicki",
"Daniel",
""
],
[
"Wróblewska",
"Anna",
""
],
[
"Seweryn",
"Karolina",
""
]
] | TITLE: Polish Natural Language Inference and Factivity -- an Expert-based
Dataset and Benchmarks
ABSTRACT: Despite recent breakthroughs in Machine Learning for Natural Language
Processing, the Natural Language Inference (NLI) problems still constitute a
challenge. To this purpose we contribute a new dataset that focuses exclusively
on the factivity phenomenon; however, our task remains the same as other NLI
tasks, i.e. prediction of entailment, contradiction or neutral (ECN). The
dataset contains entirely natural language utterances in Polish and gathers
2,432 verb-complement pairs and 309 unique verbs. The dataset is based on the
National Corpus of Polish (NKJP) and is a representative sample in regards to
frequency of main verbs and other linguistic features (e.g. occurrence of
internal negation). We found that transformer BERT-based models working on
sentences obtained relatively good results ($\approx89\%$ F1 score). Even
though better results were achieved using linguistic features ($\approx91\%$ F1
score), this model requires more human labour (humans in the loop) because
features were prepared manually by expert linguists. BERT-based models
consuming only the input sentences show that they capture most of the
complexity of NLI/factivity. Complex cases in the phenomenon - e.g. cases with
entitlement (E) and non-factive verbs - remain an open issue for further
research.
|
2201.05978 | Varun Ramamohan | Varun Ramamohan, Shobhit Singhal, Aditya Raj Gupta, Nomesh Bhojkumar
Bolia | Discrete Simulation Optimization for Tuning Machine Learning Method
Hyperparameters | null | Journal of Simulation (2023) | 10.1080/17477778.2023.2219401 | null | cs.LG cs.SY eess.SY | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Machine learning (ML) methods are used in most technical areas such as image
recognition, product recommendation, financial analysis, medical diagnosis, and
predictive maintenance. An important aspect of implementing ML methods involves
controlling the learning process for the ML method so as to maximize the
performance of the method under consideration. Hyperparameter tuning is the
process of selecting a suitable set of ML method parameters that control its
learning process. In this work, we demonstrate the use of discrete simulation
optimization methods such as ranking and selection (R&S) and random search for
identifying a hyperparameter set that maximizes the performance of a ML method.
Specifically, we use the KN R&S method and the stochastic ruler random search
method and one of its variations for this purpose. We also construct the
theoretical basis for applying the KN method, which determines the optimal
solution with a statistical guarantee via solution space enumeration. In
comparison, the stochastic ruler method asymptotically converges to global
optima and incurs smaller computational overheads. We demonstrate the
application of these methods to a wide variety of machine learning models,
including deep neural network models used for time series prediction and image
classification. We benchmark our application of these methods with
state-of-the-art hyperparameter optimization libraries such as $hyperopt$ and
$mango$. The KN method consistently outperforms $hyperopt$'s random search (RS)
and Tree of Parzen Estimators (TPE) methods. The stochastic ruler method
outperforms the $hyperopt$ RS method and offers statistically comparable
performance with respect to $hyperopt$'s TPE method and the $mango$ algorithm.
| [
{
"version": "v1",
"created": "Sun, 16 Jan 2022 06:01:58 GMT"
},
{
"version": "v2",
"created": "Wed, 4 May 2022 06:07:01 GMT"
},
{
"version": "v3",
"created": "Tue, 20 Jun 2023 05:16:51 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Ramamohan",
"Varun",
""
],
[
"Singhal",
"Shobhit",
""
],
[
"Gupta",
"Aditya Raj",
""
],
[
"Bolia",
"Nomesh Bhojkumar",
""
]
] | TITLE: Discrete Simulation Optimization for Tuning Machine Learning Method
Hyperparameters
ABSTRACT: Machine learning (ML) methods are used in most technical areas such as image
recognition, product recommendation, financial analysis, medical diagnosis, and
predictive maintenance. An important aspect of implementing ML methods involves
controlling the learning process for the ML method so as to maximize the
performance of the method under consideration. Hyperparameter tuning is the
process of selecting a suitable set of ML method parameters that control its
learning process. In this work, we demonstrate the use of discrete simulation
optimization methods such as ranking and selection (R&S) and random search for
identifying a hyperparameter set that maximizes the performance of a ML method.
Specifically, we use the KN R&S method and the stochastic ruler random search
method and one of its variations for this purpose. We also construct the
theoretical basis for applying the KN method, which determines the optimal
solution with a statistical guarantee via solution space enumeration. In
comparison, the stochastic ruler method asymptotically converges to global
optima and incurs smaller computational overheads. We demonstrate the
application of these methods to a wide variety of machine learning models,
including deep neural network models used for time series prediction and image
classification. We benchmark our application of these methods with
state-of-the-art hyperparameter optimization libraries such as $hyperopt$ and
$mango$. The KN method consistently outperforms $hyperopt$'s random search (RS)
and Tree of Parzen Estimators (TPE) methods. The stochastic ruler method
outperforms the $hyperopt$ RS method and offers statistically comparable
performance with respect to $hyperopt$'s TPE method and the $mango$ algorithm.
|
2202.07165 | Fumiyuki Kato | Fumiyuki Kato, Yang Cao, Masatoshi Yoshikawa | OLIVE: Oblivious Federated Learning on Trusted Execution Environment
against the risk of sparsification | This paper is the full version of a paper accepted at VLDB 2023 | null | null | null | cs.LG cs.CR | http://creativecommons.org/licenses/by/4.0/ | Combining Federated Learning (FL) with a Trusted Execution Environment (TEE)
is a promising approach for realizing privacy-preserving FL, which has garnered
significant academic attention in recent years. Implementing the TEE on the
server side enables each round of FL to proceed without exposing the client's
gradient information to untrusted servers. This addresses usability gaps in
existing secure aggregation schemes as well as utility gaps in differentially
private FL. However, to address the issue using a TEE, the vulnerabilities of
server-side TEEs need to be considered -- this has not been sufficiently
investigated in the context of FL. The main technical contribution of this
study is the analysis of the vulnerabilities of TEE in FL and the defense.
First, we theoretically analyze the leakage of memory access patterns,
revealing the risk of sparsified gradients, which are commonly used in FL to
enhance communication efficiency and model accuracy. Second, we devise an
inference attack to link memory access patterns to sensitive information in the
training dataset. Finally, we propose an oblivious yet efficient aggregation
algorithm to prevent memory access pattern leakage. Our experiments on
real-world data demonstrate that the proposed method functions efficiently in
practical scales.
| [
{
"version": "v1",
"created": "Tue, 15 Feb 2022 03:23:57 GMT"
},
{
"version": "v2",
"created": "Wed, 16 Feb 2022 07:51:15 GMT"
},
{
"version": "v3",
"created": "Thu, 1 Dec 2022 12:38:50 GMT"
},
{
"version": "v4",
"created": "Wed, 12 Apr 2023 02:30:32 GMT"
},
{
"version": "v5",
"created": "Mon, 19 Jun 2023 13:54:11 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Kato",
"Fumiyuki",
""
],
[
"Cao",
"Yang",
""
],
[
"Yoshikawa",
"Masatoshi",
""
]
] | TITLE: OLIVE: Oblivious Federated Learning on Trusted Execution Environment
against the risk of sparsification
ABSTRACT: Combining Federated Learning (FL) with a Trusted Execution Environment (TEE)
is a promising approach for realizing privacy-preserving FL, which has garnered
significant academic attention in recent years. Implementing the TEE on the
server side enables each round of FL to proceed without exposing the client's
gradient information to untrusted servers. This addresses usability gaps in
existing secure aggregation schemes as well as utility gaps in differentially
private FL. However, to address the issue using a TEE, the vulnerabilities of
server-side TEEs need to be considered -- this has not been sufficiently
investigated in the context of FL. The main technical contribution of this
study is the analysis of the vulnerabilities of TEE in FL and the defense.
First, we theoretically analyze the leakage of memory access patterns,
revealing the risk of sparsified gradients, which are commonly used in FL to
enhance communication efficiency and model accuracy. Second, we devise an
inference attack to link memory access patterns to sensitive information in the
training dataset. Finally, we propose an oblivious yet efficient aggregation
algorithm to prevent memory access pattern leakage. Our experiments on
real-world data demonstrate that the proposed method functions efficiently in
practical scales.
|
2203.04301 | Tong Yu | Tong Yu, Pietro Mascagni, Juan Verde, Jacques Marescaux, Didier
Mutter, Nicolas Padoy | Live Laparoscopic Video Retrieval with Compressed Uncertainty | 16 pages, 13 figures | Medical Image Analysis 88 (2023) 102866 | 10.1016/j.media.2023.102866 | null | eess.IV cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Searching through large volumes of medical data to retrieve relevant
information is a challenging yet crucial task for clinical care. However the
primitive and most common approach to retrieval, involving text in the form of
keywords, is severely limited when dealing with complex media formats.
Content-based retrieval offers a way to overcome this limitation, by using rich
media as the query itself. Surgical video-to-video retrieval in particular is a
new and largely unexplored research problem with high clinical value,
especially in the real-time case: using real-time video hashing, search can be
achieved directly inside of the operating room. Indeed, the process of hashing
converts large data entries into compact binary arrays or hashes, enabling
large-scale search operations at a very fast rate. However, due to fluctuations
over the course of a video, not all bits in a given hash are equally reliable.
In this work, we propose a method capable of mitigating this uncertainty while
maintaining a light computational footprint. We present superior retrieval
results (3-4 % top 10 mean average precision) on a multi-task evaluation
protocol for surgery, using cholecystectomy phases, bypass phases, and coming
from an entirely new dataset introduced here, critical events across six
different surgery types. Success on this multi-task benchmark shows the
generalizability of our approach for surgical video retrieval.
| [
{
"version": "v1",
"created": "Tue, 8 Mar 2022 09:37:57 GMT"
},
{
"version": "v2",
"created": "Mon, 12 Jun 2023 12:51:28 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Yu",
"Tong",
""
],
[
"Mascagni",
"Pietro",
""
],
[
"Verde",
"Juan",
""
],
[
"Marescaux",
"Jacques",
""
],
[
"Mutter",
"Didier",
""
],
[
"Padoy",
"Nicolas",
""
]
] | TITLE: Live Laparoscopic Video Retrieval with Compressed Uncertainty
ABSTRACT: Searching through large volumes of medical data to retrieve relevant
information is a challenging yet crucial task for clinical care. However the
primitive and most common approach to retrieval, involving text in the form of
keywords, is severely limited when dealing with complex media formats.
Content-based retrieval offers a way to overcome this limitation, by using rich
media as the query itself. Surgical video-to-video retrieval in particular is a
new and largely unexplored research problem with high clinical value,
especially in the real-time case: using real-time video hashing, search can be
achieved directly inside of the operating room. Indeed, the process of hashing
converts large data entries into compact binary arrays or hashes, enabling
large-scale search operations at a very fast rate. However, due to fluctuations
over the course of a video, not all bits in a given hash are equally reliable.
In this work, we propose a method capable of mitigating this uncertainty while
maintaining a light computational footprint. We present superior retrieval
results (3-4 % top 10 mean average precision) on a multi-task evaluation
protocol for surgery, using cholecystectomy phases, bypass phases, and coming
from an entirely new dataset introduced here, critical events across six
different surgery types. Success on this multi-task benchmark shows the
generalizability of our approach for surgical video retrieval.
|
2203.06410 | Shi-Xue Zhang | Shi-Xue Zhang, Xiaobin Zhu, Jie-Bo Hou, Chun Yang, Xu-Cheng Yin | Kernel Proposal Network for Arbitrary Shape Text Detection | This paper was completed in 2020-11.It was first submitted to CVPR
2021 and then ICCV 2021. Finally, it has been accepted by TNNLS in 2022-02
after major revision. Here, I thank Dr.Hou for his important contributions | null | null | null | cs.CV | http://creativecommons.org/licenses/by-sa/4.0/ | Segmentation-based methods have achieved great success for arbitrary shape
text detection. However, separating neighboring text instances is still one of
the most challenging problems due to the complexity of texts in scene images.
In this paper, we propose an innovative Kernel Proposal Network (dubbed KPN)
for arbitrary shape text detection. The proposed KPN can separate neighboring
text instances by classifying different texts into instance-independent feature
maps, meanwhile avoiding the complex aggregation process existing in
segmentation-based arbitrary shape text detection methods. To be concrete, our
KPN will predict a Gaussian center map for each text image, which will be used
to extract a series of candidate kernel proposals (i.e., dynamic convolution
kernel) from the embedding feature maps according to their corresponding
keypoint positions. To enforce the independence between kernel proposals, we
propose a novel orthogonal learning loss (OLL) via orthogonal constraints.
Specifically, our kernel proposals contain important self-information learned
by network and location information by position embedding. Finally, kernel
proposals will individually convolve all embedding feature maps for generating
individual embedded maps of text instances. In this way, our KPN can
effectively separate neighboring text instances and improve the robustness
against unclear boundaries. To our knowledge, our work is the first to
introduce the dynamic convolution kernel strategy to efficiently and
effectively tackle the adhesion problem of neighboring text instances in text
detection. Experimental results on challenging datasets verify the impressive
performance and efficiency of our method. The code and model are available at
https://github.com/GXYM/KPN.
| [
{
"version": "v1",
"created": "Sat, 12 Mar 2022 11:02:32 GMT"
},
{
"version": "v2",
"created": "Tue, 20 Jun 2023 03:18:52 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Zhang",
"Shi-Xue",
""
],
[
"Zhu",
"Xiaobin",
""
],
[
"Hou",
"Jie-Bo",
""
],
[
"Yang",
"Chun",
""
],
[
"Yin",
"Xu-Cheng",
""
]
] | TITLE: Kernel Proposal Network for Arbitrary Shape Text Detection
ABSTRACT: Segmentation-based methods have achieved great success for arbitrary shape
text detection. However, separating neighboring text instances is still one of
the most challenging problems due to the complexity of texts in scene images.
In this paper, we propose an innovative Kernel Proposal Network (dubbed KPN)
for arbitrary shape text detection. The proposed KPN can separate neighboring
text instances by classifying different texts into instance-independent feature
maps, meanwhile avoiding the complex aggregation process existing in
segmentation-based arbitrary shape text detection methods. To be concrete, our
KPN will predict a Gaussian center map for each text image, which will be used
to extract a series of candidate kernel proposals (i.e., dynamic convolution
kernel) from the embedding feature maps according to their corresponding
keypoint positions. To enforce the independence between kernel proposals, we
propose a novel orthogonal learning loss (OLL) via orthogonal constraints.
Specifically, our kernel proposals contain important self-information learned
by network and location information by position embedding. Finally, kernel
proposals will individually convolve all embedding feature maps for generating
individual embedded maps of text instances. In this way, our KPN can
effectively separate neighboring text instances and improve the robustness
against unclear boundaries. To our knowledge, our work is the first to
introduce the dynamic convolution kernel strategy to efficiently and
effectively tackle the adhesion problem of neighboring text instances in text
detection. Experimental results on challenging datasets verify the impressive
performance and efficiency of our method. The code and model are available at
https://github.com/GXYM/KPN.
|
2203.10247 | Qing Cai | Qing Cai, Yiming Qian, Jinxing Li, Jun Lv, Yee-Hong Yang, Feng Wu,
David Zhang | HIPA: Hierarchical Patch Transformer for Single Image Super Resolution | null | null | 10.1109/TIP.2023.3279977 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Transformer-based architectures start to emerge in single image super
resolution (SISR) and have achieved promising performance. Most existing Vision
Transformers divide images into the same number of patches with a fixed size,
which may not be optimal for restoring patches with different levels of texture
richness. This paper presents HIPA, a novel Transformer architecture that
progressively recovers the high resolution image using a hierarchical patch
partition. Specifically, we build a cascaded model that processes an input
image in multiple stages, where we start with tokens with small patch sizes and
gradually merge to the full resolution. Such a hierarchical patch mechanism not
only explicitly enables feature aggregation at multiple resolutions but also
adaptively learns patch-aware features for different image regions, e.g., using
a smaller patch for areas with fine details and a larger patch for textureless
regions. Meanwhile, a new attention-based position encoding scheme for
Transformer is proposed to let the network focus on which tokens should be paid
more attention by assigning different weights to different tokens, which is the
first time to our best knowledge. Furthermore, we also propose a new
multi-reception field attention module to enlarge the convolution reception
field from different branches. The experimental results on several public
datasets demonstrate the superior performance of the proposed HIPA over
previous methods quantitatively and qualitatively.
| [
{
"version": "v1",
"created": "Sat, 19 Mar 2022 05:09:34 GMT"
},
{
"version": "v2",
"created": "Wed, 7 Jun 2023 01:39:31 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Cai",
"Qing",
""
],
[
"Qian",
"Yiming",
""
],
[
"Li",
"Jinxing",
""
],
[
"Lv",
"Jun",
""
],
[
"Yang",
"Yee-Hong",
""
],
[
"Wu",
"Feng",
""
],
[
"Zhang",
"David",
""
]
] | TITLE: HIPA: Hierarchical Patch Transformer for Single Image Super Resolution
ABSTRACT: Transformer-based architectures start to emerge in single image super
resolution (SISR) and have achieved promising performance. Most existing Vision
Transformers divide images into the same number of patches with a fixed size,
which may not be optimal for restoring patches with different levels of texture
richness. This paper presents HIPA, a novel Transformer architecture that
progressively recovers the high resolution image using a hierarchical patch
partition. Specifically, we build a cascaded model that processes an input
image in multiple stages, where we start with tokens with small patch sizes and
gradually merge to the full resolution. Such a hierarchical patch mechanism not
only explicitly enables feature aggregation at multiple resolutions but also
adaptively learns patch-aware features for different image regions, e.g., using
a smaller patch for areas with fine details and a larger patch for textureless
regions. Meanwhile, a new attention-based position encoding scheme for
Transformer is proposed to let the network focus on which tokens should be paid
more attention by assigning different weights to different tokens, which is the
first time to our best knowledge. Furthermore, we also propose a new
multi-reception field attention module to enlarge the convolution reception
field from different branches. The experimental results on several public
datasets demonstrate the superior performance of the proposed HIPA over
previous methods quantitatively and qualitatively.
|
2203.16952 | Swalpa Kumar Roy Dr. | Swalpa Kumar Roy, Ankur Deria, Danfeng Hong, Behnood Rasti, Antonio
Plaza, Jocelyn Chanussot | Multimodal Fusion Transformer for Remote Sensing Image Classification | Published in IEEE Transactions on Geoscience and Remote Sensing | null | 10.1109/TGRS.2023.3286826 | null | cs.CV cs.LG eess.IV | http://creativecommons.org/licenses/by/4.0/ | Vision transformers (ViTs) have been trending in image classification tasks
due to their promising performance when compared to convolutional neural
networks (CNNs). As a result, many researchers have tried to incorporate ViTs
in hyperspectral image (HSI) classification tasks. To achieve satisfactory
performance, close to that of CNNs, transformers need fewer parameters. ViTs
and other similar transformers use an external classification (CLS) token which
is randomly initialized and often fails to generalize well, whereas other
sources of multimodal datasets, such as light detection and ranging (LiDAR)
offer the potential to improve these models by means of a CLS. In this paper,
we introduce a new multimodal fusion transformer (MFT) network which comprises
a multihead cross patch attention (mCrossPA) for HSI land-cover classification.
Our mCrossPA utilizes other sources of complementary information in addition to
the HSI in the transformer encoder to achieve better generalization. The
concept of tokenization is used to generate CLS and HSI patch tokens, helping
to learn a {distinctive representation} in a reduced and hierarchical feature
space. Extensive experiments are carried out on {widely used benchmark}
datasets {i.e.,} the University of Houston, Trento, University of Southern
Mississippi Gulfpark (MUUFL), and Augsburg. We compare the results of the
proposed MFT model with other state-of-the-art transformers, classical CNNs,
and conventional classifiers models. The superior performance achieved by the
proposed model is due to the use of multihead cross patch attention. The source
code will be made available publicly at
\url{https://github.com/AnkurDeria/MFT}.}
| [
{
"version": "v1",
"created": "Thu, 31 Mar 2022 11:18:41 GMT"
},
{
"version": "v2",
"created": "Tue, 20 Jun 2023 17:58:25 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Roy",
"Swalpa Kumar",
""
],
[
"Deria",
"Ankur",
""
],
[
"Hong",
"Danfeng",
""
],
[
"Rasti",
"Behnood",
""
],
[
"Plaza",
"Antonio",
""
],
[
"Chanussot",
"Jocelyn",
""
]
] | TITLE: Multimodal Fusion Transformer for Remote Sensing Image Classification
ABSTRACT: Vision transformers (ViTs) have been trending in image classification tasks
due to their promising performance when compared to convolutional neural
networks (CNNs). As a result, many researchers have tried to incorporate ViTs
in hyperspectral image (HSI) classification tasks. To achieve satisfactory
performance, close to that of CNNs, transformers need fewer parameters. ViTs
and other similar transformers use an external classification (CLS) token which
is randomly initialized and often fails to generalize well, whereas other
sources of multimodal datasets, such as light detection and ranging (LiDAR)
offer the potential to improve these models by means of a CLS. In this paper,
we introduce a new multimodal fusion transformer (MFT) network which comprises
a multihead cross patch attention (mCrossPA) for HSI land-cover classification.
Our mCrossPA utilizes other sources of complementary information in addition to
the HSI in the transformer encoder to achieve better generalization. The
concept of tokenization is used to generate CLS and HSI patch tokens, helping
to learn a {distinctive representation} in a reduced and hierarchical feature
space. Extensive experiments are carried out on {widely used benchmark}
datasets {i.e.,} the University of Houston, Trento, University of Southern
Mississippi Gulfpark (MUUFL), and Augsburg. We compare the results of the
proposed MFT model with other state-of-the-art transformers, classical CNNs,
and conventional classifiers models. The superior performance achieved by the
proposed model is due to the use of multihead cross patch attention. The source
code will be made available publicly at
\url{https://github.com/AnkurDeria/MFT}.}
|
2205.03562 | Shi-Xue Zhang | Shi-Xue Zhang, Xiaobin Zhu, Jie-Bo Hou, Xu-Cheng Yin | Graph Fusion Network for Multi-Oriented Object Detection | Accepted by Applied Intelligence (APIN 2022) | null | null | null | cs.CV | http://creativecommons.org/licenses/by-sa/4.0/ | In object detection, non-maximum suppression (NMS) methods are extensively
adopted to remove horizontal duplicates of detected dense boxes for generating
final object instances. However, due to the degraded quality of dense detection
boxes and not explicit exploration of the context information, existing NMS
methods via simple intersection-over-union (IoU) metrics tend to underperform
on multi-oriented and long-size objects detection. Distinguishing with general
NMS methods via duplicate removal, we propose a novel graph fusion network,
named GFNet, for multi-oriented object detection. Our GFNet is extensible and
adaptively fuse dense detection boxes to detect more accurate and holistic
multi-oriented object instances. Specifically, we first adopt a locality-aware
clustering algorithm to group dense detection boxes into different clusters. We
will construct an instance sub-graph for the detection boxes belonging to one
cluster. Then, we propose a graph-based fusion network via Graph Convolutional
Network (GCN) to learn to reason and fuse the detection boxes for generating
final instance boxes. Extensive experiments both on public available
multi-oriented text datasets (including MSRA-TD500, ICDAR2015, ICDAR2017-MLT)
and multi-oriented object datasets (DOTA) verify the effectiveness and
robustness of our method against general NMS methods in multi-oriented object
detection.
| [
{
"version": "v1",
"created": "Sat, 7 May 2022 05:47:22 GMT"
},
{
"version": "v2",
"created": "Fri, 26 Aug 2022 03:01:40 GMT"
},
{
"version": "v3",
"created": "Tue, 20 Jun 2023 03:50:38 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Zhang",
"Shi-Xue",
""
],
[
"Zhu",
"Xiaobin",
""
],
[
"Hou",
"Jie-Bo",
""
],
[
"Yin",
"Xu-Cheng",
""
]
] | TITLE: Graph Fusion Network for Multi-Oriented Object Detection
ABSTRACT: In object detection, non-maximum suppression (NMS) methods are extensively
adopted to remove horizontal duplicates of detected dense boxes for generating
final object instances. However, due to the degraded quality of dense detection
boxes and not explicit exploration of the context information, existing NMS
methods via simple intersection-over-union (IoU) metrics tend to underperform
on multi-oriented and long-size objects detection. Distinguishing with general
NMS methods via duplicate removal, we propose a novel graph fusion network,
named GFNet, for multi-oriented object detection. Our GFNet is extensible and
adaptively fuse dense detection boxes to detect more accurate and holistic
multi-oriented object instances. Specifically, we first adopt a locality-aware
clustering algorithm to group dense detection boxes into different clusters. We
will construct an instance sub-graph for the detection boxes belonging to one
cluster. Then, we propose a graph-based fusion network via Graph Convolutional
Network (GCN) to learn to reason and fuse the detection boxes for generating
final instance boxes. Extensive experiments both on public available
multi-oriented text datasets (including MSRA-TD500, ICDAR2015, ICDAR2017-MLT)
and multi-oriented object datasets (DOTA) verify the effectiveness and
robustness of our method against general NMS methods in multi-oriented object
detection.
|
2205.05320 | Shi-Xue Zhang | Shi-Xue Zhang, Chun Yang, Xiaobin Zhu, Xu-Cheng Yin | Arbitrary Shape Text Detection via Boundary Transformer | It is an extend version (TextBPN++) to our preliminary conference
version TextBPN(ICCV 2021) [arXiv:2107.12664], which has been accepted by
IEEE Transactions on Multimedia (T-MM 2023) | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | In arbitrary shape text detection, locating accurate text boundaries is
challenging and non-trivial. Existing methods often suffer from indirect text
boundary modeling or complex post-processing. In this paper, we systematically
present a unified coarse-to-fine framework via boundary learning for arbitrary
shape text detection, which can accurately and efficiently locate text
boundaries without post-processing. In our method, we explicitly model the text
boundary via an innovative iterative boundary transformer in a coarse-to-fine
manner. In this way, our method can directly gain accurate text boundaries and
abandon complex post-processing to improve efficiency. Specifically, our method
mainly consists of a feature extraction backbone, a boundary proposal module,
and an iteratively optimized boundary transformer module. The boundary proposal
module consisting of multi-layer dilated convolutions will compute important
prior information (including classification map, distance field, and direction
field) for generating coarse boundary proposals while guiding the boundary
transformer's optimization. The boundary transformer module adopts an
encoder-decoder structure, in which the encoder is constructed by multi-layer
transformer blocks with residual connection while the decoder is a simple
multi-layer perceptron network (MLP). Under the guidance of prior information,
the boundary transformer module will gradually refine the coarse boundary
proposals via iterative boundary deformation. Furthermore, we propose a novel
boundary energy loss (BEL) which introduces an energy minimization constraint
and an energy monotonically decreasing constraint to further optimize and
stabilize the learning of boundary refinement. Extensive experiments on
publicly available and challenging datasets demonstrate the state-of-the-art
performance and promising efficiency of our method.
| [
{
"version": "v1",
"created": "Wed, 11 May 2022 07:59:13 GMT"
},
{
"version": "v2",
"created": "Tue, 30 Aug 2022 08:08:19 GMT"
},
{
"version": "v3",
"created": "Tue, 13 Jun 2023 15:17:51 GMT"
},
{
"version": "v4",
"created": "Tue, 20 Jun 2023 03:00:29 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Zhang",
"Shi-Xue",
""
],
[
"Yang",
"Chun",
""
],
[
"Zhu",
"Xiaobin",
""
],
[
"Yin",
"Xu-Cheng",
""
]
] | TITLE: Arbitrary Shape Text Detection via Boundary Transformer
ABSTRACT: In arbitrary shape text detection, locating accurate text boundaries is
challenging and non-trivial. Existing methods often suffer from indirect text
boundary modeling or complex post-processing. In this paper, we systematically
present a unified coarse-to-fine framework via boundary learning for arbitrary
shape text detection, which can accurately and efficiently locate text
boundaries without post-processing. In our method, we explicitly model the text
boundary via an innovative iterative boundary transformer in a coarse-to-fine
manner. In this way, our method can directly gain accurate text boundaries and
abandon complex post-processing to improve efficiency. Specifically, our method
mainly consists of a feature extraction backbone, a boundary proposal module,
and an iteratively optimized boundary transformer module. The boundary proposal
module consisting of multi-layer dilated convolutions will compute important
prior information (including classification map, distance field, and direction
field) for generating coarse boundary proposals while guiding the boundary
transformer's optimization. The boundary transformer module adopts an
encoder-decoder structure, in which the encoder is constructed by multi-layer
transformer blocks with residual connection while the decoder is a simple
multi-layer perceptron network (MLP). Under the guidance of prior information,
the boundary transformer module will gradually refine the coarse boundary
proposals via iterative boundary deformation. Furthermore, we propose a novel
boundary energy loss (BEL) which introduces an energy minimization constraint
and an energy monotonically decreasing constraint to further optimize and
stabilize the learning of boundary refinement. Extensive experiments on
publicly available and challenging datasets demonstrate the state-of-the-art
performance and promising efficiency of our method.
|
2205.07050 | Vicky Kouni Ms | Vicky Kouni, Yannis Panagakis | DECONET: an Unfolding Network for Analysis-based Compressed Sensing with
Generalization Error Bounds | Accepted in IEEE Transactions on Signal Processing | null | 10.1109/TSP.2023.3272286 | null | cs.IT cs.IR cs.LG math.IT | http://creativecommons.org/licenses/by/4.0/ | We present a new deep unfolding network for analysis-sparsity-based
Compressed Sensing. The proposed network coined Decoding Network (DECONET)
jointly learns a decoder that reconstructs vectors from their incomplete, noisy
measurements and a redundant sparsifying analysis operator, which is shared
across the layers of DECONET. Moreover, we formulate the hypothesis class of
DECONET and estimate its associated Rademacher complexity. Then, we use this
estimate to deliver meaningful upper bounds for the generalization error of
DECONET. Finally, the validity of our theoretical results is assessed and
comparisons to state-of-the-art unfolding networks are made, on both synthetic
and real-world datasets. Experimental results indicate that our proposed
network outperforms the baselines, consistently for all datasets, and its
behaviour complies with our theoretical findings.
| [
{
"version": "v1",
"created": "Sat, 14 May 2022 12:50:48 GMT"
},
{
"version": "v2",
"created": "Sat, 21 May 2022 14:50:15 GMT"
},
{
"version": "v3",
"created": "Sat, 11 Jun 2022 18:33:23 GMT"
},
{
"version": "v4",
"created": "Wed, 10 Aug 2022 20:30:39 GMT"
},
{
"version": "v5",
"created": "Sat, 13 Aug 2022 14:37:08 GMT"
},
{
"version": "v6",
"created": "Wed, 26 Apr 2023 10:22:07 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Kouni",
"Vicky",
""
],
[
"Panagakis",
"Yannis",
""
]
] | TITLE: DECONET: an Unfolding Network for Analysis-based Compressed Sensing with
Generalization Error Bounds
ABSTRACT: We present a new deep unfolding network for analysis-sparsity-based
Compressed Sensing. The proposed network coined Decoding Network (DECONET)
jointly learns a decoder that reconstructs vectors from their incomplete, noisy
measurements and a redundant sparsifying analysis operator, which is shared
across the layers of DECONET. Moreover, we formulate the hypothesis class of
DECONET and estimate its associated Rademacher complexity. Then, we use this
estimate to deliver meaningful upper bounds for the generalization error of
DECONET. Finally, the validity of our theoretical results is assessed and
comparisons to state-of-the-art unfolding networks are made, on both synthetic
and real-world datasets. Experimental results indicate that our proposed
network outperforms the baselines, consistently for all datasets, and its
behaviour complies with our theoretical findings.
|
2205.12424 | Hazim Hanif | Hazim Hanif and Sergio Maffeis | VulBERTa: Simplified Source Code Pre-Training for Vulnerability
Detection | Accepted as a conference paper at IJCNN 2022 | International Joint Conference on Neural Networks (IJCNN), 2022 | 10.1109/IJCNN55064.2022.9892280 | null | cs.CR cs.AI cs.LG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This paper presents VulBERTa, a deep learning approach to detect security
vulnerabilities in source code. Our approach pre-trains a RoBERTa model with a
custom tokenisation pipeline on real-world code from open-source C/C++
projects. The model learns a deep knowledge representation of the code syntax
and semantics, which we leverage to train vulnerability detection classifiers.
We evaluate our approach on binary and multi-class vulnerability detection
tasks across several datasets (Vuldeepecker, Draper, REVEAL and muVuldeepecker)
and benchmarks (CodeXGLUE and D2A). The evaluation results show that VulBERTa
achieves state-of-the-art performance and outperforms existing approaches
across different datasets, despite its conceptual simplicity, and limited cost
in terms of size of training data and number of model parameters.
| [
{
"version": "v1",
"created": "Wed, 25 May 2022 00:56:43 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Hanif",
"Hazim",
""
],
[
"Maffeis",
"Sergio",
""
]
] | TITLE: VulBERTa: Simplified Source Code Pre-Training for Vulnerability
Detection
ABSTRACT: This paper presents VulBERTa, a deep learning approach to detect security
vulnerabilities in source code. Our approach pre-trains a RoBERTa model with a
custom tokenisation pipeline on real-world code from open-source C/C++
projects. The model learns a deep knowledge representation of the code syntax
and semantics, which we leverage to train vulnerability detection classifiers.
We evaluate our approach on binary and multi-class vulnerability detection
tasks across several datasets (Vuldeepecker, Draper, REVEAL and muVuldeepecker)
and benchmarks (CodeXGLUE and D2A). The evaluation results show that VulBERTa
achieves state-of-the-art performance and outperforms existing approaches
across different datasets, despite its conceptual simplicity, and limited cost
in terms of size of training data and number of model parameters.
|
2205.13005 | Ryan Kortvelesy | Ryan Kortvelesy and Amanda Prorok | QGNN: Value Function Factorisation with Graph Neural Networks | null | null | null | null | cs.LG cs.MA | http://creativecommons.org/licenses/by/4.0/ | In multi-agent reinforcement learning, the use of a global objective is a
powerful tool for incentivising cooperation. Unfortunately, it is not
sample-efficient to train individual agents with a global reward, because it
does not necessarily correlate with an agent's individual actions. This problem
can be solved by factorising the global value function into local value
functions. Early work in this domain performed factorisation by conditioning
local value functions purely on local information. Recently, it has been shown
that providing both local information and an encoding of the global state can
promote cooperative behaviour. In this paper we propose QGNN, the first value
factorisation method to use a graph neural network (GNN) based model. The
multi-layer message passing architecture of QGNN provides more representational
complexity than models in prior work, allowing it to produce a more effective
factorisation. QGNN also introduces a permutation invariant mixer which is able
to match the performance of other methods, even with significantly fewer
parameters. We evaluate our method against several baselines, including
QMIX-Att, GraphMIX, QMIX, VDN, and hybrid architectures. Our experiments
include Starcraft, the standard benchmark for credit assignment; Estimate Game,
a custom environment that explicitly models inter-agent dependencies; and
Coalition Structure Generation, a foundational problem with real-world
applications. The results show that QGNN outperforms state-of-the-art value
factorisation baselines consistently.
| [
{
"version": "v1",
"created": "Wed, 25 May 2022 18:35:11 GMT"
},
{
"version": "v2",
"created": "Tue, 20 Jun 2023 16:53:06 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Kortvelesy",
"Ryan",
""
],
[
"Prorok",
"Amanda",
""
]
] | TITLE: QGNN: Value Function Factorisation with Graph Neural Networks
ABSTRACT: In multi-agent reinforcement learning, the use of a global objective is a
powerful tool for incentivising cooperation. Unfortunately, it is not
sample-efficient to train individual agents with a global reward, because it
does not necessarily correlate with an agent's individual actions. This problem
can be solved by factorising the global value function into local value
functions. Early work in this domain performed factorisation by conditioning
local value functions purely on local information. Recently, it has been shown
that providing both local information and an encoding of the global state can
promote cooperative behaviour. In this paper we propose QGNN, the first value
factorisation method to use a graph neural network (GNN) based model. The
multi-layer message passing architecture of QGNN provides more representational
complexity than models in prior work, allowing it to produce a more effective
factorisation. QGNN also introduces a permutation invariant mixer which is able
to match the performance of other methods, even with significantly fewer
parameters. We evaluate our method against several baselines, including
QMIX-Att, GraphMIX, QMIX, VDN, and hybrid architectures. Our experiments
include Starcraft, the standard benchmark for credit assignment; Estimate Game,
a custom environment that explicitly models inter-agent dependencies; and
Coalition Structure Generation, a foundational problem with real-world
applications. The results show that QGNN outperforms state-of-the-art value
factorisation baselines consistently.
|
2205.13094 | Niladri Chatterji | Niladri S. Chatterji, Saminul Haque, Tatsunori Hashimoto | Undersampling is a Minimax Optimal Robustness Intervention in
Nonparametric Classification | null | null | null | null | cs.LG cs.AI math.ST stat.ML stat.TH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While a broad range of techniques have been proposed to tackle distribution
shift, the simple baseline of training on an $\textit{undersampled}$ balanced
dataset often achieves close to state-of-the-art-accuracy across several
popular benchmarks. This is rather surprising, since undersampling algorithms
discard excess majority group data. To understand this phenomenon, we ask if
learning is fundamentally constrained by a lack of minority group samples. We
prove that this is indeed the case in the setting of nonparametric binary
classification. Our results show that in the worst case, an algorithm cannot
outperform undersampling unless there is a high degree of overlap between the
train and test distributions (which is unlikely to be the case in real-world
datasets), or if the algorithm leverages additional structure about the
distribution shift. In particular, in the case of label shift we show that
there is always an undersampling algorithm that is minimax optimal. In the case
of group-covariate shift we show that there is an undersampling algorithm that
is minimax optimal when the overlap between the group distributions is small.
We also perform an experimental case study on a label shift dataset and find
that in line with our theory, the test accuracy of robust neural network
classifiers is constrained by the number of minority samples.
| [
{
"version": "v1",
"created": "Thu, 26 May 2022 00:35:11 GMT"
},
{
"version": "v2",
"created": "Sun, 21 Aug 2022 20:22:28 GMT"
},
{
"version": "v3",
"created": "Sun, 25 Sep 2022 06:08:55 GMT"
},
{
"version": "v4",
"created": "Mon, 19 Jun 2023 15:44:18 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Chatterji",
"Niladri S.",
""
],
[
"Haque",
"Saminul",
""
],
[
"Hashimoto",
"Tatsunori",
""
]
] | TITLE: Undersampling is a Minimax Optimal Robustness Intervention in
Nonparametric Classification
ABSTRACT: While a broad range of techniques have been proposed to tackle distribution
shift, the simple baseline of training on an $\textit{undersampled}$ balanced
dataset often achieves close to state-of-the-art-accuracy across several
popular benchmarks. This is rather surprising, since undersampling algorithms
discard excess majority group data. To understand this phenomenon, we ask if
learning is fundamentally constrained by a lack of minority group samples. We
prove that this is indeed the case in the setting of nonparametric binary
classification. Our results show that in the worst case, an algorithm cannot
outperform undersampling unless there is a high degree of overlap between the
train and test distributions (which is unlikely to be the case in real-world
datasets), or if the algorithm leverages additional structure about the
distribution shift. In particular, in the case of label shift we show that
there is always an undersampling algorithm that is minimax optimal. In the case
of group-covariate shift we show that there is an undersampling algorithm that
is minimax optimal when the overlap between the group distributions is small.
We also perform an experimental case study on a label shift dataset and find
that in line with our theory, the test accuracy of robust neural network
classifiers is constrained by the number of minority samples.
|
2205.13616 | Xiangyu Qi | Xiangyu Qi, Tinghao Xie, Jiachen T. Wang, Tong Wu, Saeed Mahloujifar,
Prateek Mittal | Towards A Proactive ML Approach for Detecting Backdoor Poison Samples | USENIX Security 2023 | null | null | null | cs.LG cs.CR cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Adversaries can embed backdoors in deep learning models by introducing
backdoor poison samples into training datasets. In this work, we investigate
how to detect such poison samples to mitigate the threat of backdoor attacks.
First, we uncover a post-hoc workflow underlying most prior work, where
defenders passively allow the attack to proceed and then leverage the
characteristics of the post-attacked model to uncover poison samples. We reveal
that this workflow does not fully exploit defenders' capabilities, and defense
pipelines built on it are prone to failure or performance degradation in many
scenarios. Second, we suggest a paradigm shift by promoting a proactive mindset
in which defenders engage proactively with the entire model training and poison
detection pipeline, directly enforcing and magnifying distinctive
characteristics of the post-attacked model to facilitate poison detection.
Based on this, we formulate a unified framework and provide practical insights
on designing detection pipelines that are more robust and generalizable. Third,
we introduce the technique of Confusion Training (CT) as a concrete
instantiation of our framework. CT applies an additional poisoning attack to
the already poisoned dataset, actively decoupling benign correlation while
exposing backdoor patterns to detection. Empirical evaluations on 4 datasets
and 14 types of attacks validate the superiority of CT over 14 baseline
defenses.
| [
{
"version": "v1",
"created": "Thu, 26 May 2022 20:44:15 GMT"
},
{
"version": "v2",
"created": "Fri, 10 Feb 2023 23:02:40 GMT"
},
{
"version": "v3",
"created": "Sun, 18 Jun 2023 02:11:20 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Qi",
"Xiangyu",
""
],
[
"Xie",
"Tinghao",
""
],
[
"Wang",
"Jiachen T.",
""
],
[
"Wu",
"Tong",
""
],
[
"Mahloujifar",
"Saeed",
""
],
[
"Mittal",
"Prateek",
""
]
] | TITLE: Towards A Proactive ML Approach for Detecting Backdoor Poison Samples
ABSTRACT: Adversaries can embed backdoors in deep learning models by introducing
backdoor poison samples into training datasets. In this work, we investigate
how to detect such poison samples to mitigate the threat of backdoor attacks.
First, we uncover a post-hoc workflow underlying most prior work, where
defenders passively allow the attack to proceed and then leverage the
characteristics of the post-attacked model to uncover poison samples. We reveal
that this workflow does not fully exploit defenders' capabilities, and defense
pipelines built on it are prone to failure or performance degradation in many
scenarios. Second, we suggest a paradigm shift by promoting a proactive mindset
in which defenders engage proactively with the entire model training and poison
detection pipeline, directly enforcing and magnifying distinctive
characteristics of the post-attacked model to facilitate poison detection.
Based on this, we formulate a unified framework and provide practical insights
on designing detection pipelines that are more robust and generalizable. Third,
we introduce the technique of Confusion Training (CT) as a concrete
instantiation of our framework. CT applies an additional poisoning attack to
the already poisoned dataset, actively decoupling benign correlation while
exposing backdoor patterns to detection. Empirical evaluations on 4 datasets
and 14 types of attacks validate the superiority of CT over 14 baseline
defenses.
|
2205.14474 | Alexander Avery | Alexander Avery, Andreas Savakis | DeepRM: Deep Recurrent Matching for 6D Pose Refinement | 9 pages, 3 figures, CVPR 2023 RHOBIN Workshop | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Precise 6D pose estimation of rigid objects from RGB images is a critical but
challenging task in robotics, augmented reality and human-computer interaction.
To address this problem, we propose DeepRM, a novel recurrent network
architecture for 6D pose refinement. DeepRM leverages initial coarse pose
estimates to render synthetic images of target objects. The rendered images are
then matched with the observed images to predict a rigid transform for updating
the previous pose estimate. This process is repeated to incrementally refine
the estimate at each iteration. The DeepRM architecture incorporates LSTM units
to propagate information through each refinement step, significantly improving
overall performance. In contrast to current 2-stage Perspective-n-Point based
solutions, DeepRM is trained end-to-end, and uses a scalable backbone that can
be tuned via a single parameter for accuracy and efficiency. During training, a
multi-scale optical flow head is added to predict the optical flow between the
observed and synthetic images. Optical flow prediction stabilizes the training
process, and enforces the learning of features that are relevant to the task of
pose estimation. Our results demonstrate that DeepRM achieves state-of-the-art
performance on two widely accepted challenging datasets.
| [
{
"version": "v1",
"created": "Sat, 28 May 2022 16:18:08 GMT"
},
{
"version": "v2",
"created": "Thu, 2 Jun 2022 15:13:05 GMT"
},
{
"version": "v3",
"created": "Tue, 7 Jun 2022 14:29:53 GMT"
},
{
"version": "v4",
"created": "Wed, 31 Aug 2022 18:36:31 GMT"
},
{
"version": "v5",
"created": "Fri, 16 Jun 2023 20:26:55 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Avery",
"Alexander",
""
],
[
"Savakis",
"Andreas",
""
]
] | TITLE: DeepRM: Deep Recurrent Matching for 6D Pose Refinement
ABSTRACT: Precise 6D pose estimation of rigid objects from RGB images is a critical but
challenging task in robotics, augmented reality and human-computer interaction.
To address this problem, we propose DeepRM, a novel recurrent network
architecture for 6D pose refinement. DeepRM leverages initial coarse pose
estimates to render synthetic images of target objects. The rendered images are
then matched with the observed images to predict a rigid transform for updating
the previous pose estimate. This process is repeated to incrementally refine
the estimate at each iteration. The DeepRM architecture incorporates LSTM units
to propagate information through each refinement step, significantly improving
overall performance. In contrast to current 2-stage Perspective-n-Point based
solutions, DeepRM is trained end-to-end, and uses a scalable backbone that can
be tuned via a single parameter for accuracy and efficiency. During training, a
multi-scale optical flow head is added to predict the optical flow between the
observed and synthetic images. Optical flow prediction stabilizes the training
process, and enforces the learning of features that are relevant to the task of
pose estimation. Our results demonstrate that DeepRM achieves state-of-the-art
performance on two widely accepted challenging datasets.
|
2205.15827 | Marnix Suilen | Marnix Suilen, Thiago D. Sim\~ao, David Parker, Nils Jansen | Robust Anytime Learning of Markov Decision Processes | Accepted at NeurIPS 2022 | null | null | null | cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Markov decision processes (MDPs) are formal models commonly used in
sequential decision-making. MDPs capture the stochasticity that may arise, for
instance, from imprecise actuators via probabilities in the transition
function. However, in data-driven applications, deriving precise probabilities
from (limited) data introduces statistical errors that may lead to unexpected
or undesirable outcomes. Uncertain MDPs (uMDPs) do not require precise
probabilities but instead use so-called uncertainty sets in the transitions,
accounting for such limited data. Tools from the formal verification community
efficiently compute robust policies that provably adhere to formal
specifications, like safety constraints, under the worst-case instance in the
uncertainty set. We continuously learn the transition probabilities of an MDP
in a robust anytime-learning approach that combines a dedicated Bayesian
inference scheme with the computation of robust policies. In particular, our
method (1) approximates probabilities as intervals, (2) adapts to new data that
may be inconsistent with an intermediate model, and (3) may be stopped at any
time to compute a robust policy on the uMDP that faithfully captures the data
so far. Furthermore, our method is capable of adapting to changes in the
environment. We show the effectiveness of our approach and compare it to robust
policies computed on uMDPs learned by the UCRL2 reinforcement learning
algorithm in an experimental evaluation on several benchmarks.
| [
{
"version": "v1",
"created": "Tue, 31 May 2022 14:29:55 GMT"
},
{
"version": "v2",
"created": "Thu, 15 Sep 2022 08:07:37 GMT"
},
{
"version": "v3",
"created": "Mon, 24 Oct 2022 13:13:34 GMT"
},
{
"version": "v4",
"created": "Mon, 19 Jun 2023 09:07:48 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Suilen",
"Marnix",
""
],
[
"Simão",
"Thiago D.",
""
],
[
"Parker",
"David",
""
],
[
"Jansen",
"Nils",
""
]
] | TITLE: Robust Anytime Learning of Markov Decision Processes
ABSTRACT: Markov decision processes (MDPs) are formal models commonly used in
sequential decision-making. MDPs capture the stochasticity that may arise, for
instance, from imprecise actuators via probabilities in the transition
function. However, in data-driven applications, deriving precise probabilities
from (limited) data introduces statistical errors that may lead to unexpected
or undesirable outcomes. Uncertain MDPs (uMDPs) do not require precise
probabilities but instead use so-called uncertainty sets in the transitions,
accounting for such limited data. Tools from the formal verification community
efficiently compute robust policies that provably adhere to formal
specifications, like safety constraints, under the worst-case instance in the
uncertainty set. We continuously learn the transition probabilities of an MDP
in a robust anytime-learning approach that combines a dedicated Bayesian
inference scheme with the computation of robust policies. In particular, our
method (1) approximates probabilities as intervals, (2) adapts to new data that
may be inconsistent with an intermediate model, and (3) may be stopped at any
time to compute a robust policy on the uMDP that faithfully captures the data
so far. Furthermore, our method is capable of adapting to changes in the
environment. We show the effectiveness of our approach and compare it to robust
policies computed on uMDPs learned by the UCRL2 reinforcement learning
algorithm in an experimental evaluation on several benchmarks.
|
2205.15987 | Wenjie Li | Wenjie Li, Qiaolin Xia, Junfeng Deng, Hao Cheng, Jiangming Liu,
Kouying Xue, Yong Cheng and Shu-Tao Xia | VFed-SSD: Towards Practical Vertical Federated Advertising | Accepted to the Trustworthy Federated Learning workshop of IJCAI2022
(FL-IJCAI22). Old version: Semi-Supervised Cross-Silo Advertising with
Partial Knowledge Transfer | null | null | null | cs.LG cs.DC cs.IR | http://creativecommons.org/licenses/by/4.0/ | As an emerging secure learning paradigm in lever-aging cross-agency private
data, vertical federatedlearning (VFL) is expected to improve advertising
models by enabling the joint learning of complementary user attributes
privately owned by the advertiser and the publisher. However, there are two key
challenges in applying it to advertising systems: a) the limited scale of
labeled overlapping samples, and b) the high cost of real-time cross-agency
serving. In this paper, we propose a semi-supervised split distillation
framework VFed-SSD to alleviate the two limitations. We identify that: i)there
are massive unlabeled overlapped data available in advertising systems, and ii)
we can keep a balance between model performance and inference cost by
decomposing the federated model. Specifically, we develop a self-supervised
task MatchedPair Detection (MPD) to exploit the vertically partitioned
unlabeled data and propose the Split Knowledge Distillation (SplitKD) schema to
avoid cross-agency serving. Empirical studies on three industrial datasets
exhibit the effectiveness of ourmethods, with the median AUC over all datasets
improved by 0.86% and 2.6% in the local andthe federated deployment mode
respectively. Overall, our framework provides an efficient federation-enhanced
solution for real-time display advertising with minimal deploying cost and
significant performance lift.
| [
{
"version": "v1",
"created": "Tue, 31 May 2022 17:45:30 GMT"
},
{
"version": "v2",
"created": "Tue, 21 Jun 2022 12:06:23 GMT"
},
{
"version": "v3",
"created": "Fri, 30 Sep 2022 18:27:05 GMT"
},
{
"version": "v4",
"created": "Sat, 17 Jun 2023 13:17:36 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Li",
"Wenjie",
""
],
[
"Xia",
"Qiaolin",
""
],
[
"Deng",
"Junfeng",
""
],
[
"Cheng",
"Hao",
""
],
[
"Liu",
"Jiangming",
""
],
[
"Xue",
"Kouying",
""
],
[
"Cheng",
"Yong",
""
],
[
"Xia",
"Shu-Tao",
""
]
] | TITLE: VFed-SSD: Towards Practical Vertical Federated Advertising
ABSTRACT: As an emerging secure learning paradigm in lever-aging cross-agency private
data, vertical federatedlearning (VFL) is expected to improve advertising
models by enabling the joint learning of complementary user attributes
privately owned by the advertiser and the publisher. However, there are two key
challenges in applying it to advertising systems: a) the limited scale of
labeled overlapping samples, and b) the high cost of real-time cross-agency
serving. In this paper, we propose a semi-supervised split distillation
framework VFed-SSD to alleviate the two limitations. We identify that: i)there
are massive unlabeled overlapped data available in advertising systems, and ii)
we can keep a balance between model performance and inference cost by
decomposing the federated model. Specifically, we develop a self-supervised
task MatchedPair Detection (MPD) to exploit the vertically partitioned
unlabeled data and propose the Split Knowledge Distillation (SplitKD) schema to
avoid cross-agency serving. Empirical studies on three industrial datasets
exhibit the effectiveness of ourmethods, with the median AUC over all datasets
improved by 0.86% and 2.6% in the local andthe federated deployment mode
respectively. Overall, our framework provides an efficient federation-enhanced
solution for real-time display advertising with minimal deploying cost and
significant performance lift.
|
2206.00356 | Kumar Abhishek | Zahra Mirikharaji, Kumar Abhishek, Alceu Bissoto, Catarina Barata,
Sandra Avila, Eduardo Valle, M. Emre Celebi, Ghassan Hamarneh | A Survey on Deep Learning for Skin Lesion Segmentation | Published in Medical Image Analysis (2023); 55 pages, 10 figures;
Mirikharaji and Abhishek: Joint first authors; Celebi and Hamarneh: Joint
senior authors | Medical Image Analysis (2023): 102863 | 10.1016/j.media.2023.102863 | null | eess.IV cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Skin cancer is a major public health problem that could benefit from
computer-aided diagnosis to reduce the burden of this common disease. Skin
lesion segmentation from images is an important step toward achieving this
goal. However, the presence of natural and artificial artifacts (e.g., hair and
air bubbles), intrinsic factors (e.g., lesion shape and contrast), and
variations in image acquisition conditions make skin lesion segmentation a
challenging task. Recently, various researchers have explored the applicability
of deep learning models to skin lesion segmentation. In this survey, we
cross-examine 177 research papers that deal with deep learning-based
segmentation of skin lesions. We analyze these works along several dimensions,
including input data (datasets, preprocessing, and synthetic data generation),
model design (architecture, modules, and losses), and evaluation aspects (data
annotation requirements and segmentation performance). We discuss these
dimensions both from the viewpoint of select seminal works, and from a
systematic viewpoint, examining how those choices have influenced current
trends, and how their limitations should be addressed. To facilitate
comparisons, we summarize all examined works in a comprehensive table as well
as an interactive table available online at
https://github.com/sfu-mial/skin-lesion-segmentation-survey.
| [
{
"version": "v1",
"created": "Wed, 1 Jun 2022 09:43:10 GMT"
},
{
"version": "v2",
"created": "Mon, 27 Feb 2023 06:58:28 GMT"
},
{
"version": "v3",
"created": "Tue, 20 Jun 2023 08:28:38 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Mirikharaji",
"Zahra",
""
],
[
"Abhishek",
"Kumar",
""
],
[
"Bissoto",
"Alceu",
""
],
[
"Barata",
"Catarina",
""
],
[
"Avila",
"Sandra",
""
],
[
"Valle",
"Eduardo",
""
],
[
"Celebi",
"M. Emre",
""
],
[
"Hamarneh",
"Ghassan",
""
]
] | TITLE: A Survey on Deep Learning for Skin Lesion Segmentation
ABSTRACT: Skin cancer is a major public health problem that could benefit from
computer-aided diagnosis to reduce the burden of this common disease. Skin
lesion segmentation from images is an important step toward achieving this
goal. However, the presence of natural and artificial artifacts (e.g., hair and
air bubbles), intrinsic factors (e.g., lesion shape and contrast), and
variations in image acquisition conditions make skin lesion segmentation a
challenging task. Recently, various researchers have explored the applicability
of deep learning models to skin lesion segmentation. In this survey, we
cross-examine 177 research papers that deal with deep learning-based
segmentation of skin lesions. We analyze these works along several dimensions,
including input data (datasets, preprocessing, and synthetic data generation),
model design (architecture, modules, and losses), and evaluation aspects (data
annotation requirements and segmentation performance). We discuss these
dimensions both from the viewpoint of select seminal works, and from a
systematic viewpoint, examining how those choices have influenced current
trends, and how their limitations should be addressed. To facilitate
comparisons, we summarize all examined works in a comprehensive table as well
as an interactive table available online at
https://github.com/sfu-mial/skin-lesion-segmentation-survey.
|
2206.06686 | Ruslan Shaydulin | Abdulkadir Canatar, Evan Peters, Cengiz Pehlevan, Stefan M. Wild,
Ruslan Shaydulin | Bandwidth Enables Generalization in Quantum Kernel Models | Accepted version | null | null | null | quant-ph cs.LG | http://creativecommons.org/licenses/by/4.0/ | Quantum computers are known to provide speedups over classical
state-of-the-art machine learning methods in some specialized settings. For
example, quantum kernel methods have been shown to provide an exponential
speedup on a learning version of the discrete logarithm problem. Understanding
the generalization of quantum models is essential to realizing similar speedups
on problems of practical interest. Recent results demonstrate that
generalization is hindered by the exponential size of the quantum feature
space. Although these results suggest that quantum models cannot generalize
when the number of qubits is large, in this paper we show that these results
rely on overly restrictive assumptions. We consider a wider class of models by
varying a hyperparameter that we call quantum kernel bandwidth. We analyze the
large-qubit limit and provide explicit formulas for the generalization of a
quantum model that can be solved in closed form. Specifically, we show that
changing the value of the bandwidth can take a model from provably not being
able to generalize to any target function to good generalization for
well-aligned targets. Our analysis shows how the bandwidth controls the
spectrum of the kernel integral operator and thereby the inductive bias of the
model. We demonstrate empirically that our theory correctly predicts how
varying the bandwidth affects generalization of quantum models on challenging
datasets, including those far outside our theoretical assumptions. We discuss
the implications of our results for quantum advantage in machine learning.
| [
{
"version": "v1",
"created": "Tue, 14 Jun 2022 08:41:08 GMT"
},
{
"version": "v2",
"created": "Wed, 15 Jun 2022 13:20:06 GMT"
},
{
"version": "v3",
"created": "Sun, 18 Jun 2023 17:21:58 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Canatar",
"Abdulkadir",
""
],
[
"Peters",
"Evan",
""
],
[
"Pehlevan",
"Cengiz",
""
],
[
"Wild",
"Stefan M.",
""
],
[
"Shaydulin",
"Ruslan",
""
]
] | TITLE: Bandwidth Enables Generalization in Quantum Kernel Models
ABSTRACT: Quantum computers are known to provide speedups over classical
state-of-the-art machine learning methods in some specialized settings. For
example, quantum kernel methods have been shown to provide an exponential
speedup on a learning version of the discrete logarithm problem. Understanding
the generalization of quantum models is essential to realizing similar speedups
on problems of practical interest. Recent results demonstrate that
generalization is hindered by the exponential size of the quantum feature
space. Although these results suggest that quantum models cannot generalize
when the number of qubits is large, in this paper we show that these results
rely on overly restrictive assumptions. We consider a wider class of models by
varying a hyperparameter that we call quantum kernel bandwidth. We analyze the
large-qubit limit and provide explicit formulas for the generalization of a
quantum model that can be solved in closed form. Specifically, we show that
changing the value of the bandwidth can take a model from provably not being
able to generalize to any target function to good generalization for
well-aligned targets. Our analysis shows how the bandwidth controls the
spectrum of the kernel integral operator and thereby the inductive bias of the
model. We demonstrate empirically that our theory correctly predicts how
varying the bandwidth affects generalization of quantum models on challenging
datasets, including those far outside our theoretical assumptions. We discuss
the implications of our results for quantum advantage in machine learning.
|
2206.07587 | Pere-Llu\'is Huguet Cabot | Abelardo Carlos Mart\'inez Lorenzo, Pere-Llu\'is Huguet Cabot, Roberto
Navigli | Cross-lingual AMR Aligner: Paying Attention to Cross-Attention | ACL 2023. Please cite authors correctly using both lastnames
("Mart\'inez Lorenzo", "Huguet Cabot") | null | null | null | cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | This paper introduces a novel aligner for Abstract Meaning Representation
(AMR) graphs that can scale cross-lingually, and is thus capable of aligning
units and spans in sentences of different languages. Our approach leverages
modern Transformer-based parsers, which inherently encode alignment information
in their cross-attention weights, allowing us to extract this information
during parsing. This eliminates the need for English-specific rules or the
Expectation Maximization (EM) algorithm that have been used in previous
approaches. In addition, we propose a guided supervised method using alignment
to further enhance the performance of our aligner. We achieve state-of-the-art
results in the benchmarks for AMR alignment and demonstrate our aligner's
ability to obtain them across multiple languages. Our code will be available at
\href{https://www.github.com/Babelscape/AMR-alignment}{github.com/Babelscape/AMR-alignment}.
| [
{
"version": "v1",
"created": "Wed, 15 Jun 2022 15:12:24 GMT"
},
{
"version": "v2",
"created": "Mon, 19 Jun 2023 16:05:35 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Lorenzo",
"Abelardo Carlos Martínez",
""
],
[
"Cabot",
"Pere-Lluís Huguet",
""
],
[
"Navigli",
"Roberto",
""
]
] | TITLE: Cross-lingual AMR Aligner: Paying Attention to Cross-Attention
ABSTRACT: This paper introduces a novel aligner for Abstract Meaning Representation
(AMR) graphs that can scale cross-lingually, and is thus capable of aligning
units and spans in sentences of different languages. Our approach leverages
modern Transformer-based parsers, which inherently encode alignment information
in their cross-attention weights, allowing us to extract this information
during parsing. This eliminates the need for English-specific rules or the
Expectation Maximization (EM) algorithm that have been used in previous
approaches. In addition, we propose a guided supervised method using alignment
to further enhance the performance of our aligner. We achieve state-of-the-art
results in the benchmarks for AMR alignment and demonstrate our aligner's
ability to obtain them across multiple languages. Our code will be available at
\href{https://www.github.com/Babelscape/AMR-alignment}{github.com/Babelscape/AMR-alignment}.
|
2207.01054 | Kristian Miok | Kristian Miok, Encarnacion Hidalgo-Tenorio, Petya Osenova,
Miguel-Angel Benitez-Castro and Marko Robnik-Sikonja | Multi-aspect Multilingual and Cross-lingual Parliamentary Speech
Analysis | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Parliamentary and legislative debate transcripts provide informative insight
into elected politicians' opinions, positions, and policy preferences. They are
interesting for political and social sciences as well as linguistics and
natural language processing (NLP) research. While existing research studied
individual parliaments, we apply advanced NLP methods to a joint and
comparative analysis of six national parliaments (Bulgarian, Czech, French,
Slovene, Spanish, and United Kingdom) between 2017 and 2020. We analyze
emotions and sentiment in the transcripts from the ParlaMint dataset collection
and assess if the age, gender, and political orientation of speakers can be
detected from their speeches. The results show some commonalities and many
surprising differences among the analyzed countries.
| [
{
"version": "v1",
"created": "Sun, 3 Jul 2022 14:31:32 GMT"
},
{
"version": "v2",
"created": "Tue, 20 Jun 2023 13:32:02 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Miok",
"Kristian",
""
],
[
"Hidalgo-Tenorio",
"Encarnacion",
""
],
[
"Osenova",
"Petya",
""
],
[
"Benitez-Castro",
"Miguel-Angel",
""
],
[
"Robnik-Sikonja",
"Marko",
""
]
] | TITLE: Multi-aspect Multilingual and Cross-lingual Parliamentary Speech
Analysis
ABSTRACT: Parliamentary and legislative debate transcripts provide informative insight
into elected politicians' opinions, positions, and policy preferences. They are
interesting for political and social sciences as well as linguistics and
natural language processing (NLP) research. While existing research studied
individual parliaments, we apply advanced NLP methods to a joint and
comparative analysis of six national parliaments (Bulgarian, Czech, French,
Slovene, Spanish, and United Kingdom) between 2017 and 2020. We analyze
emotions and sentiment in the transcripts from the ParlaMint dataset collection
and assess if the age, gender, and political orientation of speakers can be
detected from their speeches. The results show some commonalities and many
surprising differences among the analyzed countries.
|
2207.04153 | Abhinav Kumar | Abhinav Kumar, Chenhao Tan, Amit Sharma | Probing Classifiers are Unreliable for Concept Removal and Detection | null | Advances in Neural Information Processing Systems, Year 2022,
Volume 35, Pages 17994-18008 | null | null | cs.LG cs.CL | http://creativecommons.org/licenses/by/4.0/ | Neural network models trained on text data have been found to encode
undesirable linguistic or sensitive concepts in their representation. Removing
such concepts is non-trivial because of a complex relationship between the
concept, text input, and the learnt representation. Recent work has proposed
post-hoc and adversarial methods to remove such unwanted concepts from a
model's representation. Through an extensive theoretical and empirical
analysis, we show that these methods can be counter-productive: they are unable
to remove the concepts entirely, and in the worst case may end up destroying
all task-relevant features. The reason is the methods' reliance on a probing
classifier as a proxy for the concept. Even under the most favorable conditions
for learning a probing classifier when a concept's relevant features in
representation space alone can provide 100% accuracy, we prove that a probing
classifier is likely to use non-concept features and thus post-hoc or
adversarial methods will fail to remove the concept correctly. These
theoretical implications are confirmed by experiments on models trained on
synthetic, Multi-NLI, and Twitter datasets. For sensitive applications of
concept removal such as fairness, we recommend caution against using these
methods and propose a spuriousness metric to gauge the quality of the final
classifier.
| [
{
"version": "v1",
"created": "Fri, 8 Jul 2022 23:15:26 GMT"
},
{
"version": "v2",
"created": "Fri, 21 Oct 2022 11:41:30 GMT"
},
{
"version": "v3",
"created": "Mon, 19 Jun 2023 17:37:02 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Kumar",
"Abhinav",
""
],
[
"Tan",
"Chenhao",
""
],
[
"Sharma",
"Amit",
""
]
] | TITLE: Probing Classifiers are Unreliable for Concept Removal and Detection
ABSTRACT: Neural network models trained on text data have been found to encode
undesirable linguistic or sensitive concepts in their representation. Removing
such concepts is non-trivial because of a complex relationship between the
concept, text input, and the learnt representation. Recent work has proposed
post-hoc and adversarial methods to remove such unwanted concepts from a
model's representation. Through an extensive theoretical and empirical
analysis, we show that these methods can be counter-productive: they are unable
to remove the concepts entirely, and in the worst case may end up destroying
all task-relevant features. The reason is the methods' reliance on a probing
classifier as a proxy for the concept. Even under the most favorable conditions
for learning a probing classifier when a concept's relevant features in
representation space alone can provide 100% accuracy, we prove that a probing
classifier is likely to use non-concept features and thus post-hoc or
adversarial methods will fail to remove the concept correctly. These
theoretical implications are confirmed by experiments on models trained on
synthetic, Multi-NLI, and Twitter datasets. For sensitive applications of
concept removal such as fairness, we recommend caution against using these
methods and propose a spuriousness metric to gauge the quality of the final
classifier.
|
2208.01582 | Junru Gu | Junru Gu, Chenxu Hu, Tianyuan Zhang, Xuanyao Chen, Yilun Wang, Yue
Wang, Hang Zhao | ViP3D: End-to-end Visual Trajectory Prediction via 3D Agent Queries | CVPR 2023 | null | null | null | cs.CV cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Perception and prediction are two separate modules in the existing autonomous
driving systems. They interact with each other via hand-picked features such as
agent bounding boxes and trajectories. Due to this separation, prediction, as a
downstream module, only receives limited information from the perception
module. To make matters worse, errors from the perception modules can propagate
and accumulate, adversely affecting the prediction results. In this work, we
propose ViP3D, a query-based visual trajectory prediction pipeline that
exploits rich information from raw videos to directly predict future
trajectories of agents in a scene. ViP3D employs sparse agent queries to
detect, track, and predict throughout the pipeline, making it the first fully
differentiable vision-based trajectory prediction approach. Instead of using
historical feature maps and trajectories, useful information from previous
timestamps is encoded in agent queries, which makes ViP3D a concise streaming
prediction method. Furthermore, extensive experimental results on the nuScenes
dataset show the strong vision-based prediction performance of ViP3D over
traditional pipelines and previous end-to-end models.
| [
{
"version": "v1",
"created": "Tue, 2 Aug 2022 16:38:28 GMT"
},
{
"version": "v2",
"created": "Thu, 13 Oct 2022 17:05:36 GMT"
},
{
"version": "v3",
"created": "Mon, 19 Jun 2023 11:50:41 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Gu",
"Junru",
""
],
[
"Hu",
"Chenxu",
""
],
[
"Zhang",
"Tianyuan",
""
],
[
"Chen",
"Xuanyao",
""
],
[
"Wang",
"Yilun",
""
],
[
"Wang",
"Yue",
""
],
[
"Zhao",
"Hang",
""
]
] | TITLE: ViP3D: End-to-end Visual Trajectory Prediction via 3D Agent Queries
ABSTRACT: Perception and prediction are two separate modules in the existing autonomous
driving systems. They interact with each other via hand-picked features such as
agent bounding boxes and trajectories. Due to this separation, prediction, as a
downstream module, only receives limited information from the perception
module. To make matters worse, errors from the perception modules can propagate
and accumulate, adversely affecting the prediction results. In this work, we
propose ViP3D, a query-based visual trajectory prediction pipeline that
exploits rich information from raw videos to directly predict future
trajectories of agents in a scene. ViP3D employs sparse agent queries to
detect, track, and predict throughout the pipeline, making it the first fully
differentiable vision-based trajectory prediction approach. Instead of using
historical feature maps and trajectories, useful information from previous
timestamps is encoded in agent queries, which makes ViP3D a concise streaming
prediction method. Furthermore, extensive experimental results on the nuScenes
dataset show the strong vision-based prediction performance of ViP3D over
traditional pipelines and previous end-to-end models.
|
2208.02108 | Qihang Zhou | Qihang Zhou, Jiming Chen, Haoyu Liu, Shibo He, Wenchao Meng | Detecting Multivariate Time Series Anomalies with Zero Known Label | null | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multivariate time series anomaly detection has been extensively studied under
the semi-supervised setting, where a training dataset with all normal instances
is required. However, preparing such a dataset is very laborious since each
single data instance should be fully guaranteed to be normal. It is, therefore,
desired to explore multivariate time series anomaly detection methods based on
the dataset without any label knowledge. In this paper, we propose MTGFlow, an
unsupervised anomaly detection approach for multivariate time series anomaly
detection via dynamic graph and entity-aware normalizing flow, leaning only on
a widely accepted hypothesis that abnormal instances exhibit sparse densities
than the normal. However, the complex interdependencies among entities and the
diverse inherent characteristics of each entity pose significant challenges on
the density estimation, let alone to detect anomalies based on the estimated
possibility distribution. To tackle these problems, we propose to learn the
mutual and dynamic relations among entities via a graph structure learning
model, which helps to model accurate distribution of multivariate time series.
Moreover, taking account of distinct characteristics of the individual
entities, an entity-aware normalizing flow is developed to describe each entity
into a parameterized normal distribution, thereby producing fine-grained
density estimation. Incorporating these two strategies, MTGFlow achieves
superior anomaly detection performance. Experiments on five public datasets
with seven baselines are conducted, MTGFlow outperforms the SOTA methods by up
to 5.0 AUROC\%. Codes will be released at
https://github.com/zqhang/Detecting-Multivariate-Time-Series-Anomalies-with-Zero-Known-Label.
| [
{
"version": "v1",
"created": "Wed, 3 Aug 2022 14:38:19 GMT"
},
{
"version": "v2",
"created": "Wed, 30 Nov 2022 05:10:55 GMT"
},
{
"version": "v3",
"created": "Sat, 17 Jun 2023 13:12:43 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Zhou",
"Qihang",
""
],
[
"Chen",
"Jiming",
""
],
[
"Liu",
"Haoyu",
""
],
[
"He",
"Shibo",
""
],
[
"Meng",
"Wenchao",
""
]
] | TITLE: Detecting Multivariate Time Series Anomalies with Zero Known Label
ABSTRACT: Multivariate time series anomaly detection has been extensively studied under
the semi-supervised setting, where a training dataset with all normal instances
is required. However, preparing such a dataset is very laborious since each
single data instance should be fully guaranteed to be normal. It is, therefore,
desired to explore multivariate time series anomaly detection methods based on
the dataset without any label knowledge. In this paper, we propose MTGFlow, an
unsupervised anomaly detection approach for multivariate time series anomaly
detection via dynamic graph and entity-aware normalizing flow, leaning only on
a widely accepted hypothesis that abnormal instances exhibit sparse densities
than the normal. However, the complex interdependencies among entities and the
diverse inherent characteristics of each entity pose significant challenges on
the density estimation, let alone to detect anomalies based on the estimated
possibility distribution. To tackle these problems, we propose to learn the
mutual and dynamic relations among entities via a graph structure learning
model, which helps to model accurate distribution of multivariate time series.
Moreover, taking account of distinct characteristics of the individual
entities, an entity-aware normalizing flow is developed to describe each entity
into a parameterized normal distribution, thereby producing fine-grained
density estimation. Incorporating these two strategies, MTGFlow achieves
superior anomaly detection performance. Experiments on five public datasets
with seven baselines are conducted, MTGFlow outperforms the SOTA methods by up
to 5.0 AUROC\%. Codes will be released at
https://github.com/zqhang/Detecting-Multivariate-Time-Series-Anomalies-with-Zero-Known-Label.
|
2208.12118 | Satender Gunwal | Ankur Sinha, Satender Gunwal and Shivam Kumar | A Globally Convergent Gradient-based Bilevel Hyperparameter Optimization
Method | null | null | null | null | cs.LG math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Hyperparameter optimization in machine learning is often achieved using naive
techniques that only lead to an approximate set of hyperparameters. Although
techniques such as Bayesian optimization perform an intelligent search on a
given domain of hyperparameters, it does not guarantee an optimal solution. A
major drawback of most of these approaches is an exponential increase of their
search domain with number of hyperparameters, increasing the computational cost
and making the approaches slow. The hyperparameter optimization problem is
inherently a bilevel optimization task, and some studies have attempted bilevel
solution methodologies for solving this problem. However, these studies assume
a unique set of model weights that minimize the training loss, which is
generally violated by deep learning architectures. This paper discusses a
gradient-based bilevel method addressing these drawbacks for solving the
hyperparameter optimization problem. The proposed method can handle continuous
hyperparameters for which we have chosen the regularization hyperparameter in
our experiments. The method guarantees convergence to the set of optimal
hyperparameters that this study has theoretically proven. The idea is based on
approximating the lower-level optimal value function using Gaussian process
regression. As a result, the bilevel problem is reduced to a single level
constrained optimization task that is solved using the augmented Lagrangian
method. We have performed an extensive computational study on the MNIST and
CIFAR-10 datasets on multi-layer perceptron and LeNet architectures that
confirms the efficiency of the proposed method. A comparative study against
grid search, random search, Bayesian optimization, and HyberBand method on
various hyperparameter problems shows that the proposed algorithm converges
with lower computation and leads to models that generalize better on the
testing set.
| [
{
"version": "v1",
"created": "Thu, 25 Aug 2022 14:25:16 GMT"
},
{
"version": "v2",
"created": "Sun, 18 Jun 2023 12:40:24 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Sinha",
"Ankur",
""
],
[
"Gunwal",
"Satender",
""
],
[
"Kumar",
"Shivam",
""
]
] | TITLE: A Globally Convergent Gradient-based Bilevel Hyperparameter Optimization
Method
ABSTRACT: Hyperparameter optimization in machine learning is often achieved using naive
techniques that only lead to an approximate set of hyperparameters. Although
techniques such as Bayesian optimization perform an intelligent search on a
given domain of hyperparameters, it does not guarantee an optimal solution. A
major drawback of most of these approaches is an exponential increase of their
search domain with number of hyperparameters, increasing the computational cost
and making the approaches slow. The hyperparameter optimization problem is
inherently a bilevel optimization task, and some studies have attempted bilevel
solution methodologies for solving this problem. However, these studies assume
a unique set of model weights that minimize the training loss, which is
generally violated by deep learning architectures. This paper discusses a
gradient-based bilevel method addressing these drawbacks for solving the
hyperparameter optimization problem. The proposed method can handle continuous
hyperparameters for which we have chosen the regularization hyperparameter in
our experiments. The method guarantees convergence to the set of optimal
hyperparameters that this study has theoretically proven. The idea is based on
approximating the lower-level optimal value function using Gaussian process
regression. As a result, the bilevel problem is reduced to a single level
constrained optimization task that is solved using the augmented Lagrangian
method. We have performed an extensive computational study on the MNIST and
CIFAR-10 datasets on multi-layer perceptron and LeNet architectures that
confirms the efficiency of the proposed method. A comparative study against
grid search, random search, Bayesian optimization, and HyberBand method on
various hyperparameter problems shows that the proposed algorithm converges
with lower computation and leads to models that generalize better on the
testing set.
|
2209.00942 | Gabriel Kronberger | Gabriel Kronberger | Local Optimization Often is Ill-conditioned in Genetic Programming for
Symbolic Regression | Submitted to International Symposium on Symbolic and Numeric
Algorithms for Scientific Computing 2022 https://synasc.ro/ | null | 10.1109/SYNASC57785.2022.00055 | null | cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Gradient-based local optimization has been shown to improve results of
genetic programming (GP) for symbolic regression. Several state-of-the-art GP
implementations use iterative nonlinear least squares (NLS) algorithms such as
the Levenberg-Marquardt algorithm for local optimization. The effectiveness of
NLS algorithms depends on appropriate scaling and conditioning of the
optimization problem. This has so far been ignored in symbolic regression and
GP literature. In this study we use a singular value decomposition of NLS
Jacobian matrices to determine the numeric rank and the condition number. We
perform experiments with a GP implementation and six different benchmark
datasets. Our results show that rank-deficient and ill-conditioned Jacobian
matrices occur frequently and for all datasets. The issue is less extreme when
restricting GP tree size and when using many non-linear functions in the
function set.
| [
{
"version": "v1",
"created": "Fri, 2 Sep 2022 10:39:26 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Kronberger",
"Gabriel",
""
]
] | TITLE: Local Optimization Often is Ill-conditioned in Genetic Programming for
Symbolic Regression
ABSTRACT: Gradient-based local optimization has been shown to improve results of
genetic programming (GP) for symbolic regression. Several state-of-the-art GP
implementations use iterative nonlinear least squares (NLS) algorithms such as
the Levenberg-Marquardt algorithm for local optimization. The effectiveness of
NLS algorithms depends on appropriate scaling and conditioning of the
optimization problem. This has so far been ignored in symbolic regression and
GP literature. In this study we use a singular value decomposition of NLS
Jacobian matrices to determine the numeric rank and the condition number. We
perform experiments with a GP implementation and six different benchmark
datasets. Our results show that rank-deficient and ill-conditioned Jacobian
matrices occur frequently and for all datasets. The issue is less extreme when
restricting GP tree size and when using many non-linear functions in the
function set.
|
2209.02544 | Junliang Yu | Junliang Yu, Xin Xia, Tong Chen, Lizhen Cui, Nguyen Quoc Viet Hung,
Hongzhi Yin | XSimGCL: Towards Extremely Simple Graph Contrastive Learning for
Recommendation | Extension of our SIGIR'22 paper "Are Graph Augmentations Necessary?
Simple Graph Contrastive Learning for Recommendation", Accepted by TKDE | null | null | null | cs.IR | http://creativecommons.org/licenses/by/4.0/ | Contrastive learning (CL) has recently been demonstrated critical in
improving recommendation performance. The underlying principle of CL-based
recommendation models is to ensure the consistency between representations
derived from different graph augmentations of the user-item bipartite graph.
This self-supervised approach allows for the extraction of general features
from raw data, thereby mitigating the issue of data sparsity. Despite the
effectiveness of this paradigm, the factors contributing to its performance
gains have yet to be fully understood. This paper provides novel insights into
the impact of CL on recommendation. Our findings indicate that CL enables the
model to learn more evenly distributed user and item representations, which
alleviates the prevalent popularity bias and promoting long-tail items. Our
analysis also suggests that the graph augmentations, previously considered
essential, are relatively unreliable and of limited significance in CL-based
recommendation. Based on these findings, we put forward an eXtremely Simple
Graph Contrastive Learning method (XSimGCL) for recommendation, which discards
the ineffective graph augmentations and instead employs a simple yet effective
noise-based embedding augmentation to generate views for CL. A comprehensive
experimental study on four large and highly sparse benchmark datasets
demonstrates that, though the proposed method is extremely simple, it can
smoothly adjust the uniformity of learned representations and outperforms its
graph augmentation-based counterparts by a large margin in both recommendation
accuracy and training efficiency. The code and used datasets are released at
https://github.com/Coder-Yu/SELFRec.
| [
{
"version": "v1",
"created": "Tue, 6 Sep 2022 14:50:51 GMT"
},
{
"version": "v2",
"created": "Thu, 8 Sep 2022 23:06:47 GMT"
},
{
"version": "v3",
"created": "Tue, 11 Apr 2023 11:33:11 GMT"
},
{
"version": "v4",
"created": "Mon, 19 Jun 2023 01:33:54 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Yu",
"Junliang",
""
],
[
"Xia",
"Xin",
""
],
[
"Chen",
"Tong",
""
],
[
"Cui",
"Lizhen",
""
],
[
"Hung",
"Nguyen Quoc Viet",
""
],
[
"Yin",
"Hongzhi",
""
]
] | TITLE: XSimGCL: Towards Extremely Simple Graph Contrastive Learning for
Recommendation
ABSTRACT: Contrastive learning (CL) has recently been demonstrated critical in
improving recommendation performance. The underlying principle of CL-based
recommendation models is to ensure the consistency between representations
derived from different graph augmentations of the user-item bipartite graph.
This self-supervised approach allows for the extraction of general features
from raw data, thereby mitigating the issue of data sparsity. Despite the
effectiveness of this paradigm, the factors contributing to its performance
gains have yet to be fully understood. This paper provides novel insights into
the impact of CL on recommendation. Our findings indicate that CL enables the
model to learn more evenly distributed user and item representations, which
alleviates the prevalent popularity bias and promoting long-tail items. Our
analysis also suggests that the graph augmentations, previously considered
essential, are relatively unreliable and of limited significance in CL-based
recommendation. Based on these findings, we put forward an eXtremely Simple
Graph Contrastive Learning method (XSimGCL) for recommendation, which discards
the ineffective graph augmentations and instead employs a simple yet effective
noise-based embedding augmentation to generate views for CL. A comprehensive
experimental study on four large and highly sparse benchmark datasets
demonstrates that, though the proposed method is extremely simple, it can
smoothly adjust the uniformity of learned representations and outperforms its
graph augmentation-based counterparts by a large margin in both recommendation
accuracy and training efficiency. The code and used datasets are released at
https://github.com/Coder-Yu/SELFRec.
|
2209.07857 | Hao Cheng | Hao Cheng, Mengmeng Liu, Lin Chen, Hellward Broszio, Monika Sester,
Michael Ying Yang | GATraj: A Graph- and Attention-based Multi-Agent Trajectory Prediction
Model | null | null | null | null | cs.RO cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Trajectory prediction has been a long-standing problem in intelligent systems
like autonomous driving and robot navigation. Models trained on large-scale
benchmarks have made significant progress in improving prediction accuracy.
However, the importance on efficiency for real-time applications has been less
emphasized. This paper proposes an attention-based graph model, named GATraj,
which achieves a good balance of prediction accuracy and inference speed. We
use attention mechanisms to model the spatial-temporal dynamics of agents, such
as pedestrians or vehicles, and a graph convolutional network to model their
interactions. Additionally, a Laplacian mixture decoder is implemented to
mitigate mode collapse and generate diverse multimodal predictions for each
agent. GATraj achieves state-of-the-art prediction performance at a much higher
speed when tested on the ETH/UCY datasets for pedestrian trajectories, and good
performance at about 100 Hz inference speed when tested on the nuScenes dataset
for autonomous driving. We conduct extensive experiments to analyze the
probability estimation of the Laplacian mixture decoder and compare it with a
Gaussian mixture decoder for predicting different multimodalities. Furthermore,
comprehensive ablation studies demonstrate the effectiveness of each proposed
module in GATraj. The code is released at
https://github.com/mengmengliu1998/GATraj.
| [
{
"version": "v1",
"created": "Fri, 16 Sep 2022 11:29:19 GMT"
},
{
"version": "v2",
"created": "Mon, 19 Jun 2023 13:05:02 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Cheng",
"Hao",
""
],
[
"Liu",
"Mengmeng",
""
],
[
"Chen",
"Lin",
""
],
[
"Broszio",
"Hellward",
""
],
[
"Sester",
"Monika",
""
],
[
"Yang",
"Michael Ying",
""
]
] | TITLE: GATraj: A Graph- and Attention-based Multi-Agent Trajectory Prediction
Model
ABSTRACT: Trajectory prediction has been a long-standing problem in intelligent systems
like autonomous driving and robot navigation. Models trained on large-scale
benchmarks have made significant progress in improving prediction accuracy.
However, the importance on efficiency for real-time applications has been less
emphasized. This paper proposes an attention-based graph model, named GATraj,
which achieves a good balance of prediction accuracy and inference speed. We
use attention mechanisms to model the spatial-temporal dynamics of agents, such
as pedestrians or vehicles, and a graph convolutional network to model their
interactions. Additionally, a Laplacian mixture decoder is implemented to
mitigate mode collapse and generate diverse multimodal predictions for each
agent. GATraj achieves state-of-the-art prediction performance at a much higher
speed when tested on the ETH/UCY datasets for pedestrian trajectories, and good
performance at about 100 Hz inference speed when tested on the nuScenes dataset
for autonomous driving. We conduct extensive experiments to analyze the
probability estimation of the Laplacian mixture decoder and compare it with a
Gaussian mixture decoder for predicting different multimodalities. Furthermore,
comprehensive ablation studies demonstrate the effectiveness of each proposed
module in GATraj. The code is released at
https://github.com/mengmengliu1998/GATraj.
|
2209.09722 | Orlando Amaral Cejas | Orlando Amaral, Muhammad Ilyas Azeem, Sallam Abualhaija and Lionel C
Briand | NLP-based Automated Compliance Checking of Data Processing Agreements
against GDPR | 26 pages, 5 figures, 10 tables, 1 Algorithm, Accepted for publication
in TSE | null | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Processing personal data is regulated in Europe by the General Data
Protection Regulation (GDPR) through data processing agreements (DPAs).
Checking the compliance of DPAs contributes to the compliance verification of
software systems as DPAs are an important source of requirements for software
development involving the processing of personal data. However, manually
checking whether a given DPA complies with GDPR is challenging as it requires
significant time and effort for understanding and identifying DPA-relevant
compliance requirements in GDPR and then verifying these requirements in the
DPA. In this paper, we propose an automated solution to check the compliance of
a given DPA against GDPR. In close interaction with legal experts, we first
built two artifacts: (i) the "shall" requirements extracted from the GDPR
provisions relevant to DPA compliance and (ii) a glossary table defining the
legal concepts in the requirements. Then, we developed an automated solution
that leverages natural language processing (NLP) technologies to check the
compliance of a given DPA against these "shall" requirements. Specifically, our
approach automatically generates phrasal-level representations for the textual
content of the DPA and compares it against predefined representations of the
"shall" requirements. Over a dataset of 30 actual DPAs, the approach correctly
finds 618 out of 750 genuine violations while raising 76 false violations, and
further correctly identifies 524 satisfied requirements. The approach has thus
an average precision of 89.1%, a recall of 82.4%, and an accuracy of 84.6%.
Compared to a baseline that relies on off-the-shelf NLP tools, our approach
provides an average accuracy gain of ~20 percentage points. The accuracy of our
approach can be improved to ~94% with limited manual verification effort.
| [
{
"version": "v1",
"created": "Tue, 20 Sep 2022 13:50:58 GMT"
},
{
"version": "v2",
"created": "Sun, 18 Jun 2023 12:59:12 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Amaral",
"Orlando",
""
],
[
"Azeem",
"Muhammad Ilyas",
""
],
[
"Abualhaija",
"Sallam",
""
],
[
"Briand",
"Lionel C",
""
]
] | TITLE: NLP-based Automated Compliance Checking of Data Processing Agreements
against GDPR
ABSTRACT: Processing personal data is regulated in Europe by the General Data
Protection Regulation (GDPR) through data processing agreements (DPAs).
Checking the compliance of DPAs contributes to the compliance verification of
software systems as DPAs are an important source of requirements for software
development involving the processing of personal data. However, manually
checking whether a given DPA complies with GDPR is challenging as it requires
significant time and effort for understanding and identifying DPA-relevant
compliance requirements in GDPR and then verifying these requirements in the
DPA. In this paper, we propose an automated solution to check the compliance of
a given DPA against GDPR. In close interaction with legal experts, we first
built two artifacts: (i) the "shall" requirements extracted from the GDPR
provisions relevant to DPA compliance and (ii) a glossary table defining the
legal concepts in the requirements. Then, we developed an automated solution
that leverages natural language processing (NLP) technologies to check the
compliance of a given DPA against these "shall" requirements. Specifically, our
approach automatically generates phrasal-level representations for the textual
content of the DPA and compares it against predefined representations of the
"shall" requirements. Over a dataset of 30 actual DPAs, the approach correctly
finds 618 out of 750 genuine violations while raising 76 false violations, and
further correctly identifies 524 satisfied requirements. The approach has thus
an average precision of 89.1%, a recall of 82.4%, and an accuracy of 84.6%.
Compared to a baseline that relies on off-the-shelf NLP tools, our approach
provides an average accuracy gain of ~20 percentage points. The accuracy of our
approach can be improved to ~94% with limited manual verification effort.
|
2209.12221 | Chenglong Li | Chenglong Li, Qiwen Zhu, Tubiao Liu, Jin Tang, and Yu Su | Hand Hygiene Assessment via Joint Step Segmentation and Key Action
Scorer | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Hand hygiene is a standard six-step hand-washing action proposed by the World
Health Organization (WHO). However, there is no good way to supervise medical
staff to do hand hygiene, which brings the potential risk of disease spread.
Existing action assessment works usually make an overall quality prediction on
an entire video. However, the internal structures of hand hygiene action are
important in hand hygiene assessment. Therefore, we propose a novel
fine-grained learning framework to perform step segmentation and key action
scorer in a joint manner for accurate hand hygiene assessment. Existing
temporal segmentation methods usually employ multi-stage convolutional network
to improve the segmentation robustness, but easily lead to over-segmentation
due to the lack of the long-range dependence. To address this issue, we design
a multi-stage convolution-transformer network for step segmentation. Based on
the observation that each hand-washing step involves several key actions which
determine the hand-washing quality, we design a set of key action scorers to
evaluate the quality of key actions in each step. In addition, there lacks a
unified dataset in hand hygiene assessment. Therefore, under the supervision of
medical staff, we contribute a video dataset that contains 300 video sequences
with fine-grained annotations. Extensive experiments on the dataset suggest
that our method well assesses hand hygiene videos and achieves outstanding
performance.
| [
{
"version": "v1",
"created": "Sun, 25 Sep 2022 13:47:21 GMT"
},
{
"version": "v2",
"created": "Tue, 18 Oct 2022 11:50:30 GMT"
},
{
"version": "v3",
"created": "Wed, 16 Nov 2022 16:06:28 GMT"
},
{
"version": "v4",
"created": "Tue, 20 Jun 2023 09:23:11 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Li",
"Chenglong",
""
],
[
"Zhu",
"Qiwen",
""
],
[
"Liu",
"Tubiao",
""
],
[
"Tang",
"Jin",
""
],
[
"Su",
"Yu",
""
]
] | TITLE: Hand Hygiene Assessment via Joint Step Segmentation and Key Action
Scorer
ABSTRACT: Hand hygiene is a standard six-step hand-washing action proposed by the World
Health Organization (WHO). However, there is no good way to supervise medical
staff to do hand hygiene, which brings the potential risk of disease spread.
Existing action assessment works usually make an overall quality prediction on
an entire video. However, the internal structures of hand hygiene action are
important in hand hygiene assessment. Therefore, we propose a novel
fine-grained learning framework to perform step segmentation and key action
scorer in a joint manner for accurate hand hygiene assessment. Existing
temporal segmentation methods usually employ multi-stage convolutional network
to improve the segmentation robustness, but easily lead to over-segmentation
due to the lack of the long-range dependence. To address this issue, we design
a multi-stage convolution-transformer network for step segmentation. Based on
the observation that each hand-washing step involves several key actions which
determine the hand-washing quality, we design a set of key action scorers to
evaluate the quality of key actions in each step. In addition, there lacks a
unified dataset in hand hygiene assessment. Therefore, under the supervision of
medical staff, we contribute a video dataset that contains 300 video sequences
with fine-grained annotations. Extensive experiments on the dataset suggest
that our method well assesses hand hygiene videos and achieves outstanding
performance.
|
2209.13012 | Guilherme Alves | Guilherme Alves, Fabien Bernier, Miguel Couceiro, Karima Makhlouf,
Catuscia Palamidessi, Sami Zhioua | Survey on Fairness Notions and Related Tensions | null | null | null | null | cs.CY cs.AI cs.LG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Automated decision systems are increasingly used to take consequential
decisions in problems such as job hiring and loan granting with the hope of
replacing subjective human decisions with objective machine learning (ML)
algorithms. However, ML-based decision systems are prone to bias, which results
in yet unfair decisions. Several notions of fairness have been defined in the
literature to capture the different subtleties of this ethical and social
concept (e.g., statistical parity, equal opportunity, etc.). Fairness
requirements to be satisfied while learning models created several types of
tensions among the different notions of fairness and other desirable properties
such as privacy and classification accuracy. This paper surveys the commonly
used fairness notions and discusses the tensions among them with privacy and
accuracy. Different methods to address the fairness-accuracy trade-off
(classified into four approaches, namely, pre-processing, in-processing,
post-processing, and hybrid) are reviewed. The survey is consolidated with
experimental analysis carried out on fairness benchmark datasets to illustrate
the relationship between fairness measures and accuracy in real-world
scenarios.
| [
{
"version": "v1",
"created": "Fri, 16 Sep 2022 13:36:05 GMT"
},
{
"version": "v2",
"created": "Mon, 19 Jun 2023 11:19:52 GMT"
}
] | 2023-06-21T00:00:00 | [
[
"Alves",
"Guilherme",
""
],
[
"Bernier",
"Fabien",
""
],
[
"Couceiro",
"Miguel",
""
],
[
"Makhlouf",
"Karima",
""
],
[
"Palamidessi",
"Catuscia",
""
],
[
"Zhioua",
"Sami",
""
]
] | TITLE: Survey on Fairness Notions and Related Tensions
ABSTRACT: Automated decision systems are increasingly used to take consequential
decisions in problems such as job hiring and loan granting with the hope of
replacing subjective human decisions with objective machine learning (ML)
algorithms. However, ML-based decision systems are prone to bias, which results
in yet unfair decisions. Several notions of fairness have been defined in the
literature to capture the different subtleties of this ethical and social
concept (e.g., statistical parity, equal opportunity, etc.). Fairness
requirements to be satisfied while learning models created several types of
tensions among the different notions of fairness and other desirable properties
such as privacy and classification accuracy. This paper surveys the commonly
used fairness notions and discusses the tensions among them with privacy and
accuracy. Different methods to address the fairness-accuracy trade-off
(classified into four approaches, namely, pre-processing, in-processing,
post-processing, and hybrid) are reviewed. The survey is consolidated with
experimental analysis carried out on fairness benchmark datasets to illustrate
the relationship between fairness measures and accuracy in real-world
scenarios.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.