id
stringlengths 9
10
| submitter
stringlengths 1
64
⌀ | authors
stringlengths 4
20.7k
| title
stringlengths 4
246
| comments
stringlengths 1
523
⌀ | journal-ref
stringlengths 4
404
⌀ | doi
stringlengths 11
153
⌀ | report-no
stringlengths 2
254
⌀ | categories
stringlengths 5
98
| license
stringclasses 9
values | orig_abstract
stringlengths 14
3.35k
| versions
listlengths 1
60
| update_date
stringlengths 10
10
| authors_parsed
listlengths 1
1.35k
| abstract
stringlengths 11
3.34k
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1207.5742
|
Andrei Romashchenko
|
Tarik Kaced and Andrei Romashchenko
|
Conditional Information Inequalities for Entropic and Almost Entropic
Points
|
Submitted to the IEEE Transactions on Information Theory
|
IEEE Transactions on Information Theory 59(11), 2013, pp.
7149-7167
|
10.1109/TIT.2013.2274614
| null |
cs.IT cs.DM math.IT math.PR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We study conditional linear information inequalities, i.e., linear
inequalities for Shannon entropy that hold for distributions whose entropies
meet some linear constraints. We prove that some conditional information
inequalities cannot be extended to any unconditional linear inequalities. Some
of these conditional inequalities hold for almost entropic points, while others
do not. We also discuss some counterparts of conditional information
inequalities for Kolmogorov complexity.
|
[
{
"created": "Tue, 24 Jul 2012 16:31:05 GMT",
"version": "v1"
},
{
"created": "Sun, 29 Jul 2012 19:20:14 GMT",
"version": "v2"
},
{
"created": "Mon, 22 Jul 2013 16:46:19 GMT",
"version": "v3"
},
{
"created": "Fri, 16 Aug 2013 09:17:51 GMT",
"version": "v4"
}
] |
2013-10-30
|
[
[
"Kaced",
"Tarik",
""
],
[
"Romashchenko",
"Andrei",
""
]
] |
We study conditional linear information inequalities, i.e., linear inequalities for Shannon entropy that hold for distributions whose entropies meet some linear constraints. We prove that some conditional information inequalities cannot be extended to any unconditional linear inequalities. Some of these conditional inequalities hold for almost entropic points, while others do not. We also discuss some counterparts of conditional information inequalities for Kolmogorov complexity.
|
2008.11714
|
Jia-Bin Huang
|
Chen Gao, Jiarui Xu, Yuliang Zou, Jia-Bin Huang
|
DRG: Dual Relation Graph for Human-Object Interaction Detection
|
ECCV 2020. Project: http://chengao.vision/DRG/ Code:
https://github.com/vt-vl-lab/DRG
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We tackle the challenging problem of human-object interaction (HOI)
detection. Existing methods either recognize the interaction of each
human-object pair in isolation or perform joint inference based on complex
appearance-based features. In this paper, we leverage an abstract
spatial-semantic representation to describe each human-object pair and
aggregate the contextual information of the scene via a dual relation graph
(one human-centric and one object-centric). Our proposed dual relation graph
effectively captures discriminative cues from the scene to resolve ambiguity
from local predictions. Our model is conceptually simple and leads to favorable
results compared to the state-of-the-art HOI detection algorithms on two
large-scale benchmark datasets.
|
[
{
"created": "Wed, 26 Aug 2020 17:59:40 GMT",
"version": "v1"
}
] |
2020-08-27
|
[
[
"Gao",
"Chen",
""
],
[
"Xu",
"Jiarui",
""
],
[
"Zou",
"Yuliang",
""
],
[
"Huang",
"Jia-Bin",
""
]
] |
We tackle the challenging problem of human-object interaction (HOI) detection. Existing methods either recognize the interaction of each human-object pair in isolation or perform joint inference based on complex appearance-based features. In this paper, we leverage an abstract spatial-semantic representation to describe each human-object pair and aggregate the contextual information of the scene via a dual relation graph (one human-centric and one object-centric). Our proposed dual relation graph effectively captures discriminative cues from the scene to resolve ambiguity from local predictions. Our model is conceptually simple and leads to favorable results compared to the state-of-the-art HOI detection algorithms on two large-scale benchmark datasets.
|
2407.09033
|
Byeonghyun Pak
|
Byeonghyun Pak, Byeongju Woo, Sunghwan Kim, Dae-hwan Kim, Hoseong Kim
|
Textual Query-Driven Mask Transformer for Domain Generalized
Segmentation
|
ECCV 2024
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this paper, we introduce a method to tackle Domain Generalized Semantic
Segmentation (DGSS) by utilizing domain-invariant semantic knowledge from text
embeddings of vision-language models. We employ the text embeddings as object
queries within a transformer-based segmentation framework (textual object
queries). These queries are regarded as a domain-invariant basis for pixel
grouping in DGSS. To leverage the power of textual object queries, we introduce
a novel framework named the textual query-driven mask transformer (tqdm). Our
tqdm aims to (1) generate textual object queries that maximally encode
domain-invariant semantics and (2) enhance the semantic clarity of dense visual
features. Additionally, we suggest three regularization losses to improve the
efficacy of tqdm by aligning between visual and textual features. By utilizing
our method, the model can comprehend inherent semantic information for classes
of interest, enabling it to generalize to extreme domains (e.g., sketch style).
Our tqdm achieves 68.9 mIoU on GTA5$\rightarrow$Cityscapes, outperforming the
prior state-of-the-art method by 2.5 mIoU. The project page is available at
https://byeonghyunpak.github.io/tqdm.
|
[
{
"created": "Fri, 12 Jul 2024 06:49:16 GMT",
"version": "v1"
},
{
"created": "Wed, 31 Jul 2024 14:27:06 GMT",
"version": "v2"
}
] |
2024-08-01
|
[
[
"Pak",
"Byeonghyun",
""
],
[
"Woo",
"Byeongju",
""
],
[
"Kim",
"Sunghwan",
""
],
[
"Kim",
"Dae-hwan",
""
],
[
"Kim",
"Hoseong",
""
]
] |
In this paper, we introduce a method to tackle Domain Generalized Semantic Segmentation (DGSS) by utilizing domain-invariant semantic knowledge from text embeddings of vision-language models. We employ the text embeddings as object queries within a transformer-based segmentation framework (textual object queries). These queries are regarded as a domain-invariant basis for pixel grouping in DGSS. To leverage the power of textual object queries, we introduce a novel framework named the textual query-driven mask transformer (tqdm). Our tqdm aims to (1) generate textual object queries that maximally encode domain-invariant semantics and (2) enhance the semantic clarity of dense visual features. Additionally, we suggest three regularization losses to improve the efficacy of tqdm by aligning between visual and textual features. By utilizing our method, the model can comprehend inherent semantic information for classes of interest, enabling it to generalize to extreme domains (e.g., sketch style). Our tqdm achieves 68.9 mIoU on GTA5$\rightarrow$Cityscapes, outperforming the prior state-of-the-art method by 2.5 mIoU. The project page is available at https://byeonghyunpak.github.io/tqdm.
|
1812.10998
|
Preeti Gopal Ms.
|
Preeti Gopal and Sharat Chandran and Imants Svalbe and Ajit Rajwade
|
Learning from past scans: Tomographic reconstruction to detect new
structures
|
5 pages, 8 figures, 1 table
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The need for tomographic reconstruction from sparse measurements arises when
the measurement process is potentially harmful, needs to be rapid, or is
uneconomical. In such cases, prior information from previous longitudinal scans
of the same or similar objects helps to reconstruct the current object whilst
requiring significantly fewer `updating' measurements. However, a significant
limitation of all prior-based methods is the possible dominance of the prior
over the reconstruction of new localised information that has evolved within
the test object. In this paper, we improve the state of the art by (1)
detecting potential regions where new changes may have occurred, and (2)
effectively reconstructing both the old and new structures by computing
regional weights that moderate the local influence of the priors. We have
tested the efficacy of our method on synthetic as well as real volume data. The
results demonstrate that using weighted priors significantly improves the
overall quality of the reconstructed data whilst minimising their impact on
regions that contain new information.
|
[
{
"created": "Sun, 23 Dec 2018 09:45:15 GMT",
"version": "v1"
}
] |
2018-12-31
|
[
[
"Gopal",
"Preeti",
""
],
[
"Chandran",
"Sharat",
""
],
[
"Svalbe",
"Imants",
""
],
[
"Rajwade",
"Ajit",
""
]
] |
The need for tomographic reconstruction from sparse measurements arises when the measurement process is potentially harmful, needs to be rapid, or is uneconomical. In such cases, prior information from previous longitudinal scans of the same or similar objects helps to reconstruct the current object whilst requiring significantly fewer `updating' measurements. However, a significant limitation of all prior-based methods is the possible dominance of the prior over the reconstruction of new localised information that has evolved within the test object. In this paper, we improve the state of the art by (1) detecting potential regions where new changes may have occurred, and (2) effectively reconstructing both the old and new structures by computing regional weights that moderate the local influence of the priors. We have tested the efficacy of our method on synthetic as well as real volume data. The results demonstrate that using weighted priors significantly improves the overall quality of the reconstructed data whilst minimising their impact on regions that contain new information.
|
2310.19848
|
Lenart Treven
|
Lenart Treven, Jonas H\"ubotter, Bhavya Sukhija, Florian D\"orfler,
Andreas Krause
|
Efficient Exploration in Continuous-time Model-based Reinforcement
Learning
| null | null | null | null |
cs.LG cs.RO math.OC
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Reinforcement learning algorithms typically consider discrete-time dynamics,
even though the underlying systems are often continuous in time. In this paper,
we introduce a model-based reinforcement learning algorithm that represents
continuous-time dynamics using nonlinear ordinary differential equations
(ODEs). We capture epistemic uncertainty using well-calibrated probabilistic
models, and use the optimistic principle for exploration. Our regret bounds
surface the importance of the measurement selection strategy(MSS), since in
continuous time we not only must decide how to explore, but also when to
observe the underlying system. Our analysis demonstrates that the regret is
sublinear when modeling ODEs with Gaussian Processes (GP) for common choices of
MSS, such as equidistant sampling. Additionally, we propose an adaptive,
data-dependent, practical MSS that, when combined with GP dynamics, also
achieves sublinear regret with significantly fewer samples. We showcase the
benefits of continuous-time modeling over its discrete-time counterpart, as
well as our proposed adaptive MSS over standard baselines, on several
applications.
|
[
{
"created": "Mon, 30 Oct 2023 15:04:40 GMT",
"version": "v1"
}
] |
2023-11-01
|
[
[
"Treven",
"Lenart",
""
],
[
"Hübotter",
"Jonas",
""
],
[
"Sukhija",
"Bhavya",
""
],
[
"Dörfler",
"Florian",
""
],
[
"Krause",
"Andreas",
""
]
] |
Reinforcement learning algorithms typically consider discrete-time dynamics, even though the underlying systems are often continuous in time. In this paper, we introduce a model-based reinforcement learning algorithm that represents continuous-time dynamics using nonlinear ordinary differential equations (ODEs). We capture epistemic uncertainty using well-calibrated probabilistic models, and use the optimistic principle for exploration. Our regret bounds surface the importance of the measurement selection strategy(MSS), since in continuous time we not only must decide how to explore, but also when to observe the underlying system. Our analysis demonstrates that the regret is sublinear when modeling ODEs with Gaussian Processes (GP) for common choices of MSS, such as equidistant sampling. Additionally, we propose an adaptive, data-dependent, practical MSS that, when combined with GP dynamics, also achieves sublinear regret with significantly fewer samples. We showcase the benefits of continuous-time modeling over its discrete-time counterpart, as well as our proposed adaptive MSS over standard baselines, on several applications.
|
2306.09109
|
Kevis-Kokitsi Maninis
|
Varun Jampani, Kevis-Kokitsi Maninis, Andreas Engelhardt, Arjun
Karpur, Karen Truong, Kyle Sargent, Stefan Popov, Andr\'e Araujo, Ricardo
Martin-Brualla, Kaushal Patel, Daniel Vlasic, Vittorio Ferrari, Ameesh
Makadia, Ce Liu, Yuanzhen Li, Howard Zhou
|
NAVI: Category-Agnostic Image Collections with High-Quality 3D Shape and
Pose Annotations
|
NeurIPS 2023 camera ready. Project page:
https://navidataset.github.io
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Recent advances in neural reconstruction enable high-quality 3D object
reconstruction from casually captured image collections. Current techniques
mostly analyze their progress on relatively simple image collections where
Structure-from-Motion (SfM) techniques can provide ground-truth (GT) camera
poses. We note that SfM techniques tend to fail on in-the-wild image
collections such as image search results with varying backgrounds and
illuminations. To enable systematic research progress on 3D reconstruction from
casual image captures, we propose NAVI: a new dataset of category-agnostic
image collections of objects with high-quality 3D scans along with per-image
2D-3D alignments providing near-perfect GT camera parameters. These 2D-3D
alignments allow us to extract accurate derivative annotations such as dense
pixel correspondences, depth and segmentation maps. We demonstrate the use of
NAVI image collections on different problem settings and show that NAVI enables
more thorough evaluations that were not possible with existing datasets. We
believe NAVI is beneficial for systematic research progress on 3D
reconstruction and correspondence estimation. Project page:
https://navidataset.github.io
|
[
{
"created": "Thu, 15 Jun 2023 13:11:30 GMT",
"version": "v1"
},
{
"created": "Fri, 13 Oct 2023 16:12:32 GMT",
"version": "v2"
}
] |
2023-10-16
|
[
[
"Jampani",
"Varun",
""
],
[
"Maninis",
"Kevis-Kokitsi",
""
],
[
"Engelhardt",
"Andreas",
""
],
[
"Karpur",
"Arjun",
""
],
[
"Truong",
"Karen",
""
],
[
"Sargent",
"Kyle",
""
],
[
"Popov",
"Stefan",
""
],
[
"Araujo",
"André",
""
],
[
"Martin-Brualla",
"Ricardo",
""
],
[
"Patel",
"Kaushal",
""
],
[
"Vlasic",
"Daniel",
""
],
[
"Ferrari",
"Vittorio",
""
],
[
"Makadia",
"Ameesh",
""
],
[
"Liu",
"Ce",
""
],
[
"Li",
"Yuanzhen",
""
],
[
"Zhou",
"Howard",
""
]
] |
Recent advances in neural reconstruction enable high-quality 3D object reconstruction from casually captured image collections. Current techniques mostly analyze their progress on relatively simple image collections where Structure-from-Motion (SfM) techniques can provide ground-truth (GT) camera poses. We note that SfM techniques tend to fail on in-the-wild image collections such as image search results with varying backgrounds and illuminations. To enable systematic research progress on 3D reconstruction from casual image captures, we propose NAVI: a new dataset of category-agnostic image collections of objects with high-quality 3D scans along with per-image 2D-3D alignments providing near-perfect GT camera parameters. These 2D-3D alignments allow us to extract accurate derivative annotations such as dense pixel correspondences, depth and segmentation maps. We demonstrate the use of NAVI image collections on different problem settings and show that NAVI enables more thorough evaluations that were not possible with existing datasets. We believe NAVI is beneficial for systematic research progress on 3D reconstruction and correspondence estimation. Project page: https://navidataset.github.io
|
1610.05726
|
Amol Patwardhan
|
Amol Patwardhan
|
Structured Unit Testable Templated Code for Efficient Code Review
Process
|
13 pages, 12 figures
| null | null | null |
cs.SE
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Modern software development teams are distributed across onsite and off-shore
locations. Each team has developers with varying experience levels and English
communication skills. In such a diverse development environment it is important
to maintain the software quality, coding standards, timely delivery of features
and bug fixes. It is also important to reduce testing effort, minimize side
effects such as change in functionality, user experience or application
performance. Code reviews are intended to control code quality. Unfortunately,
many projects lack enforcement of processes and standards because of
approaching deadlines, live production issues and lack of resource
availability. This study examines a novel structured, unit testable templated
code method to enforce code review standards with an intent to reduce coding
effort, minimize revisions and eliminate functional and performance side
effects on the system. The proposed method would also result in unit-testable
code that can also be easily rolled back and increase team productivity. The
baseline for traditional code review processes using metrics such as code
review duration, bug regression rate, revision count was measured. These
metrics were then compared with results from the proposed code review process
that used structured unit testable templated code. The performance on 2 large
enterprise level applications spanning over 2 years and 9 feature and
maintenance release cycles was evaluated. The structured unit testable
templated code method resulted in a decrease in total code review time,
revision count and coding effort. It also decreased the number of live
production issues caused by code churn or side effects of bug fix when compared
to traditional code review process.
|
[
{
"created": "Tue, 9 Aug 2016 05:26:21 GMT",
"version": "v1"
}
] |
2016-10-19
|
[
[
"Patwardhan",
"Amol",
""
]
] |
Modern software development teams are distributed across onsite and off-shore locations. Each team has developers with varying experience levels and English communication skills. In such a diverse development environment it is important to maintain the software quality, coding standards, timely delivery of features and bug fixes. It is also important to reduce testing effort, minimize side effects such as change in functionality, user experience or application performance. Code reviews are intended to control code quality. Unfortunately, many projects lack enforcement of processes and standards because of approaching deadlines, live production issues and lack of resource availability. This study examines a novel structured, unit testable templated code method to enforce code review standards with an intent to reduce coding effort, minimize revisions and eliminate functional and performance side effects on the system. The proposed method would also result in unit-testable code that can also be easily rolled back and increase team productivity. The baseline for traditional code review processes using metrics such as code review duration, bug regression rate, revision count was measured. These metrics were then compared with results from the proposed code review process that used structured unit testable templated code. The performance on 2 large enterprise level applications spanning over 2 years and 9 feature and maintenance release cycles was evaluated. The structured unit testable templated code method resulted in a decrease in total code review time, revision count and coding effort. It also decreased the number of live production issues caused by code churn or side effects of bug fix when compared to traditional code review process.
|
2212.07720
|
Benny Kimelfeld
|
Majd Khalil, Benny Kimelfeld
|
The Complexity of the Shapley Value for Regular Path Queries
| null | null | null | null |
cs.DB
|
http://creativecommons.org/licenses/by/4.0/
|
A path query extracts vertex tuples from a labeled graph, based on the words
that are formed by the paths connecting the vertices. We study the
computational complexity of measuring the contribution of edges and vertices to
an answer to a path query, focusing on the class of conjunctive regular path
queries. To measure this contribution, we adopt the traditional Shapley value
from cooperative game theory. This value has been recently proposed and studied
in the context of relational database queries and has uses in a plethora of
other domains.
We first study the contribution of edges and show that the exact Shapley
value is almost always hard to compute. Specifically, it is #P-hard to
calculate the contribution of an edge whenever at least one (non-redundant)
conjunct allows for a word of length three or more. In the case of regular path
queries (i.e., no conjunction), the problem is tractable if the query has only
words of length at most two; hence, this property fully characterizes the
tractability of the problem. On the other hand, if we allow for an
approximation error, then it is straightforward to obtain an efficient scheme
(FPRAS) for an additive approximation. Yet, a multiplicative approximation is
harder to obtain. We establish that in the case of conjunctive regular path
queries, a multiplicative approximation of the Shapley value of an edge can be
computed in polynomial time if and only if all query atoms are finite languages
(assuming non-redundancy and conventional complexity limitations). We also
study the analogous situation where we wish to determine the contribution of a
vertex, rather than an edge, and establish complexity results of similar
nature.
|
[
{
"created": "Thu, 15 Dec 2022 10:55:04 GMT",
"version": "v1"
}
] |
2022-12-16
|
[
[
"Khalil",
"Majd",
""
],
[
"Kimelfeld",
"Benny",
""
]
] |
A path query extracts vertex tuples from a labeled graph, based on the words that are formed by the paths connecting the vertices. We study the computational complexity of measuring the contribution of edges and vertices to an answer to a path query, focusing on the class of conjunctive regular path queries. To measure this contribution, we adopt the traditional Shapley value from cooperative game theory. This value has been recently proposed and studied in the context of relational database queries and has uses in a plethora of other domains. We first study the contribution of edges and show that the exact Shapley value is almost always hard to compute. Specifically, it is #P-hard to calculate the contribution of an edge whenever at least one (non-redundant) conjunct allows for a word of length three or more. In the case of regular path queries (i.e., no conjunction), the problem is tractable if the query has only words of length at most two; hence, this property fully characterizes the tractability of the problem. On the other hand, if we allow for an approximation error, then it is straightforward to obtain an efficient scheme (FPRAS) for an additive approximation. Yet, a multiplicative approximation is harder to obtain. We establish that in the case of conjunctive regular path queries, a multiplicative approximation of the Shapley value of an edge can be computed in polynomial time if and only if all query atoms are finite languages (assuming non-redundancy and conventional complexity limitations). We also study the analogous situation where we wish to determine the contribution of a vertex, rather than an edge, and establish complexity results of similar nature.
|
1206.0430
|
Richard Southwell
|
Richard Southwell, Jianwei Huang, Biying Shou
|
Congestion Games on Weighted Directed Graphs, with Applications to
Spectrum Sharing
| null | null | null | null |
cs.NI cs.GT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
With the advance of complex large-scale networks, it is becoming increasingly
important to understand how selfish and spatially distributed individuals will
share network resources without centralized coordinations. In this paper, we
introduce the graphical congestion game with weighted edges (GCGWE) as a
general theoretical model to study this problem. In GCGWE, we view the players
as vertices in a weighted graph. The amount of negative impact (e.g.
congestion) caused by two close-by players to each other is determined by the
weight of the edge linking them. The GCGWE unifies and significantly
generalizes several simpler models considered in the previous literature, and
is well suited for modeling a wide range of networking scenarios. One good
example is to use the GCGWE to model spectrum sharing in wireless networks,
where we can properly define the edge weights and payoff functions to capture
the rather complicated interference relationship between wireless nodes. By
identifying which GCGWEs possess pure Nash equilibria and the very desirable
finite improvement property, we gain insight into when spatially distributed
wireless nodes will be able to self-organize into a mutually acceptable
resource allocation. We also consider the efficiency of the pure Nash
equilibria, and the computational complexity of finding them.
|
[
{
"created": "Sun, 3 Jun 2012 10:57:16 GMT",
"version": "v1"
}
] |
2012-06-05
|
[
[
"Southwell",
"Richard",
""
],
[
"Huang",
"Jianwei",
""
],
[
"Shou",
"Biying",
""
]
] |
With the advance of complex large-scale networks, it is becoming increasingly important to understand how selfish and spatially distributed individuals will share network resources without centralized coordinations. In this paper, we introduce the graphical congestion game with weighted edges (GCGWE) as a general theoretical model to study this problem. In GCGWE, we view the players as vertices in a weighted graph. The amount of negative impact (e.g. congestion) caused by two close-by players to each other is determined by the weight of the edge linking them. The GCGWE unifies and significantly generalizes several simpler models considered in the previous literature, and is well suited for modeling a wide range of networking scenarios. One good example is to use the GCGWE to model spectrum sharing in wireless networks, where we can properly define the edge weights and payoff functions to capture the rather complicated interference relationship between wireless nodes. By identifying which GCGWEs possess pure Nash equilibria and the very desirable finite improvement property, we gain insight into when spatially distributed wireless nodes will be able to self-organize into a mutually acceptable resource allocation. We also consider the efficiency of the pure Nash equilibria, and the computational complexity of finding them.
|
1906.07887
|
Kedar Tatwawadi
|
Kedar Tatwawadi, Shubham Chandak
|
Tutorial on algebraic deletion correction codes
| null | null | null | null |
cs.DS cs.IT eess.SP math.IT
|
http://creativecommons.org/licenses/by/4.0/
|
The deletion channel is known to be a notoriously diffcult channel to design
error-correction codes for. In spite of this difficulty, there are some
beautiful code constructions which give some intuition about the channel and
about what good deletion codes look like. In this tutorial we will take a look
at some of them. This document is a transcript of my talk at the coding theory
reading group on some interesting works on deletion channel. It is not intended
to be an exhaustive survey of works on deletion channel, but more as a tutorial
to some of the important and cute ideas in this area. For a comprehensive
survey, we refer the reader to the cited sources and surveys.
We also provide an implementation of VT codes that correct single
insertion/deletion errors for general alphabets at
https://github.com/shubhamchandak94/VT_codes/.
|
[
{
"created": "Wed, 19 Jun 2019 02:56:11 GMT",
"version": "v1"
}
] |
2019-06-20
|
[
[
"Tatwawadi",
"Kedar",
""
],
[
"Chandak",
"Shubham",
""
]
] |
The deletion channel is known to be a notoriously diffcult channel to design error-correction codes for. In spite of this difficulty, there are some beautiful code constructions which give some intuition about the channel and about what good deletion codes look like. In this tutorial we will take a look at some of them. This document is a transcript of my talk at the coding theory reading group on some interesting works on deletion channel. It is not intended to be an exhaustive survey of works on deletion channel, but more as a tutorial to some of the important and cute ideas in this area. For a comprehensive survey, we refer the reader to the cited sources and surveys. We also provide an implementation of VT codes that correct single insertion/deletion errors for general alphabets at https://github.com/shubhamchandak94/VT_codes/.
|
2403.10484
|
Rosana Montes
|
Rosana Montes and Liliana Herrera and Emilio Crisol
|
Moodle Usability Assessment Methodology using the Universal Design for
Learning perspective
|
preprint second version
| null | null | null |
cs.CY
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
The application of the Universal Design for Learning framework favors the
creation of virtual educational environments for all. It requires developing
accessible content, having a usable platform, and the use of flexible didactics
and evaluations that promote constant student motivation. The present study
aims to design a methodology to evaluate the usability of the Moodle platform
based on the principles of Universal Design for Learning, recognizing the
importance of accessibility, usability and the availability of Assistive
Technologies. We developed and applied a methodology to assess the usability
level of Moodle platforms, taking into consideration that they integrate
Assistive Technologies or are used for MOOC contexts. We provide the results of
a use case that assesses two instances for the respective Moodle v.2.x and
v.3.x family versions. We employed the framework of mixed design research in
order to assess a MOOC-type educational program devised under the principles of
Universal Design for Learning. As a result of the assessment of Moodle v.2.x
and v.3.x, we conclude that the platforms must improve some key elements (e.g.
contrasting colors, incorporation of alternative text and links) in order to
comply with international accessibility standards. With respect to usability,
we can confirm that the principles and guidelines of Universal Design for
Learning are applicable to MOOC-type Virtual Learning Environments, are
positively valued by students, and have a positive impact on certification
rates.
|
[
{
"created": "Fri, 15 Mar 2024 17:19:04 GMT",
"version": "v1"
},
{
"created": "Tue, 2 Apr 2024 15:25:51 GMT",
"version": "v2"
}
] |
2024-04-03
|
[
[
"Montes",
"Rosana",
""
],
[
"Herrera",
"Liliana",
""
],
[
"Crisol",
"Emilio",
""
]
] |
The application of the Universal Design for Learning framework favors the creation of virtual educational environments for all. It requires developing accessible content, having a usable platform, and the use of flexible didactics and evaluations that promote constant student motivation. The present study aims to design a methodology to evaluate the usability of the Moodle platform based on the principles of Universal Design for Learning, recognizing the importance of accessibility, usability and the availability of Assistive Technologies. We developed and applied a methodology to assess the usability level of Moodle platforms, taking into consideration that they integrate Assistive Technologies or are used for MOOC contexts. We provide the results of a use case that assesses two instances for the respective Moodle v.2.x and v.3.x family versions. We employed the framework of mixed design research in order to assess a MOOC-type educational program devised under the principles of Universal Design for Learning. As a result of the assessment of Moodle v.2.x and v.3.x, we conclude that the platforms must improve some key elements (e.g. contrasting colors, incorporation of alternative text and links) in order to comply with international accessibility standards. With respect to usability, we can confirm that the principles and guidelines of Universal Design for Learning are applicable to MOOC-type Virtual Learning Environments, are positively valued by students, and have a positive impact on certification rates.
|
2112.13310
|
Wenchi Ma
|
Wenchi Ma, Tianxiao Zhang, Guanghui Wang
|
Miti-DETR: Object Detection based on Transformers with Mitigatory
Self-Attention Convergence
| null |
AAAI 2022 workshop
| null | null |
cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
Object Detection with Transformers (DETR) and related works reach or even
surpass the highly-optimized Faster-RCNN baseline with self-attention network
architectures. Inspired by the evidence that pure self-attention possesses a
strong inductive bias that leads to the transformer losing the expressive power
with respect to network depth, we propose a transformer architecture with a
mitigatory self-attention mechanism by applying possible direct mapping
connections in the transformer architecture to mitigate the rank collapse so as
to counteract feature expression loss and enhance the model performance. We
apply this proposal in object detection tasks and develop a model named
Miti-DETR. Miti-DETR reserves the inputs of each single attention layer to the
outputs of that layer so that the "non-attention" information has participated
in any attention propagation. The formed residual self-attention network
addresses two critical issues: (1) stop the self-attention networks from
degenerating to rank-1 to the maximized degree; and (2) further diversify the
path distribution of parameter update so that easier attention learning is
expected. Miti-DETR significantly enhances the average detection precision and
convergence speed towards existing DETR-based models on the challenging COCO
object detection dataset. Moreover, the proposed transformer with the residual
self-attention network can be easily generalized or plugged in other related
task models without specific customization.
|
[
{
"created": "Sun, 26 Dec 2021 03:23:59 GMT",
"version": "v1"
}
] |
2021-12-28
|
[
[
"Ma",
"Wenchi",
""
],
[
"Zhang",
"Tianxiao",
""
],
[
"Wang",
"Guanghui",
""
]
] |
Object Detection with Transformers (DETR) and related works reach or even surpass the highly-optimized Faster-RCNN baseline with self-attention network architectures. Inspired by the evidence that pure self-attention possesses a strong inductive bias that leads to the transformer losing the expressive power with respect to network depth, we propose a transformer architecture with a mitigatory self-attention mechanism by applying possible direct mapping connections in the transformer architecture to mitigate the rank collapse so as to counteract feature expression loss and enhance the model performance. We apply this proposal in object detection tasks and develop a model named Miti-DETR. Miti-DETR reserves the inputs of each single attention layer to the outputs of that layer so that the "non-attention" information has participated in any attention propagation. The formed residual self-attention network addresses two critical issues: (1) stop the self-attention networks from degenerating to rank-1 to the maximized degree; and (2) further diversify the path distribution of parameter update so that easier attention learning is expected. Miti-DETR significantly enhances the average detection precision and convergence speed towards existing DETR-based models on the challenging COCO object detection dataset. Moreover, the proposed transformer with the residual self-attention network can be easily generalized or plugged in other related task models without specific customization.
|
1311.6264
|
J\"urgen M\"unch
|
Frank Elberzhager, Alla Rosbach, J\"urgen M\"unch, Robert Eschbach
|
Inspection and Test Process Integration Based on Explicit Test
Prioritization Strategies
|
12 pages. The final publication is available at
http://link.springer.com/chapter/10.1007%2F978-3-642-27213-4_12
|
Proceedings of the Software Quality Days (SWQD), pages 181-192,
Vienna, Austria, January 17-19 2012
|
10.1007/978-3-642-27213-4_12
| null |
cs.SE
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Today's software quality assurance techniques are often applied in isolation.
Consequently, synergies resulting from systematically integrating different
quality assurance activities are often not exploited. Such combinations promise
benefits, such as a reduction in quality assurance effort or higher defect
detection rates. The integration of inspection and testing, for instance, can
be used to guide testing activities. For example, testing activities can be
focused on defect-prone parts based upon inspection results. Existing
approaches for predicting defect-prone parts do not make systematic use of the
results from inspections. This article gives an overview of an integrated
inspection and testing approach, and presents a preliminary case study aiming
at verifying a study design for evaluating the approach. First results from
this preliminary case study indicate that synergies resulting from the
integration of inspection and testing might exist, and show a trend that
testing activities could be guided based on inspection results.
|
[
{
"created": "Mon, 25 Nov 2013 11:14:07 GMT",
"version": "v1"
}
] |
2013-11-26
|
[
[
"Elberzhager",
"Frank",
""
],
[
"Rosbach",
"Alla",
""
],
[
"Münch",
"Jürgen",
""
],
[
"Eschbach",
"Robert",
""
]
] |
Today's software quality assurance techniques are often applied in isolation. Consequently, synergies resulting from systematically integrating different quality assurance activities are often not exploited. Such combinations promise benefits, such as a reduction in quality assurance effort or higher defect detection rates. The integration of inspection and testing, for instance, can be used to guide testing activities. For example, testing activities can be focused on defect-prone parts based upon inspection results. Existing approaches for predicting defect-prone parts do not make systematic use of the results from inspections. This article gives an overview of an integrated inspection and testing approach, and presents a preliminary case study aiming at verifying a study design for evaluating the approach. First results from this preliminary case study indicate that synergies resulting from the integration of inspection and testing might exist, and show a trend that testing activities could be guided based on inspection results.
|
2306.10153
|
Komal Teru
|
Komal K. Teru
|
Semi-supervised Relation Extraction via Data Augmentation and
Consistency-training
|
Previously published at INTERPOLATE @ NeurIPS 2022 workshop
|
Proceedings of the 17th Conference of the European Chapter of the
Association for Computational Linguistics, 2023, 1112--1124
| null | null |
cs.CL cs.IR
|
http://creativecommons.org/licenses/by/4.0/
|
Due to the semantic complexity of the Relation extraction (RE) task,
obtaining high-quality human labelled data is an expensive and noisy process.
To improve the sample efficiency of the models, semi-supervised learning (SSL)
methods aim to leverage unlabelled data in addition to learning from limited
labelled data points. Recently, strong data augmentation combined with
consistency-based semi-supervised learning methods have advanced the state of
the art in several SSL tasks. However, adapting these methods to the RE task
has been challenging due to the difficulty of data augmentation for RE. In this
work, we leverage the recent advances in controlled text generation to perform
high quality data augmentation for the RE task. We further introduce small but
significant changes to model architecture that allows for generation of more
training data by interpolating different data points in their latent space.
These data augmentations along with consistency training result in very
competitive results for semi-supervised relation extraction on four benchmark
datasets.
|
[
{
"created": "Fri, 16 Jun 2023 19:45:42 GMT",
"version": "v1"
}
] |
2023-06-21
|
[
[
"Teru",
"Komal K.",
""
]
] |
Due to the semantic complexity of the Relation extraction (RE) task, obtaining high-quality human labelled data is an expensive and noisy process. To improve the sample efficiency of the models, semi-supervised learning (SSL) methods aim to leverage unlabelled data in addition to learning from limited labelled data points. Recently, strong data augmentation combined with consistency-based semi-supervised learning methods have advanced the state of the art in several SSL tasks. However, adapting these methods to the RE task has been challenging due to the difficulty of data augmentation for RE. In this work, we leverage the recent advances in controlled text generation to perform high quality data augmentation for the RE task. We further introduce small but significant changes to model architecture that allows for generation of more training data by interpolating different data points in their latent space. These data augmentations along with consistency training result in very competitive results for semi-supervised relation extraction on four benchmark datasets.
|
2403.10988
|
Li-Yuan Tsao
|
Li-Yuan Tsao, Yi-Chen Lo, Chia-Che Chang, Hao-Wei Chen, Roy Tseng,
Chien Feng, Chun-Yi Lee
|
Boosting Flow-based Generative Super-Resolution Models via Learned Prior
|
Accepted to CVPR2024
| null | null | null |
cs.CV cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
Flow-based super-resolution (SR) models have demonstrated astonishing
capabilities in generating high-quality images. However, these methods
encounter several challenges during image generation, such as grid artifacts,
exploding inverses, and suboptimal results due to a fixed sampling temperature.
To overcome these issues, this work introduces a conditional learned prior to
the inference phase of a flow-based SR model. This prior is a latent code
predicted by our proposed latent module conditioned on the low-resolution
image, which is then transformed by the flow model into an SR image. Our
framework is designed to seamlessly integrate with any contemporary flow-based
SR model without modifying its architecture or pre-trained weights. We evaluate
the effectiveness of our proposed framework through extensive experiments and
ablation analyses. The proposed framework successfully addresses all the
inherent issues in flow-based SR models and enhances their performance in
various SR scenarios. Our code is available at:
https://github.com/liyuantsao/BFSR
|
[
{
"created": "Sat, 16 Mar 2024 18:04:12 GMT",
"version": "v1"
},
{
"created": "Sat, 30 Mar 2024 04:56:05 GMT",
"version": "v2"
},
{
"created": "Wed, 29 May 2024 03:12:58 GMT",
"version": "v3"
}
] |
2024-05-30
|
[
[
"Tsao",
"Li-Yuan",
""
],
[
"Lo",
"Yi-Chen",
""
],
[
"Chang",
"Chia-Che",
""
],
[
"Chen",
"Hao-Wei",
""
],
[
"Tseng",
"Roy",
""
],
[
"Feng",
"Chien",
""
],
[
"Lee",
"Chun-Yi",
""
]
] |
Flow-based super-resolution (SR) models have demonstrated astonishing capabilities in generating high-quality images. However, these methods encounter several challenges during image generation, such as grid artifacts, exploding inverses, and suboptimal results due to a fixed sampling temperature. To overcome these issues, this work introduces a conditional learned prior to the inference phase of a flow-based SR model. This prior is a latent code predicted by our proposed latent module conditioned on the low-resolution image, which is then transformed by the flow model into an SR image. Our framework is designed to seamlessly integrate with any contemporary flow-based SR model without modifying its architecture or pre-trained weights. We evaluate the effectiveness of our proposed framework through extensive experiments and ablation analyses. The proposed framework successfully addresses all the inherent issues in flow-based SR models and enhances their performance in various SR scenarios. Our code is available at: https://github.com/liyuantsao/BFSR
|
2008.09607
|
Magnus Lie Hetland
|
Magnus Lie Hetland
|
Optimal Metric Search Is Equivalent to the Minimum Dominating Set
Problem
| null | null |
10.1007/978-3-030-60936-8_9
| null |
cs.DS
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In metric search, worst-case analysis is of little value, as the search
invariably degenerates to a linear scan for ill-behaved data. Consequently,
much effort has been expended on more nuanced descriptions of what performance
might in fact be attainable, including heuristic baselines like the AESA
family, as well as statistical proxies such as intrinsic dimensionality. This
paper gets to the heart of the matter with an exact characterization of the
best performance actually achievable for any given data set and query.
Specifically, linear-time objective-preserving reductions are established in
both directions between optimal metric search and the minimum dominating set
problem, whose greedy approximation becomes the equivalent of an oracle-based
AESA, repeatedly selecting the pivot that eliminates the most of the remaining
points. As an illustration, the AESA heuristic is adapted to downplay the role
of previously eliminated points, yielding some modest performance improvements
over the original, as well as its younger relative iAESA2.
|
[
{
"created": "Fri, 21 Aug 2020 17:59:41 GMT",
"version": "v1"
}
] |
2020-11-03
|
[
[
"Hetland",
"Magnus Lie",
""
]
] |
In metric search, worst-case analysis is of little value, as the search invariably degenerates to a linear scan for ill-behaved data. Consequently, much effort has been expended on more nuanced descriptions of what performance might in fact be attainable, including heuristic baselines like the AESA family, as well as statistical proxies such as intrinsic dimensionality. This paper gets to the heart of the matter with an exact characterization of the best performance actually achievable for any given data set and query. Specifically, linear-time objective-preserving reductions are established in both directions between optimal metric search and the minimum dominating set problem, whose greedy approximation becomes the equivalent of an oracle-based AESA, repeatedly selecting the pivot that eliminates the most of the remaining points. As an illustration, the AESA heuristic is adapted to downplay the role of previously eliminated points, yielding some modest performance improvements over the original, as well as its younger relative iAESA2.
|
2406.10602
|
Daniil Gurgurov
|
Daniil Gurgurov, Tanja B\"aumel, Tatiana Anikina
|
Multilingual Large Language Models and Curse of Multilinguality
| null | null |
10.48550/arXiv.2406.10602
| null |
cs.CL
|
http://creativecommons.org/licenses/by-sa/4.0/
|
Multilingual Large Language Models (LLMs) have gained large popularity among
Natural Language Processing (NLP) researchers and practitioners. These models,
trained on huge datasets, show proficiency across various languages and
demonstrate effectiveness in numerous downstream tasks. This paper navigates
the landscape of multilingual LLMs, providing an introductory overview of their
technical aspects. It explains underlying architectures, objective functions,
pre-training data sources, and tokenization methods. This work explores the
unique features of different model types: encoder-only (mBERT, XLM-R),
decoder-only (XGLM, PALM, BLOOM, GPT-3), and encoder-decoder models (mT5,
mBART). Additionally, it addresses one of the significant limitations of
multilingual LLMs - the curse of multilinguality - and discusses current
attempts to overcome it.
|
[
{
"created": "Sat, 15 Jun 2024 11:31:39 GMT",
"version": "v1"
}
] |
2024-07-02
|
[
[
"Gurgurov",
"Daniil",
""
],
[
"Bäumel",
"Tanja",
""
],
[
"Anikina",
"Tatiana",
""
]
] |
Multilingual Large Language Models (LLMs) have gained large popularity among Natural Language Processing (NLP) researchers and practitioners. These models, trained on huge datasets, show proficiency across various languages and demonstrate effectiveness in numerous downstream tasks. This paper navigates the landscape of multilingual LLMs, providing an introductory overview of their technical aspects. It explains underlying architectures, objective functions, pre-training data sources, and tokenization methods. This work explores the unique features of different model types: encoder-only (mBERT, XLM-R), decoder-only (XGLM, PALM, BLOOM, GPT-3), and encoder-decoder models (mT5, mBART). Additionally, it addresses one of the significant limitations of multilingual LLMs - the curse of multilinguality - and discusses current attempts to overcome it.
|
2407.15570
|
Zehra Yigit
|
Zehra Yigit, Ertugrul Basar
|
Hybrid STAR-RIS Enabled Integrated Sensing and Communication
|
10 pages, 7 figures
| null | null | null |
cs.IT eess.SP math.IT
|
http://creativecommons.org/licenses/by/4.0/
|
Integrated sensing and communication (ISAC) is recognized as one of the key
enabling technologies for sixth-generation (6G) wireless communication
networks, facilitating diverse emerging applications and services in an energy
and cost-efficient manner. This paper proposes a multi-user multi-target ISAC
system to enable full-space coverage for communication and sensing tasks. The
proposed system employs a hybrid simultaneous transmission and reflection
reconfigurable intelligent surface (STAR-RIS) comprising active transmissive
and passive reflective elements. In the proposed scheme, the passive reflective
elements support communication and sensing links for nearby communication users
and sensing targets, while low-power active transmissive elements are deployed
to improve sensing performance and overcome high path attenuation due to
multi-hop transmission for remote targets. Moreover, to optimize the
transmissive/reflective coefficients of the hybrid STAR-RIS, a semi-definite
relaxation (SDR)-based algorithm is proposed. Furthermore, to evaluate sensing
performance, signal-to-interference-noise ratio (SINR) and Cramer-Rao bound
(CRB) metrics have been derived and investigated via conducting extensive
computer simulations.
|
[
{
"created": "Mon, 22 Jul 2024 11:55:39 GMT",
"version": "v1"
}
] |
2024-07-23
|
[
[
"Yigit",
"Zehra",
""
],
[
"Basar",
"Ertugrul",
""
]
] |
Integrated sensing and communication (ISAC) is recognized as one of the key enabling technologies for sixth-generation (6G) wireless communication networks, facilitating diverse emerging applications and services in an energy and cost-efficient manner. This paper proposes a multi-user multi-target ISAC system to enable full-space coverage for communication and sensing tasks. The proposed system employs a hybrid simultaneous transmission and reflection reconfigurable intelligent surface (STAR-RIS) comprising active transmissive and passive reflective elements. In the proposed scheme, the passive reflective elements support communication and sensing links for nearby communication users and sensing targets, while low-power active transmissive elements are deployed to improve sensing performance and overcome high path attenuation due to multi-hop transmission for remote targets. Moreover, to optimize the transmissive/reflective coefficients of the hybrid STAR-RIS, a semi-definite relaxation (SDR)-based algorithm is proposed. Furthermore, to evaluate sensing performance, signal-to-interference-noise ratio (SINR) and Cramer-Rao bound (CRB) metrics have been derived and investigated via conducting extensive computer simulations.
|
2007.15538
|
Filippo Gabriele Prattic\`o
|
F. Gabriele Prattic\`o, Fabrizio Lamberti (Politecnico di Torino)
|
Mixed-Reality Robotic Games: Design Guidelines for Effective
Entertainment with Consumer Robots
|
This paper is accepted for inclusion in future issue of IEEE Consumer
Electronic Magazine. Copyright IEEE 2020
|
IEEE Consumer Electronics Magazine, vol. 10, no. 1, pp. 6-16, Jan.
2021
|
10.1109/MCE.2020.2988578
| null |
cs.HC cs.GR cs.RO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In recent years, there has been an increasing interest in the use of robotic
technology at home. A number of service robots appeared on the market,
supporting customers in the execution of everyday tasks. Roughly at the same
time, consumer level robots started to be used also as toys or gaming
companions. However, gaming possibilities provided by current off-the-shelf
robotic products are generally quite limited, and this fact makes them quickly
loose their attractiveness. A way that has been proven capable to boost robotic
gaming and related devices consists in creating playful experiences in which
physical and digital elements are combined together using Mixed Reality
technologies. However, these games differ significantly from digital- or
physical only experiences, and new design principles are required to support
developers in their creative work. This papers addresses such need, by drafting
a set of guidelines which summarize developments carried out by the research
community and their findings.
|
[
{
"created": "Thu, 30 Jul 2020 15:47:17 GMT",
"version": "v1"
}
] |
2020-12-08
|
[
[
"Pratticò",
"F. Gabriele",
"",
"Politecnico di Torino"
],
[
"Lamberti",
"Fabrizio",
"",
"Politecnico di Torino"
]
] |
In recent years, there has been an increasing interest in the use of robotic technology at home. A number of service robots appeared on the market, supporting customers in the execution of everyday tasks. Roughly at the same time, consumer level robots started to be used also as toys or gaming companions. However, gaming possibilities provided by current off-the-shelf robotic products are generally quite limited, and this fact makes them quickly loose their attractiveness. A way that has been proven capable to boost robotic gaming and related devices consists in creating playful experiences in which physical and digital elements are combined together using Mixed Reality technologies. However, these games differ significantly from digital- or physical only experiences, and new design principles are required to support developers in their creative work. This papers addresses such need, by drafting a set of guidelines which summarize developments carried out by the research community and their findings.
|
1603.00536
|
EPTCS
|
C\'esar A. Mu\~noz (NASA Langley Research Center), Jorge A. P\'erez
(University of Groningen)
|
Proceedings of the Eleventh International Workshop on Developments in
Computational Models
| null |
EPTCS 204, 2016
|
10.4204/EPTCS.204
| null |
cs.LO cs.PL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
This volume contains the proceedings of DCM 2015, the 11th International
Workshop on Developments in Computational Models held on October 28, 2015 in
Cali, Colombia. DCM 2015 was organized as a one-day satellite event of the 12th
International Colloquium on Theoretical Aspects of Computing (ICTAC 2015).
Several new models of computation have emerged in the last few years, and
many developments of traditional computational models have been proposed with
the aim of taking into account the new demands of computer systems users and
the new capabilities of computation engines. A new computational model, or a
new feature in a traditional one, usually is reflected in a new family of
programming languages, and new paradigms of software development.
The aim of the DCM workshop series is to bring together researchers who are
currently developing new computational models or new features for traditional
computational models, in order to foster their interaction, to provide a forum
for presenting new ideas and work in progress, and to enable newcomers to learn
about current activities in this area. Topics of interest include all abstract
models of computation and their applications to the development of programming
languages and systems.
|
[
{
"created": "Wed, 2 Mar 2016 00:49:28 GMT",
"version": "v1"
}
] |
2016-03-03
|
[
[
"Muñoz",
"César A.",
"",
"NASA Langley Research Center"
],
[
"Pérez",
"Jorge A.",
"",
"University of Groningen"
]
] |
This volume contains the proceedings of DCM 2015, the 11th International Workshop on Developments in Computational Models held on October 28, 2015 in Cali, Colombia. DCM 2015 was organized as a one-day satellite event of the 12th International Colloquium on Theoretical Aspects of Computing (ICTAC 2015). Several new models of computation have emerged in the last few years, and many developments of traditional computational models have been proposed with the aim of taking into account the new demands of computer systems users and the new capabilities of computation engines. A new computational model, or a new feature in a traditional one, usually is reflected in a new family of programming languages, and new paradigms of software development. The aim of the DCM workshop series is to bring together researchers who are currently developing new computational models or new features for traditional computational models, in order to foster their interaction, to provide a forum for presenting new ideas and work in progress, and to enable newcomers to learn about current activities in this area. Topics of interest include all abstract models of computation and their applications to the development of programming languages and systems.
|
2405.00846
|
Duy Nguyen
|
Duy P. Nguyen and Kai-Chieh Hsu and Wenhao Yu and Jie Tan and Jaime F.
Fisac
|
Gameplay Filters: Safe Robot Walking through Adversarial Imagination
| null | null | null | null |
cs.RO cs.LG
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Ensuring the safe operation of legged robots in uncertain, novel environments
is crucial to their widespread adoption. Despite recent advances in safety
filters that can keep arbitrary task-driven policies from incurring safety
failures, existing solutions for legged robot locomotion still rely on
simplified dynamics and may fail when the robot is perturbed away from
predefined stable gaits. This paper presents a general approach that leverages
offline game-theoretic reinforcement learning to synthesize a highly robust
safety filter for high-order nonlinear dynamics. This gameplay filter then
maintains runtime safety by continually simulating adversarial futures and
precluding task-driven actions that would cause it to lose future games (and
thereby violate safety). Validated on a 36-dimensional quadruped robot
locomotion task, the gameplay safety filter exhibits inherent robustness to the
sim-to-real gap without manual tuning or heuristic designs. Physical
experiments demonstrate the effectiveness of the gameplay safety filter under
perturbations, such as tugging and unmodeled irregular terrains, while
simulation studies shed light on how to trade off computation and
conservativeness without compromising safety.
|
[
{
"created": "Wed, 1 May 2024 20:21:44 GMT",
"version": "v1"
},
{
"created": "Fri, 31 May 2024 14:26:47 GMT",
"version": "v2"
}
] |
2024-06-03
|
[
[
"Nguyen",
"Duy P.",
""
],
[
"Hsu",
"Kai-Chieh",
""
],
[
"Yu",
"Wenhao",
""
],
[
"Tan",
"Jie",
""
],
[
"Fisac",
"Jaime F.",
""
]
] |
Ensuring the safe operation of legged robots in uncertain, novel environments is crucial to their widespread adoption. Despite recent advances in safety filters that can keep arbitrary task-driven policies from incurring safety failures, existing solutions for legged robot locomotion still rely on simplified dynamics and may fail when the robot is perturbed away from predefined stable gaits. This paper presents a general approach that leverages offline game-theoretic reinforcement learning to synthesize a highly robust safety filter for high-order nonlinear dynamics. This gameplay filter then maintains runtime safety by continually simulating adversarial futures and precluding task-driven actions that would cause it to lose future games (and thereby violate safety). Validated on a 36-dimensional quadruped robot locomotion task, the gameplay safety filter exhibits inherent robustness to the sim-to-real gap without manual tuning or heuristic designs. Physical experiments demonstrate the effectiveness of the gameplay safety filter under perturbations, such as tugging and unmodeled irregular terrains, while simulation studies shed light on how to trade off computation and conservativeness without compromising safety.
|
2206.11728
|
Gunnar Kudrjavets
|
Gunnar Kudrjavets (University of Groningen), Jeff Thomas (Meta
Platforms, Inc.), Aditya Kumar (Snap, Inc.), Nachiappan Nagappan (Meta
Platforms, Inc.), and Ayushi Rastogi (University of Groningen)
|
There Ain't No Such Thing as a Free Custom Memory Allocator
|
4 pages. To be published in 38th IEEE International Conference on
Software Maintenance and Evolution (ICSME 2022), Oct 3-7, 2022, Limassol,
Cyprus
| null |
10.1109/ICSME55016.2022.00079
| null |
cs.SE
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Using custom memory allocators is an efficient performance optimization
technique. However, dependency on a custom allocator can introduce several
maintenance-related issues. We present lessons learned from the industry and
provide critical guidance for using custom memory allocators and enumerate
various challenges associated with integrating them. These recommendations are
based on years of experience incorporating custom allocators into different
industrial software projects.
|
[
{
"created": "Thu, 23 Jun 2022 14:26:50 GMT",
"version": "v1"
}
] |
2022-12-23
|
[
[
"Kudrjavets",
"Gunnar",
"",
"University of Groningen"
],
[
"Thomas",
"Jeff",
"",
"Meta\n Platforms, Inc."
],
[
"Kumar",
"Aditya",
"",
"Snap, Inc."
],
[
"Nagappan",
"Nachiappan",
"",
"Meta\n Platforms, Inc."
],
[
"Rastogi",
"Ayushi",
"",
"University of Groningen"
]
] |
Using custom memory allocators is an efficient performance optimization technique. However, dependency on a custom allocator can introduce several maintenance-related issues. We present lessons learned from the industry and provide critical guidance for using custom memory allocators and enumerate various challenges associated with integrating them. These recommendations are based on years of experience incorporating custom allocators into different industrial software projects.
|
2407.10943
|
Hanqing Wang
|
Hanqing Wang, Jiahe Chen, Wensi Huang, Qingwei Ben, Tai Wang, Boyu Mi,
Tao Huang, Siheng Zhao, Yilun Chen, Sizhe Yang, Peizhou Cao, Wenye Yu, Zichao
Ye, Jialun Li, Junfeng Long, Zirui Wang, Huiling Wang, Ying Zhao, Zhongying
Tu, Yu Qiao, Dahua Lin, Jiangmiao Pang
|
GRUtopia: Dream General Robots in a City at Scale
| null | null | null | null |
cs.RO cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Recent works have been exploring the scaling laws in the field of Embodied
AI. Given the prohibitive costs of collecting real-world data, we believe the
Simulation-to-Real (Sim2Real) paradigm is a crucial step for scaling the
learning of embodied models. This paper introduces project GRUtopia, the first
simulated interactive 3D society designed for various robots. It features
several advancements: (a) The scene dataset, GRScenes, includes 100k
interactive, finely annotated scenes, which can be freely combined into
city-scale environments. In contrast to previous works mainly focusing on home,
GRScenes covers 89 diverse scene categories, bridging the gap of
service-oriented environments where general robots would be initially deployed.
(b) GRResidents, a Large Language Model (LLM) driven Non-Player Character (NPC)
system that is responsible for social interaction, task generation, and task
assignment, thus simulating social scenarios for embodied AI applications. (c)
The benchmark, GRBench, supports various robots but focuses on legged robots as
primary agents and poses moderately challenging tasks involving Object
Loco-Navigation, Social Loco-Navigation, and Loco-Manipulation. We hope that
this work can alleviate the scarcity of high-quality data in this field and
provide a more comprehensive assessment of Embodied AI research. The project is
available at https://github.com/OpenRobotLab/GRUtopia.
|
[
{
"created": "Mon, 15 Jul 2024 17:40:46 GMT",
"version": "v1"
}
] |
2024-07-16
|
[
[
"Wang",
"Hanqing",
""
],
[
"Chen",
"Jiahe",
""
],
[
"Huang",
"Wensi",
""
],
[
"Ben",
"Qingwei",
""
],
[
"Wang",
"Tai",
""
],
[
"Mi",
"Boyu",
""
],
[
"Huang",
"Tao",
""
],
[
"Zhao",
"Siheng",
""
],
[
"Chen",
"Yilun",
""
],
[
"Yang",
"Sizhe",
""
],
[
"Cao",
"Peizhou",
""
],
[
"Yu",
"Wenye",
""
],
[
"Ye",
"Zichao",
""
],
[
"Li",
"Jialun",
""
],
[
"Long",
"Junfeng",
""
],
[
"Wang",
"Zirui",
""
],
[
"Wang",
"Huiling",
""
],
[
"Zhao",
"Ying",
""
],
[
"Tu",
"Zhongying",
""
],
[
"Qiao",
"Yu",
""
],
[
"Lin",
"Dahua",
""
],
[
"Pang",
"Jiangmiao",
""
]
] |
Recent works have been exploring the scaling laws in the field of Embodied AI. Given the prohibitive costs of collecting real-world data, we believe the Simulation-to-Real (Sim2Real) paradigm is a crucial step for scaling the learning of embodied models. This paper introduces project GRUtopia, the first simulated interactive 3D society designed for various robots. It features several advancements: (a) The scene dataset, GRScenes, includes 100k interactive, finely annotated scenes, which can be freely combined into city-scale environments. In contrast to previous works mainly focusing on home, GRScenes covers 89 diverse scene categories, bridging the gap of service-oriented environments where general robots would be initially deployed. (b) GRResidents, a Large Language Model (LLM) driven Non-Player Character (NPC) system that is responsible for social interaction, task generation, and task assignment, thus simulating social scenarios for embodied AI applications. (c) The benchmark, GRBench, supports various robots but focuses on legged robots as primary agents and poses moderately challenging tasks involving Object Loco-Navigation, Social Loco-Navigation, and Loco-Manipulation. We hope that this work can alleviate the scarcity of high-quality data in this field and provide a more comprehensive assessment of Embodied AI research. The project is available at https://github.com/OpenRobotLab/GRUtopia.
|
1903.04554
|
Ali Rahmati
|
Ali Rahmati, Seyyedali Hosseinalipour, Huaiyu Dai
|
Optimal Time Allocation in VANETs Advertising: A Price-based Approach
using Stackelberg Game
|
6 pages, 4 figures
| null | null | null |
cs.GT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Vehicular ad-hoc networks (VANETs) have recently attracted a lot of attention
due to their immense potentials and applications. Wide range of coverage and
accessibility to end users make VANETs a good target for commercial companies.
In this paper, we consider a scenario in which advertising companies aim to
disseminate their advertisements in different areas of a city by utilizing
VANETs infrastructure. These companies compete for renting the VANETs
infrastructure to spread their advertisements. We partition the city map into
different blocks, and consider a manager for all the blocks who is in charge of
splitting the time between interested advertising companies. Each advertising
company (AdC) is charged proportional to the allocated time. In order to find
the best time splitting between AdCs, we propose a Stackelberg game scheme in
which the block manager assigns the companies to the blocks and imposes the
renting prices to different companies in order to maximize its own profit.
Based on this, AdCs request the amount of time they desire to rent the
infrastructure in order to maximize their utilities. To obtain the Stackelberg
equilibrium of the game, a mixed integer nonlinear optimization problem is
solved using the proposed optimal and sub-optimal algorithms. The simulation
results demonstrate that the sub-optimal algorithm approaches the optimal one
in performance with lower complexity.
|
[
{
"created": "Mon, 11 Mar 2019 19:21:54 GMT",
"version": "v1"
}
] |
2019-03-13
|
[
[
"Rahmati",
"Ali",
""
],
[
"Hosseinalipour",
"Seyyedali",
""
],
[
"Dai",
"Huaiyu",
""
]
] |
Vehicular ad-hoc networks (VANETs) have recently attracted a lot of attention due to their immense potentials and applications. Wide range of coverage and accessibility to end users make VANETs a good target for commercial companies. In this paper, we consider a scenario in which advertising companies aim to disseminate their advertisements in different areas of a city by utilizing VANETs infrastructure. These companies compete for renting the VANETs infrastructure to spread their advertisements. We partition the city map into different blocks, and consider a manager for all the blocks who is in charge of splitting the time between interested advertising companies. Each advertising company (AdC) is charged proportional to the allocated time. In order to find the best time splitting between AdCs, we propose a Stackelberg game scheme in which the block manager assigns the companies to the blocks and imposes the renting prices to different companies in order to maximize its own profit. Based on this, AdCs request the amount of time they desire to rent the infrastructure in order to maximize their utilities. To obtain the Stackelberg equilibrium of the game, a mixed integer nonlinear optimization problem is solved using the proposed optimal and sub-optimal algorithms. The simulation results demonstrate that the sub-optimal algorithm approaches the optimal one in performance with lower complexity.
|
1703.08440
|
Kojo Sarfo Gyamfi
|
Kojo Sarfo Gyamfi, James Brusey and Andrew Hunt
|
K-Means Clustering using Tabu Search with Quantized Means
|
World Conference on Engineering and Computer Science
| null | null | null |
cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The Tabu Search (TS) metaheuristic has been proposed for K-Means clustering
as an alternative to Lloyd's algorithm, which for all its ease of
implementation and fast runtime, has the major drawback of being trapped at
local optima. While the TS approach can yield superior performance, it involves
a high computational complexity. Moreover, the difficulty in parameter
selection in the existing TS approach does not make it any more attractive.
This paper presents an alternative, low-complexity formulation of the TS
optimization procedure for K-Means clustering. This approach does not require
many parameter settings. We initially constrain the centers to points in the
dataset. We then aim at evolving these centers using a unique neighborhood
structure that makes use of gradient information of the objective function.
This results in an efficient exploration of the search space, after which the
means are refined. The proposed scheme is implemented in MATLAB and tested on
four real-world datasets, and it achieves a significant improvement over the
existing TS approach in terms of the intra cluster sum of squares and
computational time.
|
[
{
"created": "Fri, 24 Mar 2017 14:59:06 GMT",
"version": "v1"
}
] |
2017-03-27
|
[
[
"Gyamfi",
"Kojo Sarfo",
""
],
[
"Brusey",
"James",
""
],
[
"Hunt",
"Andrew",
""
]
] |
The Tabu Search (TS) metaheuristic has been proposed for K-Means clustering as an alternative to Lloyd's algorithm, which for all its ease of implementation and fast runtime, has the major drawback of being trapped at local optima. While the TS approach can yield superior performance, it involves a high computational complexity. Moreover, the difficulty in parameter selection in the existing TS approach does not make it any more attractive. This paper presents an alternative, low-complexity formulation of the TS optimization procedure for K-Means clustering. This approach does not require many parameter settings. We initially constrain the centers to points in the dataset. We then aim at evolving these centers using a unique neighborhood structure that makes use of gradient information of the objective function. This results in an efficient exploration of the search space, after which the means are refined. The proposed scheme is implemented in MATLAB and tested on four real-world datasets, and it achieves a significant improvement over the existing TS approach in terms of the intra cluster sum of squares and computational time.
|
2205.05192
|
Cynthia Liem
|
Han-Yin Huang and Cynthia C. S. Liem
|
Social Inclusion in Curated Contexts: Insights from Museum Practices
|
in Proceedings of the 2022 ACM Conference on Fairness,
Accountability, and Transparency (FAccT '22), June 21-24, 2022, Seoul,
Republic of Korea
| null |
10.1145/3531146.3533095
| null |
cs.LG cs.AI cs.CY
|
http://creativecommons.org/licenses/by/4.0/
|
Artificial intelligence literature suggests that minority and fragile
communities in society can be negatively impacted by machine learning
algorithms due to inherent biases in the design process, which lead to socially
exclusive decisions and policies. Faced with similar challenges in dealing with
an increasingly diversified audience, the museum sector has seen changes in
theory and practice, particularly in the areas of representation and
meaning-making. While rarity and grandeur used to be at the centre stage of the
early museum practices, folk life and museums' relationships with the diverse
communities they serve become a widely integrated part of the contemporary
practices. These changes address issues of diversity and accessibility in order
to offer more socially inclusive services. Drawing on these changes and
reflecting back on the AI world, we argue that the museum experience provides
useful lessons for building AI with socially inclusive approaches, especially
in situations in which both a collection and access to it will need to be
curated or filtered, as frequently happens in search engines, recommender
systems and digital libraries. We highlight three principles: (1) Instead of
upholding the value of neutrality, practitioners are aware of the influences of
their own backgrounds and those of others on their work. By not claiming to be
neutral but practising cultural humility, the chances of addressing potential
biases can be increased. (2) There should be room for situational
interpretation beyond the stages of data collection and machine learning.
Before applying models and predictions, the contexts in which relevant parties
exist should be taken into account. (3) Community participation serves the
needs of communities and has the added benefit of bringing practitioners and
communities together.
|
[
{
"created": "Tue, 10 May 2022 22:22:12 GMT",
"version": "v1"
}
] |
2022-05-12
|
[
[
"Huang",
"Han-Yin",
""
],
[
"Liem",
"Cynthia C. S.",
""
]
] |
Artificial intelligence literature suggests that minority and fragile communities in society can be negatively impacted by machine learning algorithms due to inherent biases in the design process, which lead to socially exclusive decisions and policies. Faced with similar challenges in dealing with an increasingly diversified audience, the museum sector has seen changes in theory and practice, particularly in the areas of representation and meaning-making. While rarity and grandeur used to be at the centre stage of the early museum practices, folk life and museums' relationships with the diverse communities they serve become a widely integrated part of the contemporary practices. These changes address issues of diversity and accessibility in order to offer more socially inclusive services. Drawing on these changes and reflecting back on the AI world, we argue that the museum experience provides useful lessons for building AI with socially inclusive approaches, especially in situations in which both a collection and access to it will need to be curated or filtered, as frequently happens in search engines, recommender systems and digital libraries. We highlight three principles: (1) Instead of upholding the value of neutrality, practitioners are aware of the influences of their own backgrounds and those of others on their work. By not claiming to be neutral but practising cultural humility, the chances of addressing potential biases can be increased. (2) There should be room for situational interpretation beyond the stages of data collection and machine learning. Before applying models and predictions, the contexts in which relevant parties exist should be taken into account. (3) Community participation serves the needs of communities and has the added benefit of bringing practitioners and communities together.
|
2107.04020
|
Xuejing Lei
|
Xuejing Lei, Ganning Zhao, Kaitai Zhang, C.-C. Jay Kuo
|
TGHop: An Explainable, Efficient and Lightweight Method for Texture
Generation
|
arXiv admin note: substantial text overlap with arXiv:2009.01376
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
An explainable, efficient and lightweight method for texture generation,
called TGHop (an acronym of Texture Generation PixelHop), is proposed in this
work. Although synthesis of visually pleasant texture can be achieved by deep
neural networks, the associated models are large in size, difficult to explain
in theory, and computationally expensive in training. In contrast, TGHop is
small in its model size, mathematically transparent, efficient in training and
inference, and able to generate high quality texture. Given an exemplary
texture, TGHop first crops many sample patches out of it to form a collection
of sample patches called the source. Then, it analyzes pixel statistics of
samples from the source and obtains a sequence of fine-to-coarse subspaces for
these patches by using the PixelHop++ framework. To generate texture patches
with TGHop, we begin with the coarsest subspace, which is called the core, and
attempt to generate samples in each subspace by following the distribution of
real samples. Finally, texture patches are stitched to form texture images of a
large size. It is demonstrated by experimental results that TGHop can generate
texture images of superior quality with a small model size and at a fast speed.
|
[
{
"created": "Thu, 8 Jul 2021 17:56:58 GMT",
"version": "v1"
}
] |
2021-07-09
|
[
[
"Lei",
"Xuejing",
""
],
[
"Zhao",
"Ganning",
""
],
[
"Zhang",
"Kaitai",
""
],
[
"Kuo",
"C. -C. Jay",
""
]
] |
An explainable, efficient and lightweight method for texture generation, called TGHop (an acronym of Texture Generation PixelHop), is proposed in this work. Although synthesis of visually pleasant texture can be achieved by deep neural networks, the associated models are large in size, difficult to explain in theory, and computationally expensive in training. In contrast, TGHop is small in its model size, mathematically transparent, efficient in training and inference, and able to generate high quality texture. Given an exemplary texture, TGHop first crops many sample patches out of it to form a collection of sample patches called the source. Then, it analyzes pixel statistics of samples from the source and obtains a sequence of fine-to-coarse subspaces for these patches by using the PixelHop++ framework. To generate texture patches with TGHop, we begin with the coarsest subspace, which is called the core, and attempt to generate samples in each subspace by following the distribution of real samples. Finally, texture patches are stitched to form texture images of a large size. It is demonstrated by experimental results that TGHop can generate texture images of superior quality with a small model size and at a fast speed.
|
2011.13056
|
Huseyin Birkan Yilmaz
|
Mehmet Sukru Kuran and H. Birkan Yilmaz and Ilker Demirkol and Nariman
Farsad and Andrea Goldsmith
|
A Survey on Modulation Techniques in Molecular Communication via
Diffusion
|
Preprint of the accepted manuscript for publication in IEEE Surveys
and Tutorials
| null | null | null |
cs.ET cs.IT math.IT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
This survey paper focuses on modulation aspects of molecular communication,
an emerging field focused on building biologically-inspired systems that embed
data within chemical signals. The primary challenges in designing these systems
are how to encode and modulate information onto chemical signals, and how to
design a receiver that can detect and decode the information from the corrupted
chemical signal observed at the destination. In this paper, we focus on
modulation design for molecular communication via diffusion systems. In these
systems, chemical signals are transported using diffusion, possibly assisted by
flow, from the transmitter to the receiver. This tutorial presents recent
advancements in modulation and demodulation schemes for molecular communication
via diffusion. We compare five different modulation types: concentration-based,
type-based, timing-based, spatial, and higher-order modulation techniques. The
end-to-end system designs for each modulation scheme are presented. In
addition, the key metrics used in the literature to evaluate the performance of
these techniques are also presented. Finally, we provide a numerical bit error
rate comparison of prominent modulation techniques using analytical models. We
close the tutorial with a discussion of key open issues and future research
directions for design of molecular communication via diffusion systems.
|
[
{
"created": "Wed, 25 Nov 2020 23:00:50 GMT",
"version": "v1"
},
{
"created": "Mon, 28 Dec 2020 09:32:37 GMT",
"version": "v2"
}
] |
2020-12-29
|
[
[
"Kuran",
"Mehmet Sukru",
""
],
[
"Yilmaz",
"H. Birkan",
""
],
[
"Demirkol",
"Ilker",
""
],
[
"Farsad",
"Nariman",
""
],
[
"Goldsmith",
"Andrea",
""
]
] |
This survey paper focuses on modulation aspects of molecular communication, an emerging field focused on building biologically-inspired systems that embed data within chemical signals. The primary challenges in designing these systems are how to encode and modulate information onto chemical signals, and how to design a receiver that can detect and decode the information from the corrupted chemical signal observed at the destination. In this paper, we focus on modulation design for molecular communication via diffusion systems. In these systems, chemical signals are transported using diffusion, possibly assisted by flow, from the transmitter to the receiver. This tutorial presents recent advancements in modulation and demodulation schemes for molecular communication via diffusion. We compare five different modulation types: concentration-based, type-based, timing-based, spatial, and higher-order modulation techniques. The end-to-end system designs for each modulation scheme are presented. In addition, the key metrics used in the literature to evaluate the performance of these techniques are also presented. Finally, we provide a numerical bit error rate comparison of prominent modulation techniques using analytical models. We close the tutorial with a discussion of key open issues and future research directions for design of molecular communication via diffusion systems.
|
1105.4540
|
Matt Malloy
|
Matthew Malloy and Robert Nowak
|
On the Limits of Sequential Testing in High Dimensions
|
Asilomar 2011
| null | null | null |
cs.IT math.IT math.ST stat.TH
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
This paper presents results pertaining to sequential methods for support
recovery of sparse signals in noise. Specifically, we show that any sequential
measurement procedure fails provided the average number of measurements per
dimension grows slower then log s / D(f0||f1) where s is the level of sparsity,
and D(f0||f1) the Kullback-Leibler divergence between the underlying
distributions. For comparison, we show any non-sequential procedure fails
provided the number of measurements grows at a rate less than log n /
D(f1||f0), where n is the total dimension of the problem. Lastly, we show that
a simple procedure termed sequential thresholding guarantees exact support
recovery provided the average number of measurements per dimension grows faster
than (log s + log log n) / D(f0||f1), a mere additive factor more than the
lower bound.
|
[
{
"created": "Mon, 23 May 2011 15:58:03 GMT",
"version": "v1"
},
{
"created": "Mon, 25 Jul 2011 20:13:41 GMT",
"version": "v2"
},
{
"created": "Tue, 18 Oct 2011 16:12:53 GMT",
"version": "v3"
}
] |
2011-10-19
|
[
[
"Malloy",
"Matthew",
""
],
[
"Nowak",
"Robert",
""
]
] |
This paper presents results pertaining to sequential methods for support recovery of sparse signals in noise. Specifically, we show that any sequential measurement procedure fails provided the average number of measurements per dimension grows slower then log s / D(f0||f1) where s is the level of sparsity, and D(f0||f1) the Kullback-Leibler divergence between the underlying distributions. For comparison, we show any non-sequential procedure fails provided the number of measurements grows at a rate less than log n / D(f1||f0), where n is the total dimension of the problem. Lastly, we show that a simple procedure termed sequential thresholding guarantees exact support recovery provided the average number of measurements per dimension grows faster than (log s + log log n) / D(f0||f1), a mere additive factor more than the lower bound.
|
1905.00919
|
Mohamed Baza
|
Ahmed Shafee, Mohamed Baza, Douglas A. Talbert, Mostafa M. Fouda,
Mahmoud Nabil, Mohamed Mahmoud
|
Mimic Learning to Generate a Shareable Network Intrusion Detection Model
| null | null | null | null |
cs.CR cs.LG stat.ML
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Purveyors of malicious network attacks continue to increase the complexity
and the sophistication of their techniques, and their ability to evade
detection continues to improve as well. Hence, intrusion detection systems must
also evolve to meet these increasingly challenging threats. Machine learning is
often used to support this needed improvement. However, training a good
prediction model can require a large set of labelled training data. Such
datasets are difficult to obtain because privacy concerns prevent the majority
of intrusion detection agencies from sharing their sensitive data. In this
paper, we propose the use of mimic learning to enable the transfer of intrusion
detection knowledge through a teacher model trained on private data to a
student model. This student model provides a mean of publicly sharing knowledge
extracted from private data without sharing the data itself. Our results
confirm that the proposed scheme can produce a student intrusion detection
model that mimics the teacher model without requiring access to the original
dataset.
|
[
{
"created": "Thu, 2 May 2019 18:14:24 GMT",
"version": "v1"
},
{
"created": "Sat, 5 Oct 2019 17:39:51 GMT",
"version": "v2"
},
{
"created": "Tue, 18 Feb 2020 20:14:47 GMT",
"version": "v3"
}
] |
2020-02-20
|
[
[
"Shafee",
"Ahmed",
""
],
[
"Baza",
"Mohamed",
""
],
[
"Talbert",
"Douglas A.",
""
],
[
"Fouda",
"Mostafa M.",
""
],
[
"Nabil",
"Mahmoud",
""
],
[
"Mahmoud",
"Mohamed",
""
]
] |
Purveyors of malicious network attacks continue to increase the complexity and the sophistication of their techniques, and their ability to evade detection continues to improve as well. Hence, intrusion detection systems must also evolve to meet these increasingly challenging threats. Machine learning is often used to support this needed improvement. However, training a good prediction model can require a large set of labelled training data. Such datasets are difficult to obtain because privacy concerns prevent the majority of intrusion detection agencies from sharing their sensitive data. In this paper, we propose the use of mimic learning to enable the transfer of intrusion detection knowledge through a teacher model trained on private data to a student model. This student model provides a mean of publicly sharing knowledge extracted from private data without sharing the data itself. Our results confirm that the proposed scheme can produce a student intrusion detection model that mimics the teacher model without requiring access to the original dataset.
|
1610.06721
|
Cosimo Anglano
|
Cosimo Anglano, Massimo Canonico, Marco Guazzone
|
Forensic Analysis of the ChatSecure Instant Messaging Application on
Android Smartphones
| null |
Digital Investigation, Volume 19, December 2016, Pages 44-59
|
10.1016/j.diin.2016.10.001
| null |
cs.CR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We present the forensic analysis of the artifacts generated on Android
smartphones by ChatSecure, a secure Instant Messaging application that provides
strong encryption for transmitted and locally-stored data to ensure the privacy
of its users.
We show that ChatSecure stores local copies of both exchanged messages and
files into two distinct, AES-256 encrypted databases, and we devise a technique
able to decrypt them when the secret passphrase, chosen by the user as the
initial step of the encryption process, is known.
Furthermore, we show how this passphrase can be identified and extracted from
the volatile memory of the device, where it persists for the entire execution
of ChatSecure after having been entered by the user, thus allowing one to carry
out decryption even if the passphrase is not revealed by the user.
Finally, we discuss how to analyze and correlate the data stored in the
databases used by ChatSecure to identify the IM accounts used by the user and
his/her buddies to communicate, as well as to reconstruct the chronology and
contents of the messages and files that have been exchanged among them.
For our study we devise and use an experimental methodology, based on the use
of emulated devices, that provides a very high degree of reproducibility of the
results, and we validate the results it yields against those obtained from real
smartphones.
|
[
{
"created": "Fri, 21 Oct 2016 09:34:33 GMT",
"version": "v1"
}
] |
2016-10-24
|
[
[
"Anglano",
"Cosimo",
""
],
[
"Canonico",
"Massimo",
""
],
[
"Guazzone",
"Marco",
""
]
] |
We present the forensic analysis of the artifacts generated on Android smartphones by ChatSecure, a secure Instant Messaging application that provides strong encryption for transmitted and locally-stored data to ensure the privacy of its users. We show that ChatSecure stores local copies of both exchanged messages and files into two distinct, AES-256 encrypted databases, and we devise a technique able to decrypt them when the secret passphrase, chosen by the user as the initial step of the encryption process, is known. Furthermore, we show how this passphrase can be identified and extracted from the volatile memory of the device, where it persists for the entire execution of ChatSecure after having been entered by the user, thus allowing one to carry out decryption even if the passphrase is not revealed by the user. Finally, we discuss how to analyze and correlate the data stored in the databases used by ChatSecure to identify the IM accounts used by the user and his/her buddies to communicate, as well as to reconstruct the chronology and contents of the messages and files that have been exchanged among them. For our study we devise and use an experimental methodology, based on the use of emulated devices, that provides a very high degree of reproducibility of the results, and we validate the results it yields against those obtained from real smartphones.
|
2401.11107
|
Zhen Chen
|
Zhen Chen, Jingping Liu, Deqing Yang, Yanghua Xiao, Huimin Xu, Zongyu
Wang, Rui Xie and Yunsen Xian
|
Exploiting Duality in Open Information Extraction with Predicate Prompt
| null | null | null | null |
cs.CL cs.IR
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Open information extraction (OpenIE) aims to extract the schema-free triplets
in the form of (\emph{subject}, \emph{predicate}, \emph{object}) from a given
sentence. Compared with general information extraction (IE), OpenIE poses more
challenges for the IE models, {especially when multiple complicated triplets
exist in a sentence. To extract these complicated triplets more effectively, in
this paper we propose a novel generative OpenIE model, namely \emph{DualOIE},
which achieves a dual task at the same time as extracting some triplets from
the sentence, i.e., converting the triplets into the sentence.} Such dual task
encourages the model to correctly recognize the structure of the given sentence
and thus is helpful to extract all potential triplets from the sentence.
Specifically, DualOIE extracts the triplets in two steps: 1) first extracting a
sequence of all potential predicates, 2) then using the predicate sequence as a
prompt to induce the generation of triplets. Our experiments on two benchmarks
and our dataset constructed from Meituan demonstrate that DualOIE achieves the
best performance among the state-of-the-art baselines. Furthermore, the online
A/B test on Meituan platform shows that 0.93\% improvement of QV-CTR and 0.56\%
improvement of UV-CTR have been obtained when the triplets extracted by DualOIE
were leveraged in Meituan's search system.
|
[
{
"created": "Sat, 20 Jan 2024 03:55:17 GMT",
"version": "v1"
}
] |
2024-01-23
|
[
[
"Chen",
"Zhen",
""
],
[
"Liu",
"Jingping",
""
],
[
"Yang",
"Deqing",
""
],
[
"Xiao",
"Yanghua",
""
],
[
"Xu",
"Huimin",
""
],
[
"Wang",
"Zongyu",
""
],
[
"Xie",
"Rui",
""
],
[
"Xian",
"Yunsen",
""
]
] |
Open information extraction (OpenIE) aims to extract the schema-free triplets in the form of (\emph{subject}, \emph{predicate}, \emph{object}) from a given sentence. Compared with general information extraction (IE), OpenIE poses more challenges for the IE models, {especially when multiple complicated triplets exist in a sentence. To extract these complicated triplets more effectively, in this paper we propose a novel generative OpenIE model, namely \emph{DualOIE}, which achieves a dual task at the same time as extracting some triplets from the sentence, i.e., converting the triplets into the sentence.} Such dual task encourages the model to correctly recognize the structure of the given sentence and thus is helpful to extract all potential triplets from the sentence. Specifically, DualOIE extracts the triplets in two steps: 1) first extracting a sequence of all potential predicates, 2) then using the predicate sequence as a prompt to induce the generation of triplets. Our experiments on two benchmarks and our dataset constructed from Meituan demonstrate that DualOIE achieves the best performance among the state-of-the-art baselines. Furthermore, the online A/B test on Meituan platform shows that 0.93\% improvement of QV-CTR and 0.56\% improvement of UV-CTR have been obtained when the triplets extracted by DualOIE were leveraged in Meituan's search system.
|
2404.03555
|
Botond Barta
|
Botond Barta, Dorina Lakatos, Attila Nagy, Mil\'an Konor Nyist, Judit
\'Acs
|
From News to Summaries: Building a Hungarian Corpus for Extractive and
Abstractive Summarization
| null | null | null | null |
cs.CL
|
http://creativecommons.org/licenses/by/4.0/
|
Training summarization models requires substantial amounts of training data.
However for less resourceful languages like Hungarian, openly available models
and datasets are notably scarce. To address this gap our paper introduces
HunSum-2 an open-source Hungarian corpus suitable for training abstractive and
extractive summarization models. The dataset is assembled from segments of the
Common Crawl corpus undergoing thorough cleaning, preprocessing and
deduplication. In addition to abstractive summarization we generate
sentence-level labels for extractive summarization using sentence similarity.
We train baseline models for both extractive and abstractive summarization
using the collected dataset. To demonstrate the effectiveness of the trained
models, we perform both quantitative and qualitative evaluation. Our dataset,
models and code are publicly available, encouraging replication, further
research, and real-world applications across various domains.
|
[
{
"created": "Thu, 4 Apr 2024 16:07:06 GMT",
"version": "v1"
},
{
"created": "Fri, 12 Apr 2024 08:05:13 GMT",
"version": "v2"
}
] |
2024-04-15
|
[
[
"Barta",
"Botond",
""
],
[
"Lakatos",
"Dorina",
""
],
[
"Nagy",
"Attila",
""
],
[
"Nyist",
"Milán Konor",
""
],
[
"Ács",
"Judit",
""
]
] |
Training summarization models requires substantial amounts of training data. However for less resourceful languages like Hungarian, openly available models and datasets are notably scarce. To address this gap our paper introduces HunSum-2 an open-source Hungarian corpus suitable for training abstractive and extractive summarization models. The dataset is assembled from segments of the Common Crawl corpus undergoing thorough cleaning, preprocessing and deduplication. In addition to abstractive summarization we generate sentence-level labels for extractive summarization using sentence similarity. We train baseline models for both extractive and abstractive summarization using the collected dataset. To demonstrate the effectiveness of the trained models, we perform both quantitative and qualitative evaluation. Our dataset, models and code are publicly available, encouraging replication, further research, and real-world applications across various domains.
|
2406.13012
|
Chi-Hua Wang
|
Joshua Ward, Chi-Hua Wang, Guang Cheng
|
Data Plagiarism Index: Characterizing the Privacy Risk of Data-Copying
in Tabular Generative Models
| null | null | null | null |
cs.LG cs.CR stat.ML
|
http://creativecommons.org/licenses/by/4.0/
|
The promise of tabular generative models is to produce realistic synthetic
data that can be shared and safely used without dangerous leakage of
information from the training set. In evaluating these models, a variety of
methods have been proposed to measure the tendency to copy data from the
training dataset when generating a sample. However, these methods suffer from
either not considering data-copying from a privacy threat perspective, not
being motivated by recent results in the data-copying literature or being
difficult to make compatible with the high dimensional, mixed type nature of
tabular data. This paper proposes a new similarity metric and Membership
Inference Attack called Data Plagiarism Index (DPI) for tabular data. We show
that DPI evaluates a new intuitive definition of data-copying and characterizes
the corresponding privacy risk. We show that the data-copying identified by DPI
poses both privacy and fairness threats to common, high performing
architectures; underscoring the necessity for more sophisticated generative
modeling techniques to mitigate this issue.
|
[
{
"created": "Tue, 18 Jun 2024 19:05:24 GMT",
"version": "v1"
}
] |
2024-06-21
|
[
[
"Ward",
"Joshua",
""
],
[
"Wang",
"Chi-Hua",
""
],
[
"Cheng",
"Guang",
""
]
] |
The promise of tabular generative models is to produce realistic synthetic data that can be shared and safely used without dangerous leakage of information from the training set. In evaluating these models, a variety of methods have been proposed to measure the tendency to copy data from the training dataset when generating a sample. However, these methods suffer from either not considering data-copying from a privacy threat perspective, not being motivated by recent results in the data-copying literature or being difficult to make compatible with the high dimensional, mixed type nature of tabular data. This paper proposes a new similarity metric and Membership Inference Attack called Data Plagiarism Index (DPI) for tabular data. We show that DPI evaluates a new intuitive definition of data-copying and characterizes the corresponding privacy risk. We show that the data-copying identified by DPI poses both privacy and fairness threats to common, high performing architectures; underscoring the necessity for more sophisticated generative modeling techniques to mitigate this issue.
|
2102.01723
|
Amir Yazdanbakhsh
|
Amir Yazdanbakhsh, Christof Angermueller, Berkin Akin, Yanqi Zhou,
Albin Jones, Milad Hashemi, Kevin Swersky, Satrajit Chatterjee, Ravi
Narayanaswami, James Laudon
|
Apollo: Transferable Architecture Exploration
|
10 pages, 5 figures, Accepted to Workshop on ML for Systems at the
34th Conference on Neural Information Processing Systems (NeurIPS 2020)
| null | null | null |
cs.LG cs.AR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The looming end of Moore's Law and ascending use of deep learning drives the
design of custom accelerators that are optimized for specific neural
architectures. Architecture exploration for such accelerators forms a
challenging constrained optimization problem over a complex, high-dimensional,
and structured input space with a costly to evaluate objective function.
Existing approaches for accelerator design are sample-inefficient and do not
transfer knowledge between related optimizations tasks with different design
constraints, such as area and/or latency budget, or neural architecture
configurations. In this work, we propose a transferable architecture
exploration framework, dubbed Apollo, that leverages recent advances in
black-box function optimization for sample-efficient accelerator design. We use
this framework to optimize accelerator configurations of a diverse set of
neural architectures with alternative design constraints. We show that our
framework finds high reward design configurations (up to 24.6% speedup) more
sample-efficiently than a baseline black-box optimization approach. We further
show that by transferring knowledge between target architectures with different
design constraints, Apollo is able to find optimal configurations faster and
often with better objective value (up to 25% improvements). This encouraging
outcome portrays a promising path forward to facilitate generating higher
quality accelerators.
|
[
{
"created": "Tue, 2 Feb 2021 19:36:02 GMT",
"version": "v1"
}
] |
2021-02-04
|
[
[
"Yazdanbakhsh",
"Amir",
""
],
[
"Angermueller",
"Christof",
""
],
[
"Akin",
"Berkin",
""
],
[
"Zhou",
"Yanqi",
""
],
[
"Jones",
"Albin",
""
],
[
"Hashemi",
"Milad",
""
],
[
"Swersky",
"Kevin",
""
],
[
"Chatterjee",
"Satrajit",
""
],
[
"Narayanaswami",
"Ravi",
""
],
[
"Laudon",
"James",
""
]
] |
The looming end of Moore's Law and ascending use of deep learning drives the design of custom accelerators that are optimized for specific neural architectures. Architecture exploration for such accelerators forms a challenging constrained optimization problem over a complex, high-dimensional, and structured input space with a costly to evaluate objective function. Existing approaches for accelerator design are sample-inefficient and do not transfer knowledge between related optimizations tasks with different design constraints, such as area and/or latency budget, or neural architecture configurations. In this work, we propose a transferable architecture exploration framework, dubbed Apollo, that leverages recent advances in black-box function optimization for sample-efficient accelerator design. We use this framework to optimize accelerator configurations of a diverse set of neural architectures with alternative design constraints. We show that our framework finds high reward design configurations (up to 24.6% speedup) more sample-efficiently than a baseline black-box optimization approach. We further show that by transferring knowledge between target architectures with different design constraints, Apollo is able to find optimal configurations faster and often with better objective value (up to 25% improvements). This encouraging outcome portrays a promising path forward to facilitate generating higher quality accelerators.
|
1406.1557
|
EPTCS
|
Harsh Raju Chamarthi (Northeastern Univeristy), Peter C. Dillinger
(Northeastern Univeristy), Panagiotis Manolios (Northeastern Univeristy)
|
Data Definitions in the ACL2 Sedan
|
In Proceedings ACL2 2014, arXiv:1406.1238
|
EPTCS 152, 2014, pp. 27-48
|
10.4204/EPTCS.152.3
| null |
cs.PL cs.LO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We present a data definition framework that enables the convenient
specification of data types in ACL2s, the ACL2 Sedan. Our primary motivation
for developing the data definition framework was pedagogical. We were teaching
undergraduate students how to reason about programs using ACL2s and wanted to
provide them with an effective method for defining, testing, and reasoning
about data types in the context of an untyped theorem prover. Our framework is
now routinely used not only for pedagogical purposes, but also by advanced
users.
Our framework concisely supports common data definition patterns, e.g. list
types, map types, and record types. It also provides support for polymorphic
functions. A distinguishing feature of our approach is that we maintain both a
predicative and an enumerative characterization of data definitions.
In this paper we present our data definition framework via a sequence of
examples. We give a complete characterization in terms of tau rules of the
inclusion/exclusion relations a data definition induces, under suitable
restrictions. The data definition framework is a key component of
counterexample generation support in ACL2s, but can be independently used in
ACL2, and is available as a community book.
|
[
{
"created": "Fri, 6 Jun 2014 01:47:21 GMT",
"version": "v1"
}
] |
2014-06-09
|
[
[
"Chamarthi",
"Harsh Raju",
"",
"Northeastern Univeristy"
],
[
"Dillinger",
"Peter C.",
"",
"Northeastern Univeristy"
],
[
"Manolios",
"Panagiotis",
"",
"Northeastern Univeristy"
]
] |
We present a data definition framework that enables the convenient specification of data types in ACL2s, the ACL2 Sedan. Our primary motivation for developing the data definition framework was pedagogical. We were teaching undergraduate students how to reason about programs using ACL2s and wanted to provide them with an effective method for defining, testing, and reasoning about data types in the context of an untyped theorem prover. Our framework is now routinely used not only for pedagogical purposes, but also by advanced users. Our framework concisely supports common data definition patterns, e.g. list types, map types, and record types. It also provides support for polymorphic functions. A distinguishing feature of our approach is that we maintain both a predicative and an enumerative characterization of data definitions. In this paper we present our data definition framework via a sequence of examples. We give a complete characterization in terms of tau rules of the inclusion/exclusion relations a data definition induces, under suitable restrictions. The data definition framework is a key component of counterexample generation support in ACL2s, but can be independently used in ACL2, and is available as a community book.
|
2112.01402
|
Rahul Rahaman
|
Dipika Singhania, Rahul Rahaman, Angela Yao
|
Iterative Contrast-Classify For Semi-supervised Temporal Action
Segmentation
|
AAAI-2022
| null | null | null |
cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
Temporal action segmentation classifies the action of each frame in (long)
video sequences. Due to the high cost of frame-wise labeling, we propose the
first semi-supervised method for temporal action segmentation. Our method
hinges on unsupervised representation learning, which, for temporal action
segmentation, poses unique challenges. Actions in untrimmed videos vary in
length and have unknown labels and start/end times. Ordering of actions across
videos may also vary. We propose a novel way to learn frame-wise
representations from temporal convolutional networks (TCNs) by clustering input
features with added time-proximity condition and multi-resolution similarity.
By merging representation learning with conventional supervised learning, we
develop an "Iterative-Contrast-Classify (ICC)" semi-supervised learning scheme.
With more labelled data, ICC progressively improves in performance; ICC
semi-supervised learning, with 40% labelled videos, performs similar to
fully-supervised counterparts. Our ICC improves MoF by {+1.8, +5.6, +2.5}% on
Breakfast, 50Salads and GTEA respectively for 100% labelled videos.
|
[
{
"created": "Thu, 2 Dec 2021 16:47:24 GMT",
"version": "v1"
},
{
"created": "Wed, 8 Dec 2021 14:56:40 GMT",
"version": "v2"
}
] |
2021-12-09
|
[
[
"Singhania",
"Dipika",
""
],
[
"Rahaman",
"Rahul",
""
],
[
"Yao",
"Angela",
""
]
] |
Temporal action segmentation classifies the action of each frame in (long) video sequences. Due to the high cost of frame-wise labeling, we propose the first semi-supervised method for temporal action segmentation. Our method hinges on unsupervised representation learning, which, for temporal action segmentation, poses unique challenges. Actions in untrimmed videos vary in length and have unknown labels and start/end times. Ordering of actions across videos may also vary. We propose a novel way to learn frame-wise representations from temporal convolutional networks (TCNs) by clustering input features with added time-proximity condition and multi-resolution similarity. By merging representation learning with conventional supervised learning, we develop an "Iterative-Contrast-Classify (ICC)" semi-supervised learning scheme. With more labelled data, ICC progressively improves in performance; ICC semi-supervised learning, with 40% labelled videos, performs similar to fully-supervised counterparts. Our ICC improves MoF by {+1.8, +5.6, +2.5}% on Breakfast, 50Salads and GTEA respectively for 100% labelled videos.
|
2211.02448
|
Dongchao Yang
|
Dongchao Yang, Songxiang Liu, Jianwei Yu, Helin Wang, Chao Weng,
Yuexian Zou
|
NoreSpeech: Knowledge Distillation based Conditional Diffusion Model for
Noise-robust Expressive TTS
|
Submitted to ICASSP2023
| null | null | null |
cs.SD eess.AS
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Expressive text-to-speech (TTS) can synthesize a new speaking style by
imiating prosody and timbre from a reference audio, which faces the following
challenges: (1) The highly dynamic prosody information in the reference audio
is difficult to extract, especially, when the reference audio contains
background noise. (2) The TTS systems should have good generalization for
unseen speaking styles. In this paper, we present a
\textbf{no}ise-\textbf{r}obust \textbf{e}xpressive TTS model (NoreSpeech),
which can robustly transfer speaking style in a noisy reference utterance to
synthesized speech. Specifically, our NoreSpeech includes several components:
(1) a novel DiffStyle module, which leverages powerful probabilistic denoising
diffusion models to learn noise-agnostic speaking style features from a teacher
model by knowledge distillation; (2) a VQ-VAE block, which maps the style
features into a controllable quantized latent space for improving the
generalization of style transfer; and (3) a straight-forward but effective
parameter-free text-style alignment module, which enables NoreSpeech to
transfer style to a textual input from a length-mismatched reference utterance.
Experiments demonstrate that NoreSpeech is more effective than previous
expressive TTS models in noise environments. Audio samples and code are
available at:
\href{http://dongchaoyang.top/NoreSpeech\_demo/}{http://dongchaoyang.top/NoreSpeech\_demo/}
|
[
{
"created": "Fri, 4 Nov 2022 13:32:58 GMT",
"version": "v1"
}
] |
2022-11-07
|
[
[
"Yang",
"Dongchao",
""
],
[
"Liu",
"Songxiang",
""
],
[
"Yu",
"Jianwei",
""
],
[
"Wang",
"Helin",
""
],
[
"Weng",
"Chao",
""
],
[
"Zou",
"Yuexian",
""
]
] |
Expressive text-to-speech (TTS) can synthesize a new speaking style by imiating prosody and timbre from a reference audio, which faces the following challenges: (1) The highly dynamic prosody information in the reference audio is difficult to extract, especially, when the reference audio contains background noise. (2) The TTS systems should have good generalization for unseen speaking styles. In this paper, we present a \textbf{no}ise-\textbf{r}obust \textbf{e}xpressive TTS model (NoreSpeech), which can robustly transfer speaking style in a noisy reference utterance to synthesized speech. Specifically, our NoreSpeech includes several components: (1) a novel DiffStyle module, which leverages powerful probabilistic denoising diffusion models to learn noise-agnostic speaking style features from a teacher model by knowledge distillation; (2) a VQ-VAE block, which maps the style features into a controllable quantized latent space for improving the generalization of style transfer; and (3) a straight-forward but effective parameter-free text-style alignment module, which enables NoreSpeech to transfer style to a textual input from a length-mismatched reference utterance. Experiments demonstrate that NoreSpeech is more effective than previous expressive TTS models in noise environments. Audio samples and code are available at: \href{http://dongchaoyang.top/NoreSpeech\_demo/}{http://dongchaoyang.top/NoreSpeech\_demo/}
|
2403.02810
|
Chu Wang
|
Chu Wang, Jinhong Wu, Yanzhi Wang, Zhijian Zha, Qi Zhou
|
Dynamic Gaussian Graph Operator: Learning parametric partial
differential equations in arbitrary discrete mechanics problems
|
The number of figures is 13. The number of tables is 7. The number of
words is 9854
| null | null | null |
cs.LG cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Deep learning methods have access to be employed for solving physical systems
governed by parametric partial differential equations (PDEs) due to massive
scientific data. It has been refined to operator learning that focuses on
learning non-linear mapping between infinite-dimensional function spaces,
offering interface from observations to solutions. However, state-of-the-art
neural operators are limited to constant and uniform discretization, thereby
leading to deficiency in generalization on arbitrary discretization schemes for
computational domain. In this work, we propose a novel operator learning
algorithm, referred to as Dynamic Gaussian Graph Operator (DGGO) that expands
neural operators to learning parametric PDEs in arbitrary discrete mechanics
problems. The Dynamic Gaussian Graph (DGG) kernel learns to map the observation
vectors defined in general Euclidean space to metric vectors defined in
high-dimensional uniform metric space. The DGG integral kernel is parameterized
by Gaussian kernel weighted Riemann sum approximating and using dynamic message
passing graph to depict the interrelation within the integral term. Fourier
Neural Operator is selected to localize the metric vectors on spatial and
frequency domains. Metric vectors are regarded as located on latent uniform
domain, wherein spatial and spectral transformation offer highly regular
constraints on solution space. The efficiency and robustness of DGGO are
validated by applying it to solve numerical arbitrary discrete mechanics
problems in comparison with mainstream neural operators. Ablation experiments
are implemented to demonstrate the effectiveness of spatial transformation in
the DGG kernel. The proposed method is utilized to forecast stress field of
hyper-elastic material with geometrically variable void as engineering
application.
|
[
{
"created": "Tue, 5 Mar 2024 09:25:31 GMT",
"version": "v1"
}
] |
2024-03-06
|
[
[
"Wang",
"Chu",
""
],
[
"Wu",
"Jinhong",
""
],
[
"Wang",
"Yanzhi",
""
],
[
"Zha",
"Zhijian",
""
],
[
"Zhou",
"Qi",
""
]
] |
Deep learning methods have access to be employed for solving physical systems governed by parametric partial differential equations (PDEs) due to massive scientific data. It has been refined to operator learning that focuses on learning non-linear mapping between infinite-dimensional function spaces, offering interface from observations to solutions. However, state-of-the-art neural operators are limited to constant and uniform discretization, thereby leading to deficiency in generalization on arbitrary discretization schemes for computational domain. In this work, we propose a novel operator learning algorithm, referred to as Dynamic Gaussian Graph Operator (DGGO) that expands neural operators to learning parametric PDEs in arbitrary discrete mechanics problems. The Dynamic Gaussian Graph (DGG) kernel learns to map the observation vectors defined in general Euclidean space to metric vectors defined in high-dimensional uniform metric space. The DGG integral kernel is parameterized by Gaussian kernel weighted Riemann sum approximating and using dynamic message passing graph to depict the interrelation within the integral term. Fourier Neural Operator is selected to localize the metric vectors on spatial and frequency domains. Metric vectors are regarded as located on latent uniform domain, wherein spatial and spectral transformation offer highly regular constraints on solution space. The efficiency and robustness of DGGO are validated by applying it to solve numerical arbitrary discrete mechanics problems in comparison with mainstream neural operators. Ablation experiments are implemented to demonstrate the effectiveness of spatial transformation in the DGG kernel. The proposed method is utilized to forecast stress field of hyper-elastic material with geometrically variable void as engineering application.
|
1908.07668
|
Sara Billey
|
Molly Baird, Sara C. Billey, Erik D. Demaine, Martin L. Demaine, David
Eppstein, S\'andor Fekete, Graham Gordon, Sean Griffin, Joseph S. B.
Mitchell, Joshua P. Swanson
|
Existence and hardness of conveyor belts
| null |
Electronic J. Combinatorics 27 (4), Paper 4.25, 2020
|
10.37236/9782
| null |
cs.CG math.CO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
An open problem of Manuel Abellanas asks whether every set of disjoint closed
unit disks in the plane can be connected by a conveyor belt, which means a
tight simple closed curve that touches the boundary of each disk, possibly
multiple times. We prove three main results. First, for unit disks whose
centers are both $x$-monotone and $y$-monotone, or whose centers have
$x$-coordinates that differ by at least two units, a conveyor belt always
exists and can be found efficiently. Second, it is NP-complete to determine
whether disks of varying radii have a conveyor belt, and it remains NP-complete
when we constrain the belt to touch disks exactly once. Third, any disjoint set
of $n$ disks of arbitrary radii can be augmented by $O(n)$ "guide" disks so
that the augmented system has a conveyor belt touching each disk exactly once,
answering a conjecture of Demaine, Demaine, and Palop.
|
[
{
"created": "Wed, 21 Aug 2019 01:38:33 GMT",
"version": "v1"
}
] |
2020-11-09
|
[
[
"Baird",
"Molly",
""
],
[
"Billey",
"Sara C.",
""
],
[
"Demaine",
"Erik D.",
""
],
[
"Demaine",
"Martin L.",
""
],
[
"Eppstein",
"David",
""
],
[
"Fekete",
"Sándor",
""
],
[
"Gordon",
"Graham",
""
],
[
"Griffin",
"Sean",
""
],
[
"Mitchell",
"Joseph S. B.",
""
],
[
"Swanson",
"Joshua P.",
""
]
] |
An open problem of Manuel Abellanas asks whether every set of disjoint closed unit disks in the plane can be connected by a conveyor belt, which means a tight simple closed curve that touches the boundary of each disk, possibly multiple times. We prove three main results. First, for unit disks whose centers are both $x$-monotone and $y$-monotone, or whose centers have $x$-coordinates that differ by at least two units, a conveyor belt always exists and can be found efficiently. Second, it is NP-complete to determine whether disks of varying radii have a conveyor belt, and it remains NP-complete when we constrain the belt to touch disks exactly once. Third, any disjoint set of $n$ disks of arbitrary radii can be augmented by $O(n)$ "guide" disks so that the augmented system has a conveyor belt touching each disk exactly once, answering a conjecture of Demaine, Demaine, and Palop.
|
1806.08274
|
Pallavi Athe
|
Pallavi Athe, Yatindra Nath Singh
|
Impact Zone Analysis of p-Cycle
| null | null | null | null |
cs.NI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Pre-configured cycle (p-Cycle) method has been studied in literature
extensively for optical network protection. A large p-cycle has high capacity
efficiency and can protect a large number of nodes against the single link
failure scenarios. All the links protected by such a p-cycle lose protection
when the p-cycle is consumed to restore traffic after a failure. As the
probability of multiple link failure is high for a large network, it also means
that with higher probability, on the second failure, protection may not be
there for the failed link. Thus, if the number of links protected by a p-cycle
is large, it makes the network unprotected with high probability on the advent
of the second failure. In this paper, we study the impact zone due to a first
link failure in the various configurations of the p-cycles. The study gives
insight into how to choose the p-cycle configuration to reduce the impact zone
while using minimum spare capacity. We propose few methods and compare them to
show how the impact zone analysis can be used to improve the fault tolerance in
an optical network.
|
[
{
"created": "Thu, 21 Jun 2018 14:50:15 GMT",
"version": "v1"
}
] |
2018-06-22
|
[
[
"Athe",
"Pallavi",
""
],
[
"Singh",
"Yatindra Nath",
""
]
] |
Pre-configured cycle (p-Cycle) method has been studied in literature extensively for optical network protection. A large p-cycle has high capacity efficiency and can protect a large number of nodes against the single link failure scenarios. All the links protected by such a p-cycle lose protection when the p-cycle is consumed to restore traffic after a failure. As the probability of multiple link failure is high for a large network, it also means that with higher probability, on the second failure, protection may not be there for the failed link. Thus, if the number of links protected by a p-cycle is large, it makes the network unprotected with high probability on the advent of the second failure. In this paper, we study the impact zone due to a first link failure in the various configurations of the p-cycles. The study gives insight into how to choose the p-cycle configuration to reduce the impact zone while using minimum spare capacity. We propose few methods and compare them to show how the impact zone analysis can be used to improve the fault tolerance in an optical network.
|
2105.10310
|
Arnaud Boutillon
|
Arnaud Boutillon, Pierre-Henri Conze, Christelle Pons, Val\'erie
Burdin, Bhushan Borotikar
|
Multi-Task, Multi-Domain Deep Segmentation with Shared Representations
and Contrastive Regularization for Sparse Pediatric Datasets
|
11 pages, 4 figures, 2 tables, accepted at the 24th International
Conference on Medical Image Computing and Computer-Assisted Intervention
(MICCAI 2021)
| null |
10.1007/978-3-030-87193-2_23
| null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Automatic segmentation of magnetic resonance (MR) images is crucial for
morphological evaluation of the pediatric musculoskeletal system in clinical
practice. However, the accuracy and generalization performance of individual
segmentation models are limited due to the restricted amount of annotated
pediatric data. Hence, we propose to train a segmentation model on multiple
datasets, arising from different parts of the anatomy, in a multi-task and
multi-domain learning framework. This approach allows to overcome the inherent
scarcity of pediatric data while benefiting from a more robust shared
representation. The proposed segmentation network comprises shared
convolutional filters, domain-specific batch normalization parameters that
compute the respective dataset statistics and a domain-specific segmentation
layer. Furthermore, a supervised contrastive regularization is integrated to
further improve generalization capabilities, by promoting intra-domain
similarity and impose inter-domain margins in embedded space. We evaluate our
contributions on two pediatric imaging datasets of the ankle and shoulder
joints for bone segmentation. Results demonstrate that the proposed model
outperforms state-of-the-art approaches.
|
[
{
"created": "Fri, 21 May 2021 12:26:05 GMT",
"version": "v1"
},
{
"created": "Wed, 2 Feb 2022 09:11:30 GMT",
"version": "v2"
}
] |
2022-02-03
|
[
[
"Boutillon",
"Arnaud",
""
],
[
"Conze",
"Pierre-Henri",
""
],
[
"Pons",
"Christelle",
""
],
[
"Burdin",
"Valérie",
""
],
[
"Borotikar",
"Bhushan",
""
]
] |
Automatic segmentation of magnetic resonance (MR) images is crucial for morphological evaluation of the pediatric musculoskeletal system in clinical practice. However, the accuracy and generalization performance of individual segmentation models are limited due to the restricted amount of annotated pediatric data. Hence, we propose to train a segmentation model on multiple datasets, arising from different parts of the anatomy, in a multi-task and multi-domain learning framework. This approach allows to overcome the inherent scarcity of pediatric data while benefiting from a more robust shared representation. The proposed segmentation network comprises shared convolutional filters, domain-specific batch normalization parameters that compute the respective dataset statistics and a domain-specific segmentation layer. Furthermore, a supervised contrastive regularization is integrated to further improve generalization capabilities, by promoting intra-domain similarity and impose inter-domain margins in embedded space. We evaluate our contributions on two pediatric imaging datasets of the ankle and shoulder joints for bone segmentation. Results demonstrate that the proposed model outperforms state-of-the-art approaches.
|
2107.05278
|
Erwin de Gelder
|
Erwin de Gelder, Eric Cator, Jan-Pieter Paardekooper, Olaf Op den
Camp, Bart De Schutter
|
Constrained Sampling from a Kernel Density Estimator to Generate
Scenarios for the Assessment of Automated Vehicles
|
6 pages, 3 figures, to be published in the proceedings of the IEEE
Intelligent Vehicle Symposium Workshops (IV workshop)
| null | null | null |
cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
The safety assessment of automated vehicles (AVs) is an important aspect of
the development cycle of AVs. A scenario-based assessment approach is accepted
by many players in the field as part of the complete safety assessment. A
scenario is a representation of a situation on the road to which the AV needs
to respond appropriately. One way to generate the required scenario-based test
descriptions is to parameterize the scenarios and to draw these parameters from
a probability density function (pdf). Because the shape of the pdf is unknown
beforehand, assuming a functional form of the pdf and fitting the parameters to
the data may lead to inaccurate fits. As an alternative, Kernel Density
Estimation (KDE) is a promising candidate for estimating the underlying pdf,
because it is flexible with the underlying distribution of the parameters.
Drawing random samples from a pdf estimated with KDE is possible without the
need of evaluating the actual pdf, which makes it suitable for drawing random
samples for, e.g., Monte Carlo methods. Sampling from a KDE while the samples
satisfy a linear equality constraint, however, has not been described in the
literature, as far as the authors know.
In this paper, we propose a method to sample from a pdf estimated using KDE,
such that the samples satisfy a linear equality constraint. We also present an
algorithm of our method in pseudo-code. The method can be used to generating
scenarios that have, e.g., a predetermined starting speed or to generate
different types of scenarios. This paper also shows that the method for
sampling scenarios can be used in case a Singular Value Decomposition (SVD) is
used to reduce the dimension of the parameter vectors.
|
[
{
"created": "Mon, 12 Jul 2021 09:28:25 GMT",
"version": "v1"
}
] |
2021-07-13
|
[
[
"de Gelder",
"Erwin",
""
],
[
"Cator",
"Eric",
""
],
[
"Paardekooper",
"Jan-Pieter",
""
],
[
"Camp",
"Olaf Op den",
""
],
[
"De Schutter",
"Bart",
""
]
] |
The safety assessment of automated vehicles (AVs) is an important aspect of the development cycle of AVs. A scenario-based assessment approach is accepted by many players in the field as part of the complete safety assessment. A scenario is a representation of a situation on the road to which the AV needs to respond appropriately. One way to generate the required scenario-based test descriptions is to parameterize the scenarios and to draw these parameters from a probability density function (pdf). Because the shape of the pdf is unknown beforehand, assuming a functional form of the pdf and fitting the parameters to the data may lead to inaccurate fits. As an alternative, Kernel Density Estimation (KDE) is a promising candidate for estimating the underlying pdf, because it is flexible with the underlying distribution of the parameters. Drawing random samples from a pdf estimated with KDE is possible without the need of evaluating the actual pdf, which makes it suitable for drawing random samples for, e.g., Monte Carlo methods. Sampling from a KDE while the samples satisfy a linear equality constraint, however, has not been described in the literature, as far as the authors know. In this paper, we propose a method to sample from a pdf estimated using KDE, such that the samples satisfy a linear equality constraint. We also present an algorithm of our method in pseudo-code. The method can be used to generating scenarios that have, e.g., a predetermined starting speed or to generate different types of scenarios. This paper also shows that the method for sampling scenarios can be used in case a Singular Value Decomposition (SVD) is used to reduce the dimension of the parameter vectors.
|
2307.16713
|
Haozhen Zhang
|
Haozhen Zhang, Le Yu, Xi Xiao, Qing Li, Francesco Mercaldo, Xiapu Luo,
Qixu Liu
|
TFE-GNN: A Temporal Fusion Encoder Using Graph Neural Networks for
Fine-grained Encrypted Traffic Classification
|
Accepted by The Web Conference 2023 (WWW'23). The code will be
available with our incoming future work
| null | null | null |
cs.LG cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
Encrypted traffic classification is receiving widespread attention from
researchers and industrial companies. However, the existing methods only
extract flow-level features, failing to handle short flows because of
unreliable statistical properties, or treat the header and payload equally,
failing to mine the potential correlation between bytes. Therefore, in this
paper, we propose a byte-level traffic graph construction approach based on
point-wise mutual information (PMI), and a model named Temporal Fusion Encoder
using Graph Neural Networks (TFE-GNN) for feature extraction. In particular, we
design a dual embedding layer, a GNN-based traffic graph encoder as well as a
cross-gated feature fusion mechanism, which can first embed the header and
payload bytes separately and then fuses them together to obtain a stronger
feature representation. The experimental results on two real datasets
demonstrate that TFE-GNN outperforms multiple state-of-the-art methods in
fine-grained encrypted traffic classification tasks.
|
[
{
"created": "Mon, 31 Jul 2023 14:32:40 GMT",
"version": "v1"
}
] |
2023-08-01
|
[
[
"Zhang",
"Haozhen",
""
],
[
"Yu",
"Le",
""
],
[
"Xiao",
"Xi",
""
],
[
"Li",
"Qing",
""
],
[
"Mercaldo",
"Francesco",
""
],
[
"Luo",
"Xiapu",
""
],
[
"Liu",
"Qixu",
""
]
] |
Encrypted traffic classification is receiving widespread attention from researchers and industrial companies. However, the existing methods only extract flow-level features, failing to handle short flows because of unreliable statistical properties, or treat the header and payload equally, failing to mine the potential correlation between bytes. Therefore, in this paper, we propose a byte-level traffic graph construction approach based on point-wise mutual information (PMI), and a model named Temporal Fusion Encoder using Graph Neural Networks (TFE-GNN) for feature extraction. In particular, we design a dual embedding layer, a GNN-based traffic graph encoder as well as a cross-gated feature fusion mechanism, which can first embed the header and payload bytes separately and then fuses them together to obtain a stronger feature representation. The experimental results on two real datasets demonstrate that TFE-GNN outperforms multiple state-of-the-art methods in fine-grained encrypted traffic classification tasks.
|
2407.08135
|
Yinfeng Zhu
|
Yinfeng Zhu
|
A quadratic upper bound on the reset thresholds of synchronizing
automata containing a transitive permutation group
| null | null | null | null |
cs.FL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
For any synchronizing $n$-state deterministic automaton, \v{C}ern\'{y}
conjectures the existence of a synchronizing word of length at most $(n-1)^2$.
We prove that there exists a synchronizing word of length at most $2n^2 - 7n +
7$ for every synchronizing $n$-state deterministic automaton that satisfies the
following two properties: 1. The image of the action of each letter contains at
least $n-1$ states; 2. The actions of bijective letters generate a transitive
permutation group on the state set.
|
[
{
"created": "Thu, 11 Jul 2024 02:17:49 GMT",
"version": "v1"
}
] |
2024-07-12
|
[
[
"Zhu",
"Yinfeng",
""
]
] |
For any synchronizing $n$-state deterministic automaton, \v{C}ern\'{y} conjectures the existence of a synchronizing word of length at most $(n-1)^2$. We prove that there exists a synchronizing word of length at most $2n^2 - 7n + 7$ for every synchronizing $n$-state deterministic automaton that satisfies the following two properties: 1. The image of the action of each letter contains at least $n-1$ states; 2. The actions of bijective letters generate a transitive permutation group on the state set.
|
2103.05424
|
Mauricio Aniche
|
Chris Langhout and Maur\'icio Aniche
|
Atoms of Confusion in Java
| null | null | null | null |
cs.SE
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Although writing code seems trivial at times, problems arise when humans
misinterpret what the code actually does. One of the potential causes are
"atoms of confusion", the smallest possible patterns of misinterpretable source
code. Previous research has investigated the impact of atoms of confusion in C
code. Results show that developers make significantly more mistakes in code
where atoms are present. In this paper, we replicate the work of Gopstein et
al. to the Java language. After deriving a set of atoms of confusion for Java,
we perform a two-phase experiment with 132 computer science students (i.e.,
novice developers). Our results show that participants are 2.7 up to 56 times
more likely to make mistakes in code snippets affected by 7 out of the 14
studied atoms of confusion, and when faced with both versions of the code
snippets, participants perceived the version affected by the atom of confusion
to be more confusing and/or less readable in 10 out of the 14 studied atoms of
confusion.
|
[
{
"created": "Mon, 8 Mar 2021 09:04:05 GMT",
"version": "v1"
},
{
"created": "Wed, 10 Mar 2021 07:31:27 GMT",
"version": "v2"
}
] |
2021-03-11
|
[
[
"Langhout",
"Chris",
""
],
[
"Aniche",
"Maurício",
""
]
] |
Although writing code seems trivial at times, problems arise when humans misinterpret what the code actually does. One of the potential causes are "atoms of confusion", the smallest possible patterns of misinterpretable source code. Previous research has investigated the impact of atoms of confusion in C code. Results show that developers make significantly more mistakes in code where atoms are present. In this paper, we replicate the work of Gopstein et al. to the Java language. After deriving a set of atoms of confusion for Java, we perform a two-phase experiment with 132 computer science students (i.e., novice developers). Our results show that participants are 2.7 up to 56 times more likely to make mistakes in code snippets affected by 7 out of the 14 studied atoms of confusion, and when faced with both versions of the code snippets, participants perceived the version affected by the atom of confusion to be more confusing and/or less readable in 10 out of the 14 studied atoms of confusion.
|
1003.4627
|
Dejan Spasov
|
Dejan Spasov
|
Unique and Minimum Distance Decoding of Linear Codes with Reduced
Complexity
| null | null | null | null |
cs.IT math.IT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We show that for (systematic) linear codes the time complexity of unique
decoding is O(n^{2}q^{nRH(delta/2/R)}) and the time complexity of minimum
distance decoding is O(n^{2}q^{nRH(delta/R)}). The proposed algorithm inspects
all error patterns in the information set of the received message of weight
less than d/2 or d, respectively.
|
[
{
"created": "Wed, 24 Mar 2010 12:45:27 GMT",
"version": "v1"
}
] |
2010-03-25
|
[
[
"Spasov",
"Dejan",
""
]
] |
We show that for (systematic) linear codes the time complexity of unique decoding is O(n^{2}q^{nRH(delta/2/R)}) and the time complexity of minimum distance decoding is O(n^{2}q^{nRH(delta/R)}). The proposed algorithm inspects all error patterns in the information set of the received message of weight less than d/2 or d, respectively.
|
2112.04036
|
Mohammad Wardat
|
Mohammad Wardat, Breno Dantas Cruz, Wei Le, Hridesh Rajan
|
DeepDiagnosis: Automatically Diagnosing Faults and Recommending
Actionable Fixes in Deep Learning Programs
|
Accepted at ICSE 2022
| null | null | null |
cs.SE cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
Deep Neural Networks (DNNs) are used in a wide variety of applications.
However, as in any software application, DNN-based apps are afflicted with
bugs. Previous work observed that DNN bug fix patterns are different from
traditional bug fix patterns. Furthermore, those buggy models are non-trivial
to diagnose and fix due to inexplicit errors with several options to fix them.
To support developers in locating and fixing bugs, we propose DeepDiagnosis, a
novel debugging approach that localizes the faults, reports error symptoms and
suggests fixes for DNN programs. In the first phase, our technique monitors a
training model, periodically checking for eight types of error conditions.
Then, in case of problems, it reports messages containing sufficient
information to perform actionable repairs to the model. In the evaluation, we
thoroughly examine 444 models -53 real-world from GitHub and Stack Overflow,
and 391 curated by AUTOTRAINER. DeepDiagnosis provides superior accuracy when
compared to UMLUAT and DeepLocalize. Our technique is faster than AUTOTRAINER
for fault localization. The results show that our approach can support
additional types of models, while state-of-the-art was only able to handle
classification ones. Our technique was able to report bugs that do not manifest
as numerical errors during training. Also, it can provide actionable insights
for fix whereas DeepLocalize can only report faults that lead to numerical
errors during training. DeepDiagnosis manifests the best capabilities of fault
detection, bug localization, and symptoms identification when compared to other
approaches.
|
[
{
"created": "Tue, 7 Dec 2021 23:15:23 GMT",
"version": "v1"
}
] |
2021-12-09
|
[
[
"Wardat",
"Mohammad",
""
],
[
"Cruz",
"Breno Dantas",
""
],
[
"Le",
"Wei",
""
],
[
"Rajan",
"Hridesh",
""
]
] |
Deep Neural Networks (DNNs) are used in a wide variety of applications. However, as in any software application, DNN-based apps are afflicted with bugs. Previous work observed that DNN bug fix patterns are different from traditional bug fix patterns. Furthermore, those buggy models are non-trivial to diagnose and fix due to inexplicit errors with several options to fix them. To support developers in locating and fixing bugs, we propose DeepDiagnosis, a novel debugging approach that localizes the faults, reports error symptoms and suggests fixes for DNN programs. In the first phase, our technique monitors a training model, periodically checking for eight types of error conditions. Then, in case of problems, it reports messages containing sufficient information to perform actionable repairs to the model. In the evaluation, we thoroughly examine 444 models -53 real-world from GitHub and Stack Overflow, and 391 curated by AUTOTRAINER. DeepDiagnosis provides superior accuracy when compared to UMLUAT and DeepLocalize. Our technique is faster than AUTOTRAINER for fault localization. The results show that our approach can support additional types of models, while state-of-the-art was only able to handle classification ones. Our technique was able to report bugs that do not manifest as numerical errors during training. Also, it can provide actionable insights for fix whereas DeepLocalize can only report faults that lead to numerical errors during training. DeepDiagnosis manifests the best capabilities of fault detection, bug localization, and symptoms identification when compared to other approaches.
|
2407.11852
|
Marcel Parciak
|
Marcel Parciak, Brecht Vandevoort, Frank Neven, Liesbet M. Peeters,
Stijn Vansummeren
|
Schema Matching with Large Language Models: an Experimental Study
|
Accepted at the 2nd International Workshop on Tabular Data Analysis
(TaDA24), collocated with the 50th International Conference on Very Large
Data Bases (VLDB 2024) Guangzhou, China - August 29, 2024
| null | null | null |
cs.DB cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
Large Language Models (LLMs) have shown useful applications in a variety of
tasks, including data wrangling. In this paper, we investigate the use of an
off-the-shelf LLM for schema matching. Our objective is to identify semantic
correspondences between elements of two relational schemas using only names and
descriptions. Using a newly created benchmark from the health domain, we
propose different so-called task scopes. These are methods for prompting the
LLM to do schema matching, which vary in the amount of context information
contained in the prompt. Using these task scopes we compare LLM-based schema
matching against a string similarity baseline, investigating matching quality,
verification effort, decisiveness, and complementarity of the approaches. We
find that matching quality suffers from a lack of context information, but also
from providing too much context information. In general, using newer LLM
versions increases decisiveness. We identify task scopes that have acceptable
verification effort and succeed in identifying a significant number of true
semantic matches. Our study shows that LLMs have potential in bootstrapping the
schema matching process and are able to assist data engineers in speeding up
this task solely based on schema element names and descriptions without the
need for data instances.
|
[
{
"created": "Tue, 16 Jul 2024 15:33:00 GMT",
"version": "v1"
}
] |
2024-07-17
|
[
[
"Parciak",
"Marcel",
""
],
[
"Vandevoort",
"Brecht",
""
],
[
"Neven",
"Frank",
""
],
[
"Peeters",
"Liesbet M.",
""
],
[
"Vansummeren",
"Stijn",
""
]
] |
Large Language Models (LLMs) have shown useful applications in a variety of tasks, including data wrangling. In this paper, we investigate the use of an off-the-shelf LLM for schema matching. Our objective is to identify semantic correspondences between elements of two relational schemas using only names and descriptions. Using a newly created benchmark from the health domain, we propose different so-called task scopes. These are methods for prompting the LLM to do schema matching, which vary in the amount of context information contained in the prompt. Using these task scopes we compare LLM-based schema matching against a string similarity baseline, investigating matching quality, verification effort, decisiveness, and complementarity of the approaches. We find that matching quality suffers from a lack of context information, but also from providing too much context information. In general, using newer LLM versions increases decisiveness. We identify task scopes that have acceptable verification effort and succeed in identifying a significant number of true semantic matches. Our study shows that LLMs have potential in bootstrapping the schema matching process and are able to assist data engineers in speeding up this task solely based on schema element names and descriptions without the need for data instances.
|
2301.03364
|
Rodrigo Hernang\'omez
|
Rodrigo Hernang\'omez, Alexandros Palaios, Cara Watermann, Daniel
Sch\"aufele, Philipp Geuer, Rafail Ismayilov, Mohammad Parvini, Anton Krause,
Martin Kasparick, Thomas Neugebauer, Oscar D. Ramos-Cantor, Hugues
Tchouankem, Jose Leon Calvo, Bo Chen, Gerhard Fettweis, S{\l}awomir
Sta\'nczak
|
Toward an AI-enabled Connected Industry: AGV Communication and Sensor
Measurement Datasets
|
7 pages, 3 figures. Published at IEEE Communications Magazine. IEEE
Copyright protected. Datasets available at
https://ieee-dataport.org/open-access/ai4mobile-industrial-wireless-datasets-iv2v-and-iv2i
|
in IEEE Communications Magazine, vol. 62, no. 4, pp. 90-95, April
2024
|
10.1109/MCOM.001.2300494
| null |
cs.NI cs.AI cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
This paper presents two wireless measurement campaigns in industrial
testbeds: industrial Vehicle-to-vehicle (iV2V) and industrial
Vehicle-to-infrastructure plus Sensor (iV2I+), together with detailed
information about the two captured datasets. iV2V covers sidelink communication
scenarios between Automated Guided Vehicles (AGVs), while iV2I+ is conducted at
an industrial setting where an autonomous cleaning robot is connected to a
private cellular network. The combination of different communication
technologies within a common measurement methodology provides insights that can
be exploited by Machine Learning (ML) for tasks such as fingerprinting,
line-of-sight detection, prediction of quality of service or link selection.
Moreover, the datasets are publicly available, labelled and prefiltered for
fast on-boarding and applicability.
|
[
{
"created": "Tue, 20 Dec 2022 15:04:20 GMT",
"version": "v1"
},
{
"created": "Tue, 10 Jan 2023 11:29:39 GMT",
"version": "v2"
},
{
"created": "Mon, 20 Mar 2023 13:41:37 GMT",
"version": "v3"
},
{
"created": "Tue, 29 Aug 2023 11:18:43 GMT",
"version": "v4"
},
{
"created": "Mon, 15 Apr 2024 11:42:03 GMT",
"version": "v5"
}
] |
2024-04-16
|
[
[
"Hernangómez",
"Rodrigo",
""
],
[
"Palaios",
"Alexandros",
""
],
[
"Watermann",
"Cara",
""
],
[
"Schäufele",
"Daniel",
""
],
[
"Geuer",
"Philipp",
""
],
[
"Ismayilov",
"Rafail",
""
],
[
"Parvini",
"Mohammad",
""
],
[
"Krause",
"Anton",
""
],
[
"Kasparick",
"Martin",
""
],
[
"Neugebauer",
"Thomas",
""
],
[
"Ramos-Cantor",
"Oscar D.",
""
],
[
"Tchouankem",
"Hugues",
""
],
[
"Calvo",
"Jose Leon",
""
],
[
"Chen",
"Bo",
""
],
[
"Fettweis",
"Gerhard",
""
],
[
"Stańczak",
"Sławomir",
""
]
] |
This paper presents two wireless measurement campaigns in industrial testbeds: industrial Vehicle-to-vehicle (iV2V) and industrial Vehicle-to-infrastructure plus Sensor (iV2I+), together with detailed information about the two captured datasets. iV2V covers sidelink communication scenarios between Automated Guided Vehicles (AGVs), while iV2I+ is conducted at an industrial setting where an autonomous cleaning robot is connected to a private cellular network. The combination of different communication technologies within a common measurement methodology provides insights that can be exploited by Machine Learning (ML) for tasks such as fingerprinting, line-of-sight detection, prediction of quality of service or link selection. Moreover, the datasets are publicly available, labelled and prefiltered for fast on-boarding and applicability.
|
2109.10254
|
Youngseog Chung
|
Youngseog Chung, Ian Char, Han Guo, Jeff Schneider, Willie Neiswanger
|
Uncertainty Toolbox: an Open-Source Library for Assessing, Visualizing,
and Improving Uncertainty Quantification
| null | null | null | null |
cs.LG stat.ML
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
With increasing deployment of machine learning systems in various real-world
tasks, there is a greater need for accurate quantification of predictive
uncertainty. While the common goal in uncertainty quantification (UQ) in
machine learning is to approximate the true distribution of the target data,
many works in UQ tend to be disjoint in the evaluation metrics utilized, and
disparate implementations for each metric lead to numerical results that are
not directly comparable across different works. To address this, we introduce
Uncertainty Toolbox, an open-source python library that helps to assess,
visualize, and improve UQ. Uncertainty Toolbox additionally provides
pedagogical resources, such as a glossary of key terms and an organized
collection of key paper references. We hope that this toolbox is useful for
accelerating and uniting research efforts in uncertainty in machine learning.
|
[
{
"created": "Tue, 21 Sep 2021 15:32:06 GMT",
"version": "v1"
}
] |
2021-09-22
|
[
[
"Chung",
"Youngseog",
""
],
[
"Char",
"Ian",
""
],
[
"Guo",
"Han",
""
],
[
"Schneider",
"Jeff",
""
],
[
"Neiswanger",
"Willie",
""
]
] |
With increasing deployment of machine learning systems in various real-world tasks, there is a greater need for accurate quantification of predictive uncertainty. While the common goal in uncertainty quantification (UQ) in machine learning is to approximate the true distribution of the target data, many works in UQ tend to be disjoint in the evaluation metrics utilized, and disparate implementations for each metric lead to numerical results that are not directly comparable across different works. To address this, we introduce Uncertainty Toolbox, an open-source python library that helps to assess, visualize, and improve UQ. Uncertainty Toolbox additionally provides pedagogical resources, such as a glossary of key terms and an organized collection of key paper references. We hope that this toolbox is useful for accelerating and uniting research efforts in uncertainty in machine learning.
|
2311.14007
|
Liangrun Da
|
Liangrun Da, Martin Kleppmann
|
Extending JSON CRDTs with Move Operations
|
7 pages, 4 figures
| null |
10.1145/3642976.3653030
| null |
cs.DC
|
http://creativecommons.org/licenses/by/4.0/
|
Conflict-Free Replicated Data Types (CRDTs) for JSON allow users to
concurrently update a JSON document and automatically merge the updates into a
consistent state. Moving a subtree in a map or reordering elements in a list
within a JSON CRDT is challenging: naive merge algorithms may introduce
unexpected results such as duplicates or cycles. In this paper, we introduce an
algorithm for move operations in a JSON CRDT that handles the interaction with
concurrent non-move operations, and uses novel optimisations to improve
performance. We plan to integrate this algorithm into the Automerge CRDT
library.
|
[
{
"created": "Thu, 23 Nov 2023 13:48:52 GMT",
"version": "v1"
},
{
"created": "Tue, 19 Mar 2024 21:05:16 GMT",
"version": "v2"
}
] |
2024-03-21
|
[
[
"Da",
"Liangrun",
""
],
[
"Kleppmann",
"Martin",
""
]
] |
Conflict-Free Replicated Data Types (CRDTs) for JSON allow users to concurrently update a JSON document and automatically merge the updates into a consistent state. Moving a subtree in a map or reordering elements in a list within a JSON CRDT is challenging: naive merge algorithms may introduce unexpected results such as duplicates or cycles. In this paper, we introduce an algorithm for move operations in a JSON CRDT that handles the interaction with concurrent non-move operations, and uses novel optimisations to improve performance. We plan to integrate this algorithm into the Automerge CRDT library.
|
1601.04689
|
Santhosh Kumar
|
Shrinivas Kudekar, Santhosh Kumar, Marco Mondelli, Henry D. Pfister,
Eren \c{S}a\c{s}o\u{g}lu, R\"udiger Urbanke
|
Reed-Muller Codes Achieve Capacity on Erasure Channels
|
This article combines our previous articles arXiv:1505.05123 and
arXiv:1505.05831
| null | null | null |
cs.IT math.IT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We introduce a new approach to proving that a sequence of deterministic
linear codes achieves capacity on an erasure channel under maximum a posteriori
decoding. Rather than relying on the precise structure of the codes our method
exploits code symmetry. In particular, the technique applies to any sequence of
linear codes where the blocklengths are strictly increasing, the code rates
converge, and the permutation group of each code is doubly transitive. In other
words, we show that symmetry alone implies near-optimal performance.
An important consequence of this result is that a sequence of Reed-Muller
codes with increasing blocklength and converging rate achieves capacity. This
possibility has been suggested previously in the literature but it has only
been proven for cases where the limiting code rate is 0 or 1. Moreover, these
results extend naturally to all affine-invariant codes and, thus, to extended
primitive narrow-sense BCH codes. This also resolves, in the affirmative, the
existence question for capacity-achieving sequences of binary cyclic codes. The
primary tools used in the proof are the sharp threshold property for symmetric
monotone boolean functions and the area theorem for extrinsic information
transfer functions.
|
[
{
"created": "Mon, 18 Jan 2016 20:50:08 GMT",
"version": "v1"
}
] |
2016-01-19
|
[
[
"Kudekar",
"Shrinivas",
""
],
[
"Kumar",
"Santhosh",
""
],
[
"Mondelli",
"Marco",
""
],
[
"Pfister",
"Henry D.",
""
],
[
"Şaşoğlu",
"Eren",
""
],
[
"Urbanke",
"Rüdiger",
""
]
] |
We introduce a new approach to proving that a sequence of deterministic linear codes achieves capacity on an erasure channel under maximum a posteriori decoding. Rather than relying on the precise structure of the codes our method exploits code symmetry. In particular, the technique applies to any sequence of linear codes where the blocklengths are strictly increasing, the code rates converge, and the permutation group of each code is doubly transitive. In other words, we show that symmetry alone implies near-optimal performance. An important consequence of this result is that a sequence of Reed-Muller codes with increasing blocklength and converging rate achieves capacity. This possibility has been suggested previously in the literature but it has only been proven for cases where the limiting code rate is 0 or 1. Moreover, these results extend naturally to all affine-invariant codes and, thus, to extended primitive narrow-sense BCH codes. This also resolves, in the affirmative, the existence question for capacity-achieving sequences of binary cyclic codes. The primary tools used in the proof are the sharp threshold property for symmetric monotone boolean functions and the area theorem for extrinsic information transfer functions.
|
2306.01940
|
Elijah Pelofske
|
Kyle Henke, Elijah Pelofske, Georg Hahn, Garrett T. Kenyon
|
Sampling binary sparse coding QUBO models using a spiking neuromorphic
processor
| null | null |
10.1145/3589737.3606003
| null |
cs.NE cs.CV cs.ET cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We consider the problem of computing a sparse binary representation of an
image. To be precise, given an image and an overcomplete, non-orthonormal
basis, we aim to find a sparse binary vector indicating the minimal set of
basis vectors that when added together best reconstruct the given input. We
formulate this problem with an $L_2$ loss on the reconstruction error, and an
$L_0$ (or, equivalently, an $L_1$) loss on the binary vector enforcing
sparsity. This yields a so-called Quadratic Unconstrained Binary Optimization
(QUBO) problem, whose solution is generally NP-hard to find. The contribution
of this work is twofold. First, the method of unsupervised and unnormalized
dictionary feature learning for a desired sparsity level to best match the data
is presented. Second, the binary sparse coding problem is then solved on the
Loihi 1 neuromorphic chip by the use of stochastic networks of neurons to
traverse the non-convex energy landscape. The solutions are benchmarked against
the classical heuristic simulated annealing. We demonstrate neuromorphic
computing is suitable for sampling low energy solutions of binary sparse coding
QUBO models, and although Loihi 1 is capable of sampling very sparse solutions
of the QUBO models, there needs to be improvement in the implementation in
order to be competitive with simulated annealing.
|
[
{
"created": "Fri, 2 Jun 2023 22:47:18 GMT",
"version": "v1"
},
{
"created": "Wed, 2 Aug 2023 16:55:29 GMT",
"version": "v2"
}
] |
2023-11-28
|
[
[
"Henke",
"Kyle",
""
],
[
"Pelofske",
"Elijah",
""
],
[
"Hahn",
"Georg",
""
],
[
"Kenyon",
"Garrett T.",
""
]
] |
We consider the problem of computing a sparse binary representation of an image. To be precise, given an image and an overcomplete, non-orthonormal basis, we aim to find a sparse binary vector indicating the minimal set of basis vectors that when added together best reconstruct the given input. We formulate this problem with an $L_2$ loss on the reconstruction error, and an $L_0$ (or, equivalently, an $L_1$) loss on the binary vector enforcing sparsity. This yields a so-called Quadratic Unconstrained Binary Optimization (QUBO) problem, whose solution is generally NP-hard to find. The contribution of this work is twofold. First, the method of unsupervised and unnormalized dictionary feature learning for a desired sparsity level to best match the data is presented. Second, the binary sparse coding problem is then solved on the Loihi 1 neuromorphic chip by the use of stochastic networks of neurons to traverse the non-convex energy landscape. The solutions are benchmarked against the classical heuristic simulated annealing. We demonstrate neuromorphic computing is suitable for sampling low energy solutions of binary sparse coding QUBO models, and although Loihi 1 is capable of sampling very sparse solutions of the QUBO models, there needs to be improvement in the implementation in order to be competitive with simulated annealing.
|
2001.01491
|
Waqas Ahmed
|
Saira Khan, Khalid Iqbal, Safi Faizullah, Muhammad Fahad, Jawad Ali,
Waqas Ahmed
|
Clustering based Privacy Preserving of Big Data using Fuzzification and
Anonymization Operation
|
08 Page, 07 figures
|
International Journal of Advanced Computer Science and
Applications, Volume 10 Issue 12, 2019
|
10.14569/IJACSA.2019.0101239
| null |
cs.DB cs.CR
|
http://creativecommons.org/licenses/by/4.0/
|
Big Data is used by data miner for analysis purpose which may contain
sensitive information. During the procedures it raises certain privacy
challenges for researchers. The existing privacy preserving methods use
different algorithms that results into limitation of data reconstruction while
securing the sensitive data. This paper presents a clustering based privacy
preservation probabilistic model of big data to secure sensitive
information..model to attain minimum perturbation and maximum privacy. In our
model, sensitive information is secured after identifying the sensitive data
from data clusters to modify or generalize it.The resulting dataset is analysed
to calculate the accuracy level of our model in terms of hidden data, lossed
data as result of reconstruction. Extensive experiements are carried out in
order to demonstrate the results of our proposed model. Clustering based
Privacy preservation of individual data in big data with minimum perturbation
and successful reconstruction highlights the significance of our model in
addition to the use of standard performance evaluation measures.
|
[
{
"created": "Mon, 6 Jan 2020 11:31:12 GMT",
"version": "v1"
}
] |
2020-01-07
|
[
[
"Khan",
"Saira",
""
],
[
"Iqbal",
"Khalid",
""
],
[
"Faizullah",
"Safi",
""
],
[
"Fahad",
"Muhammad",
""
],
[
"Ali",
"Jawad",
""
],
[
"Ahmed",
"Waqas",
""
]
] |
Big Data is used by data miner for analysis purpose which may contain sensitive information. During the procedures it raises certain privacy challenges for researchers. The existing privacy preserving methods use different algorithms that results into limitation of data reconstruction while securing the sensitive data. This paper presents a clustering based privacy preservation probabilistic model of big data to secure sensitive information..model to attain minimum perturbation and maximum privacy. In our model, sensitive information is secured after identifying the sensitive data from data clusters to modify or generalize it.The resulting dataset is analysed to calculate the accuracy level of our model in terms of hidden data, lossed data as result of reconstruction. Extensive experiements are carried out in order to demonstrate the results of our proposed model. Clustering based Privacy preservation of individual data in big data with minimum perturbation and successful reconstruction highlights the significance of our model in addition to the use of standard performance evaluation measures.
|
2211.09783
|
Yulong Chen
|
Yulong Chen, Yang Liu, Ruochen Xu, Ziyi Yang, Chenguang Zhu, Michael
Zeng, Yue Zhang
|
UniSumm and SummZoo: Unified Model and Diverse Benchmark for Few-Shot
Summarization
|
ACL2023 main conference
| null | null | null |
cs.CL
|
http://creativecommons.org/licenses/by/4.0/
|
The high annotation costs and diverse demands of various summarization tasks
motivate the development of few-shot summarization. However, despite the
emergence of many summarization tasks and datasets, the current training
paradigm for few-shot summarization systems ignores potentially shareable
knowledge in heterogeneous datasets. To this end, we propose \textsc{UniSumm},
a unified few-shot summarization model pre-trained with multiple summarization
tasks and can be prefix-tuned to excel at any few-shot summarization task.
Meanwhile, to better evaluate few-shot summarizers, under the principles of
diversity and robustness, we assemble and release a new benchmark
\textsc{SummZoo}. It consists of $8$ summarization tasks with multiple sets of
few-shot samples for each task, covering diverse domains. Experimental results
and analysis show that \textsc{UniSumm} outperforms strong baselines by a large
margin across all sub-tasks in \textsc{SummZoo} under both automatic and human
evaluations and achieves comparable results in human evaluation compared with a
GPT-3.5 model.
|
[
{
"created": "Thu, 17 Nov 2022 18:54:47 GMT",
"version": "v1"
},
{
"created": "Mon, 21 Nov 2022 15:16:40 GMT",
"version": "v2"
},
{
"created": "Tue, 6 Dec 2022 08:54:22 GMT",
"version": "v3"
},
{
"created": "Tue, 13 Dec 2022 14:57:14 GMT",
"version": "v4"
},
{
"created": "Mon, 19 Dec 2022 05:15:58 GMT",
"version": "v5"
},
{
"created": "Sat, 27 May 2023 19:28:00 GMT",
"version": "v6"
}
] |
2023-05-30
|
[
[
"Chen",
"Yulong",
""
],
[
"Liu",
"Yang",
""
],
[
"Xu",
"Ruochen",
""
],
[
"Yang",
"Ziyi",
""
],
[
"Zhu",
"Chenguang",
""
],
[
"Zeng",
"Michael",
""
],
[
"Zhang",
"Yue",
""
]
] |
The high annotation costs and diverse demands of various summarization tasks motivate the development of few-shot summarization. However, despite the emergence of many summarization tasks and datasets, the current training paradigm for few-shot summarization systems ignores potentially shareable knowledge in heterogeneous datasets. To this end, we propose \textsc{UniSumm}, a unified few-shot summarization model pre-trained with multiple summarization tasks and can be prefix-tuned to excel at any few-shot summarization task. Meanwhile, to better evaluate few-shot summarizers, under the principles of diversity and robustness, we assemble and release a new benchmark \textsc{SummZoo}. It consists of $8$ summarization tasks with multiple sets of few-shot samples for each task, covering diverse domains. Experimental results and analysis show that \textsc{UniSumm} outperforms strong baselines by a large margin across all sub-tasks in \textsc{SummZoo} under both automatic and human evaluations and achieves comparable results in human evaluation compared with a GPT-3.5 model.
|
2406.14756
|
Huitong Pan
|
Huitong Pan and Qi Zhang and Cornelia Caragea and Eduard Dragut and
Longin Jan Latecki
|
SciDMT: A Large-Scale Corpus for Detecting Scientific Mentions
|
LREC/COLING 2024
|
LREC-COLING. (2024) 14407-14417
| null | null |
cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
We present SciDMT, an enhanced and expanded corpus for scientific mention
detection, offering a significant advancement over existing related resources.
SciDMT contains annotated scientific documents for datasets (D), methods (M),
and tasks (T). The corpus consists of two components: 1) the SciDMT main
corpus, which includes 48 thousand scientific articles with over 1.8 million
weakly annotated mention annotations in the format of in-text span, and 2) an
evaluation set, which comprises 100 scientific articles manually annotated for
evaluation purposes. To the best of our knowledge, SciDMT is the largest corpus
for scientific entity mention detection. The corpus's scale and diversity are
instrumental in developing and refining models for tasks such as indexing
scientific papers, enhancing information retrieval, and improving the
accessibility of scientific knowledge. We demonstrate the corpus's utility
through experiments with advanced deep learning architectures like SciBERT and
GPT-3.5. Our findings establish performance baselines and highlight unresolved
challenges in scientific mention detection. SciDMT serves as a robust benchmark
for the research community, encouraging the development of innovative models to
further the field of scientific information extraction.
|
[
{
"created": "Thu, 20 Jun 2024 22:03:21 GMT",
"version": "v1"
}
] |
2024-06-24
|
[
[
"Pan",
"Huitong",
""
],
[
"Zhang",
"Qi",
""
],
[
"Caragea",
"Cornelia",
""
],
[
"Dragut",
"Eduard",
""
],
[
"Latecki",
"Longin Jan",
""
]
] |
We present SciDMT, an enhanced and expanded corpus for scientific mention detection, offering a significant advancement over existing related resources. SciDMT contains annotated scientific documents for datasets (D), methods (M), and tasks (T). The corpus consists of two components: 1) the SciDMT main corpus, which includes 48 thousand scientific articles with over 1.8 million weakly annotated mention annotations in the format of in-text span, and 2) an evaluation set, which comprises 100 scientific articles manually annotated for evaluation purposes. To the best of our knowledge, SciDMT is the largest corpus for scientific entity mention detection. The corpus's scale and diversity are instrumental in developing and refining models for tasks such as indexing scientific papers, enhancing information retrieval, and improving the accessibility of scientific knowledge. We demonstrate the corpus's utility through experiments with advanced deep learning architectures like SciBERT and GPT-3.5. Our findings establish performance baselines and highlight unresolved challenges in scientific mention detection. SciDMT serves as a robust benchmark for the research community, encouraging the development of innovative models to further the field of scientific information extraction.
|
2406.17952
|
Andrew Dennehy
|
Andrew Dennehy, Xiaoyu Zou, Shabnam J. Semnani, Yuri Fialko, Alexander
Cloninger
|
LINSCAN -- A Linearity Based Clustering Algorithm
| null | null | null | null |
cs.LG cs.CG
|
http://creativecommons.org/licenses/by/4.0/
|
DBSCAN and OPTICS are powerful algorithms for identifying clusters of points
in domains where few assumptions can be made about the structure of the data.
In this paper, we leverage these strengths and introduce a new algorithm,
LINSCAN, designed to seek lineated clusters that are difficult to find and
isolate with existing methods. In particular, by embedding points as normal
distributions approximating their local neighborhoods and leveraging a distance
function derived from the Kullback Leibler Divergence, LINSCAN can detect and
distinguish lineated clusters that are spatially close but have orthogonal
covariances. We demonstrate how LINSCAN can be applied to seismic data to
identify active faults, including intersecting faults, and determine their
orientation. Finally, we discuss the properties a generalization of DBSCAN and
OPTICS must have in order to retain the stability benefits of these algorithms.
|
[
{
"created": "Tue, 25 Jun 2024 21:58:37 GMT",
"version": "v1"
}
] |
2024-06-27
|
[
[
"Dennehy",
"Andrew",
""
],
[
"Zou",
"Xiaoyu",
""
],
[
"Semnani",
"Shabnam J.",
""
],
[
"Fialko",
"Yuri",
""
],
[
"Cloninger",
"Alexander",
""
]
] |
DBSCAN and OPTICS are powerful algorithms for identifying clusters of points in domains where few assumptions can be made about the structure of the data. In this paper, we leverage these strengths and introduce a new algorithm, LINSCAN, designed to seek lineated clusters that are difficult to find and isolate with existing methods. In particular, by embedding points as normal distributions approximating their local neighborhoods and leveraging a distance function derived from the Kullback Leibler Divergence, LINSCAN can detect and distinguish lineated clusters that are spatially close but have orthogonal covariances. We demonstrate how LINSCAN can be applied to seismic data to identify active faults, including intersecting faults, and determine their orientation. Finally, we discuss the properties a generalization of DBSCAN and OPTICS must have in order to retain the stability benefits of these algorithms.
|
1610.02526
|
Arash Shaghaghi
|
Arash Shaghaghi and Mohamed Ali (Dali) Kaafar and Sandra Scott-Hayward
and Salil S. Kanhere and Sanjay Jha
|
Towards Policy Enforcement Point as a Service (PEPS)
|
This is a copy of the paper accepted at IEEE NFV-SDN'16. An extended
work based on this paper will be submitted to a journal
| null | null | null |
cs.CR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this paper, we coin the term Policy Enforcement as a Service (PEPS), which
enables the provision of innovative inter-layer and inter-domain Access
Control. We leverage the architecture of Software-Defined-Network (SDN) to
introduce a common network-level enforcement point, which is made available to
a range of access control systems. With our PEPS model, it is possible to have
a `defense in depth' protection model and drop unsuccessful access requests
before engaging the data provider (e.g. a database system). Moreover, the
current implementation of access control within the `trusted' perimeter of an
organization is no longer a restriction so that the potential for novel,
distributed and cooperative security services can be realized. We conduct an
analysis of the security requirements and technical challenges for implementing
Policy Enforcement as a Service. To illustrate the benefits of our proposal in
practice, we include a report on our prototype PEPS-enabled location-based
access control.
|
[
{
"created": "Sat, 8 Oct 2016 13:09:47 GMT",
"version": "v1"
}
] |
2016-10-11
|
[
[
"Shaghaghi",
"Arash",
"",
"Dali"
],
[
"Ali",
"Mohamed",
"",
"Dali"
],
[
"Kaafar",
"",
""
],
[
"Scott-Hayward",
"Sandra",
""
],
[
"Kanhere",
"Salil S.",
""
],
[
"Jha",
"Sanjay",
""
]
] |
In this paper, we coin the term Policy Enforcement as a Service (PEPS), which enables the provision of innovative inter-layer and inter-domain Access Control. We leverage the architecture of Software-Defined-Network (SDN) to introduce a common network-level enforcement point, which is made available to a range of access control systems. With our PEPS model, it is possible to have a `defense in depth' protection model and drop unsuccessful access requests before engaging the data provider (e.g. a database system). Moreover, the current implementation of access control within the `trusted' perimeter of an organization is no longer a restriction so that the potential for novel, distributed and cooperative security services can be realized. We conduct an analysis of the security requirements and technical challenges for implementing Policy Enforcement as a Service. To illustrate the benefits of our proposal in practice, we include a report on our prototype PEPS-enabled location-based access control.
|
2406.17424
|
Shinwoo An
|
Shinwoo An, Eunjin Oh and Jie Xue
|
Sparse Outerstring Graphs Have Logarithmic Treewidth
|
17pages, In ESA'24
| null | null | null |
cs.CG cs.DS
|
http://creativecommons.org/licenses/by/4.0/
|
An outerstring graph is the intersection graph of curves lying inside a disk
with one endpoint on the boundary of the disk. We show that an outerstring
graph with $n$ vertices has treewidth $O(\alpha\log n)$, where $\alpha$ denotes
the arboricity of the graph, with an almost matching lower bound of
$\Omega(\alpha \log (n/\alpha))$. As a corollary, we show that a
$t$-biclique-free outerstring graph has treewidth $O(t(\log t)\log n)$. This
leads to polynomial-time algorithms for most of the central NP-complete
problems such as \textsc{Independent Set}, \textsc{Vertex Cover},
\textsc{Dominating Set}, \textsc{Feedback Vertex Set}, \textsc{Coloring} for
sparse outerstring graphs. Also, we can obtain subexponential-time (exact,
parameterized, and approximation) algorithms for various NP-complete problems
such as \textsc{Vertex Cover}, \textsc{Feedback Vertex Set} and \textsc{Cycle
Packing} for (not necessarily sparse) outerstring graphs.
|
[
{
"created": "Tue, 25 Jun 2024 09:59:24 GMT",
"version": "v1"
}
] |
2024-06-26
|
[
[
"An",
"Shinwoo",
""
],
[
"Oh",
"Eunjin",
""
],
[
"Xue",
"Jie",
""
]
] |
An outerstring graph is the intersection graph of curves lying inside a disk with one endpoint on the boundary of the disk. We show that an outerstring graph with $n$ vertices has treewidth $O(\alpha\log n)$, where $\alpha$ denotes the arboricity of the graph, with an almost matching lower bound of $\Omega(\alpha \log (n/\alpha))$. As a corollary, we show that a $t$-biclique-free outerstring graph has treewidth $O(t(\log t)\log n)$. This leads to polynomial-time algorithms for most of the central NP-complete problems such as \textsc{Independent Set}, \textsc{Vertex Cover}, \textsc{Dominating Set}, \textsc{Feedback Vertex Set}, \textsc{Coloring} for sparse outerstring graphs. Also, we can obtain subexponential-time (exact, parameterized, and approximation) algorithms for various NP-complete problems such as \textsc{Vertex Cover}, \textsc{Feedback Vertex Set} and \textsc{Cycle Packing} for (not necessarily sparse) outerstring graphs.
|
1512.09032
|
Shankara Narayanan Krishna
|
Khushraj Madnani, Shankara Narayanan Krishna and Paritosh Pandya
|
Metric Temporal Logic with Counting
| null | null | null | null |
cs.LO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Ability to count number of occurrences of events within a specified time
interval is very useful in specification of resource bounded real time
computation. In this paper, we study an extension of Metric Temporal Logic
($\mathsf{MTL}$) with two different counting modalities called $\mathsf{C}$ and
$\mathsf{UT}$ (until with threshold), which enhance the expressive power of
$\mathsf{MTL}$ in orthogonal fashion. We confine ourselves only to the future
fragment of $\mathsf{MTL}$ interpreted in a pointwise manner over finite timed
words. We provide a comprehensive study of the expressive power of logic
$\mathsf{CTMTL}$ and its fragments using the technique of EF games extended
with suitable counting moves. Finally, as our main result, we establish the
decidability of $\mathsf{CTMTL}$ by giving an equisatisfiable reduction from
$\mathsf{CTMTL}$ to $\mathsf{MTL}$. The reduction provides one more example of
the use of temporal projections with oversampling introduced earlier for
proving decidability. Our reduction also implies that $\mathsf{MITL}$ extended
with $\mathsf{C}$ and $\mathsf{UT}$ modalities is elementarily decidable.
|
[
{
"created": "Wed, 30 Dec 2015 17:42:14 GMT",
"version": "v1"
}
] |
2015-12-31
|
[
[
"Madnani",
"Khushraj",
""
],
[
"Krishna",
"Shankara Narayanan",
""
],
[
"Pandya",
"Paritosh",
""
]
] |
Ability to count number of occurrences of events within a specified time interval is very useful in specification of resource bounded real time computation. In this paper, we study an extension of Metric Temporal Logic ($\mathsf{MTL}$) with two different counting modalities called $\mathsf{C}$ and $\mathsf{UT}$ (until with threshold), which enhance the expressive power of $\mathsf{MTL}$ in orthogonal fashion. We confine ourselves only to the future fragment of $\mathsf{MTL}$ interpreted in a pointwise manner over finite timed words. We provide a comprehensive study of the expressive power of logic $\mathsf{CTMTL}$ and its fragments using the technique of EF games extended with suitable counting moves. Finally, as our main result, we establish the decidability of $\mathsf{CTMTL}$ by giving an equisatisfiable reduction from $\mathsf{CTMTL}$ to $\mathsf{MTL}$. The reduction provides one more example of the use of temporal projections with oversampling introduced earlier for proving decidability. Our reduction also implies that $\mathsf{MITL}$ extended with $\mathsf{C}$ and $\mathsf{UT}$ modalities is elementarily decidable.
|
2403.11728
|
Kevin R\"osch
|
Johannes Fischer, Kevin R\"osch, Martin Lauer, Christoph Stiller
|
PITA: Physics-Informed Trajectory Autoencoder
| null | null | null | null |
cs.LG cs.RO
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
Validating robotic systems in safety-critical appli-cations requires testing
in many scenarios including rare edgecases that are unlikely to occur,
requiring to complement real-world testing with testing in simulation.
Generative models canbe used to augment real-world datasets with generated data
toproduce edge case scenarios by sampling in a learned latentspace.
Autoencoders can learn said latent representation for aspecific domain by
learning to reconstruct the input data froma lower-dimensional intermediate
representation. However, theresulting trajectories are not necessarily
physically plausible, butinstead typically contain noise that is not present in
the inputtrajectory. To resolve this issue, we propose the novel
Physics-Informed Trajectory Autoencoder (PITA) architecture, whichincorporates
a physical dynamics model into the loss functionof the autoencoder. This
results in smooth trajectories that notonly reconstruct the input trajectory
but also adhere to thephysical model. We evaluate PITA on a real-world dataset
ofvehicle trajectories and compare its performance to a normalautoencoder and a
state-of-the-art action-space autoencoder.
|
[
{
"created": "Mon, 18 Mar 2024 12:37:41 GMT",
"version": "v1"
}
] |
2024-03-19
|
[
[
"Fischer",
"Johannes",
""
],
[
"Rösch",
"Kevin",
""
],
[
"Lauer",
"Martin",
""
],
[
"Stiller",
"Christoph",
""
]
] |
Validating robotic systems in safety-critical appli-cations requires testing in many scenarios including rare edgecases that are unlikely to occur, requiring to complement real-world testing with testing in simulation. Generative models canbe used to augment real-world datasets with generated data toproduce edge case scenarios by sampling in a learned latentspace. Autoencoders can learn said latent representation for aspecific domain by learning to reconstruct the input data froma lower-dimensional intermediate representation. However, theresulting trajectories are not necessarily physically plausible, butinstead typically contain noise that is not present in the inputtrajectory. To resolve this issue, we propose the novel Physics-Informed Trajectory Autoencoder (PITA) architecture, whichincorporates a physical dynamics model into the loss functionof the autoencoder. This results in smooth trajectories that notonly reconstruct the input trajectory but also adhere to thephysical model. We evaluate PITA on a real-world dataset ofvehicle trajectories and compare its performance to a normalautoencoder and a state-of-the-art action-space autoencoder.
|
2405.02022
|
Michael Baddeley Dr
|
Burhanuddin Rangwala, Ava Powelson, Michael Baddeley and Israat Haque
|
STX-Vote: Improving Reliability with Bit Voting in Synchronous
Transmission-based IoT Networks
| null | null | null | null |
cs.NI
|
http://creativecommons.org/licenses/by/4.0/
|
Industrial Internet of Things (IIoT) networks must meet strict reliability,
latency, and low energy consumption requirements. However, traditional
low-power wireless protocols are ineffective in finding a sweet spot for
balancing these performance metrics. Recently, network flooding protocols based
on Synchronous Transmissions (STX) have been proposed for better performance in
reliability-critical IIoT, where simultaneous transmissions are possible
without packet collisions. STX-based protocols can offer a competitive edge
over routing-based protocols, particularly in dependability. However, they
notably suffer from the beating effect, a physical layer phenomenon that
results in sinusoidal interference across a packet and, consequently, packet
loss. Thus, we introduce STX-Vote, an error correction scheme that can handle
errors caused by beating effects. Importantly, we utilize transmission
redundancy already inherent within STX protocols so do not incur additional
on-air overhead. Through simulation, we demonstrate STX-Vote can provide a 40%
increase in reliability. We subsequently implement STX-Vote on nRF52840-DK
devices and perform extensive experiments. The results confirm that STX-Vote
improves reliability by 25-28% for BLE 5 PHYs and 8% for IEEE 802.15.4; thus,
it can complement existing error correction schemes.
|
[
{
"created": "Fri, 3 May 2024 11:54:16 GMT",
"version": "v1"
},
{
"created": "Mon, 12 Aug 2024 06:12:22 GMT",
"version": "v2"
},
{
"created": "Tue, 13 Aug 2024 07:39:24 GMT",
"version": "v3"
}
] |
2024-08-14
|
[
[
"Rangwala",
"Burhanuddin",
""
],
[
"Powelson",
"Ava",
""
],
[
"Baddeley",
"Michael",
""
],
[
"Haque",
"Israat",
""
]
] |
Industrial Internet of Things (IIoT) networks must meet strict reliability, latency, and low energy consumption requirements. However, traditional low-power wireless protocols are ineffective in finding a sweet spot for balancing these performance metrics. Recently, network flooding protocols based on Synchronous Transmissions (STX) have been proposed for better performance in reliability-critical IIoT, where simultaneous transmissions are possible without packet collisions. STX-based protocols can offer a competitive edge over routing-based protocols, particularly in dependability. However, they notably suffer from the beating effect, a physical layer phenomenon that results in sinusoidal interference across a packet and, consequently, packet loss. Thus, we introduce STX-Vote, an error correction scheme that can handle errors caused by beating effects. Importantly, we utilize transmission redundancy already inherent within STX protocols so do not incur additional on-air overhead. Through simulation, we demonstrate STX-Vote can provide a 40% increase in reliability. We subsequently implement STX-Vote on nRF52840-DK devices and perform extensive experiments. The results confirm that STX-Vote improves reliability by 25-28% for BLE 5 PHYs and 8% for IEEE 802.15.4; thus, it can complement existing error correction schemes.
|
1203.5689
|
Philipp Schaer
|
Philipp Schaer, Thomas L\"uke, Wilko van Hoek
|
Building Custom Term Suggestion Web Services with OAI-Harvested Open
Data
|
8 pages, 5 figures, presented at 2. DGI-Konferenz / 64. Jahrestagung
der DGI, D\"usseldorf, Germany on 2012-03-23
|
Social Media und Web Science: Das Web als Lebensraum. 2.
DGI-Konferenz / 64. Jahrestagung der DGI, D\"usseldorf, 22. bis 23. M\"arz
2012, Proceedings (2012), p. 389-396
| null | null |
cs.DL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The problem that the same information need can be expressed in a variety of
ways is especially true for scientific literature. Each scientific discipline
has its own domain-specific language and vocabulary. This language is coded
into documentary tools like thesauri or classifications that are used to
document and describe scientific documents. When we think of information
retrieval as "fundamentally a linguistic process" (Blair, 2003) users have to
be aware of the most relevant search terms - which are the controlled thesauri
terms the documents are described with. This can be achieved with so-called
search-term-recommenders (STR) that map free search terms of a lay user to
controlled vocabulary terms which can then be used as a term suggestion or to
do an automatic query expansion (Hienert, Schaer, Schaible, & Mayr, 2011).
State-of-the-art repository software systems like DSpace or EPrints already
offer some kind of term suggestion features in search or input forms but these
implementations only work as simple auto completion mechanisms that don't
incorporate any kind of semantic mapping. Such software systems would gain a
lot in terms of usability and data consistency if tools like the proposed
domain-specific STRs would be freely available. We aim to implement a rich
toolbox of web services (like the mentioned domain-specific STRs) to support
users and providers of online Digital Library (DL) or repository systems.
|
[
{
"created": "Mon, 26 Mar 2012 14:45:45 GMT",
"version": "v1"
}
] |
2012-03-27
|
[
[
"Schaer",
"Philipp",
""
],
[
"Lüke",
"Thomas",
""
],
[
"van Hoek",
"Wilko",
""
]
] |
The problem that the same information need can be expressed in a variety of ways is especially true for scientific literature. Each scientific discipline has its own domain-specific language and vocabulary. This language is coded into documentary tools like thesauri or classifications that are used to document and describe scientific documents. When we think of information retrieval as "fundamentally a linguistic process" (Blair, 2003) users have to be aware of the most relevant search terms - which are the controlled thesauri terms the documents are described with. This can be achieved with so-called search-term-recommenders (STR) that map free search terms of a lay user to controlled vocabulary terms which can then be used as a term suggestion or to do an automatic query expansion (Hienert, Schaer, Schaible, & Mayr, 2011). State-of-the-art repository software systems like DSpace or EPrints already offer some kind of term suggestion features in search or input forms but these implementations only work as simple auto completion mechanisms that don't incorporate any kind of semantic mapping. Such software systems would gain a lot in terms of usability and data consistency if tools like the proposed domain-specific STRs would be freely available. We aim to implement a rich toolbox of web services (like the mentioned domain-specific STRs) to support users and providers of online Digital Library (DL) or repository systems.
|
2010.11786
|
Benjamin Ricaud
|
Benjamin Ricaud, Nicolas Aspert and Volodymyr Miz
|
Spikyball sampling: Exploring large networks via an inhomogeneous
filtered diffusion
| null | null | null | null |
cs.SI cs.IR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Studying real-world networks such as social networks or web networks is a
challenge. These networks often combine a complex, highly connected structure
together with a large size. We propose a new approach for large scale networks
that is able to automatically sample user-defined relevant parts of a network.
Starting from a few selected places in the network and a reduced set of
expansion rules, the method adopts a filtered breadth-first search approach,
that expands through edges and nodes matching these properties. Moreover, the
expansion is performed over a random subset of neighbors at each step to
mitigate further the overwhelming number of connections that may exist in large
graphs. This carries the image of a "spiky" expansion. We show that this
approach generalize previous exploration sampling methods, such as Snowball or
Forest Fire and extend them. We demonstrate its ability to capture groups of
nodes with high interactions while discarding weakly connected nodes that are
often numerous in social networks and may hide important structures.
|
[
{
"created": "Thu, 22 Oct 2020 15:01:13 GMT",
"version": "v1"
}
] |
2020-10-23
|
[
[
"Ricaud",
"Benjamin",
""
],
[
"Aspert",
"Nicolas",
""
],
[
"Miz",
"Volodymyr",
""
]
] |
Studying real-world networks such as social networks or web networks is a challenge. These networks often combine a complex, highly connected structure together with a large size. We propose a new approach for large scale networks that is able to automatically sample user-defined relevant parts of a network. Starting from a few selected places in the network and a reduced set of expansion rules, the method adopts a filtered breadth-first search approach, that expands through edges and nodes matching these properties. Moreover, the expansion is performed over a random subset of neighbors at each step to mitigate further the overwhelming number of connections that may exist in large graphs. This carries the image of a "spiky" expansion. We show that this approach generalize previous exploration sampling methods, such as Snowball or Forest Fire and extend them. We demonstrate its ability to capture groups of nodes with high interactions while discarding weakly connected nodes that are often numerous in social networks and may hide important structures.
|
1809.03609
|
Mohamed Ibrahim
|
Mohamed R. Ibrahim, James Haworth, Tao Cheng
|
URBAN-i: From urban scenes to mapping slums, transport modes, and
pedestrians in cities using deep learning and computer vision
|
12 pages, 9 figures
| null |
10.1177/2399808319846517
| null |
cs.CV cs.AI
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
Within the burgeoning expansion of deep learning and computer vision across
the different fields of science, when it comes to urban development, deep
learning and computer vision applications are still limited towards the notions
of smart cities and autonomous vehicles. Indeed, a wide gap of knowledge
appears when it comes to cities and urban regions in less developed countries
where the chaos of informality is the dominant scheme. How can deep learning
and Artificial Intelligence (AI) untangle the complexities of informality to
advance urban modelling and our understanding of cities? Various questions and
debates can be raised concerning the future of cities of the North and the
South in the paradigm of AI and computer vision. In this paper, we introduce a
new method for multipurpose realistic-dynamic urban modelling relying on deep
learning and computer vision, using deep Convolutional Neural Networks (CNN),
to sense and detect informality and slums in urban scenes from aerial and
street view images in addition to detection of pedestrian and transport modes.
The model has been trained on images of urban scenes in cities across the
globe. The model shows a good validation of understanding a wide spectrum of
nuances among the planned and the unplanned regions, including informal and
slum areas. We attempt to advance urban modelling for better understanding the
dynamics of city developments. We also aim to exemplify the significant impacts
of AI in cities beyond how smart cities are discussed and perceived in the
mainstream. The algorithms of the URBAN-i model are fully-coded in Python
programming with the pre-trained deep learning models to be used as a tool for
mapping and city modelling in the various corner of the globe, including
informal settlements and slum regions.
|
[
{
"created": "Mon, 10 Sep 2018 21:49:38 GMT",
"version": "v1"
}
] |
2019-10-23
|
[
[
"Ibrahim",
"Mohamed R.",
""
],
[
"Haworth",
"James",
""
],
[
"Cheng",
"Tao",
""
]
] |
Within the burgeoning expansion of deep learning and computer vision across the different fields of science, when it comes to urban development, deep learning and computer vision applications are still limited towards the notions of smart cities and autonomous vehicles. Indeed, a wide gap of knowledge appears when it comes to cities and urban regions in less developed countries where the chaos of informality is the dominant scheme. How can deep learning and Artificial Intelligence (AI) untangle the complexities of informality to advance urban modelling and our understanding of cities? Various questions and debates can be raised concerning the future of cities of the North and the South in the paradigm of AI and computer vision. In this paper, we introduce a new method for multipurpose realistic-dynamic urban modelling relying on deep learning and computer vision, using deep Convolutional Neural Networks (CNN), to sense and detect informality and slums in urban scenes from aerial and street view images in addition to detection of pedestrian and transport modes. The model has been trained on images of urban scenes in cities across the globe. The model shows a good validation of understanding a wide spectrum of nuances among the planned and the unplanned regions, including informal and slum areas. We attempt to advance urban modelling for better understanding the dynamics of city developments. We also aim to exemplify the significant impacts of AI in cities beyond how smart cities are discussed and perceived in the mainstream. The algorithms of the URBAN-i model are fully-coded in Python programming with the pre-trained deep learning models to be used as a tool for mapping and city modelling in the various corner of the globe, including informal settlements and slum regions.
|
2306.11593
|
Luigi Celona
|
Simone Bianco and Luigi Celona and Marco Donzella and Paolo Napoletano
|
Improving Image Captioning Descriptiveness by Ranking and LLM-based
Fusion
| null | null | null | null |
cs.CV cs.AI cs.CL cs.DB cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
State-of-The-Art (SoTA) image captioning models often rely on the Microsoft
COCO (MS-COCO) dataset for training. This dataset contains annotations provided
by human annotators, who typically produce captions averaging around ten
tokens. However, this constraint presents a challenge in effectively capturing
complex scenes and conveying detailed information. Furthermore, captioning
models tend to exhibit bias towards the ``average'' caption, which captures
only the more general aspects. What would happen if we were able to
automatically generate longer captions, thereby making them more detailed?
Would these captions, evaluated by humans, be more or less representative of
the image content compared to the original MS-COCO captions? In this paper, we
present a novel approach to address previous challenges by showcasing how
captions generated from different SoTA models can be effectively fused,
resulting in richer captions. Our proposed method leverages existing models
from the literature, eliminating the need for additional training. Instead, it
utilizes an image-text based metric to rank the captions generated by SoTA
models for a given image. Subsequently, the top two captions are fused using a
Large Language Model (LLM). Experimental results demonstrate the effectiveness
of our approach, as the captions generated by our model exhibit higher
consistency with human judgment when evaluated on the MS-COCO test set. By
combining the strengths of various SoTA models, our method enhances the quality
and appeal of image captions, bridging the gap between automated systems and
the rich, informative nature of human-generated descriptions. This advance
opens up new possibilities for generating captions that are more suitable for
the training of both vision-language and captioning models.
|
[
{
"created": "Tue, 20 Jun 2023 15:13:02 GMT",
"version": "v1"
}
] |
2023-06-21
|
[
[
"Bianco",
"Simone",
""
],
[
"Celona",
"Luigi",
""
],
[
"Donzella",
"Marco",
""
],
[
"Napoletano",
"Paolo",
""
]
] |
State-of-The-Art (SoTA) image captioning models often rely on the Microsoft COCO (MS-COCO) dataset for training. This dataset contains annotations provided by human annotators, who typically produce captions averaging around ten tokens. However, this constraint presents a challenge in effectively capturing complex scenes and conveying detailed information. Furthermore, captioning models tend to exhibit bias towards the ``average'' caption, which captures only the more general aspects. What would happen if we were able to automatically generate longer captions, thereby making them more detailed? Would these captions, evaluated by humans, be more or less representative of the image content compared to the original MS-COCO captions? In this paper, we present a novel approach to address previous challenges by showcasing how captions generated from different SoTA models can be effectively fused, resulting in richer captions. Our proposed method leverages existing models from the literature, eliminating the need for additional training. Instead, it utilizes an image-text based metric to rank the captions generated by SoTA models for a given image. Subsequently, the top two captions are fused using a Large Language Model (LLM). Experimental results demonstrate the effectiveness of our approach, as the captions generated by our model exhibit higher consistency with human judgment when evaluated on the MS-COCO test set. By combining the strengths of various SoTA models, our method enhances the quality and appeal of image captions, bridging the gap between automated systems and the rich, informative nature of human-generated descriptions. This advance opens up new possibilities for generating captions that are more suitable for the training of both vision-language and captioning models.
|
2111.14994
|
Niki Hrovatin
|
Niki Hrovatin, Aleksandar To\v{s}i\'c, Michael Mrissa and Jernej
Vi\v{c}i\v{c}
|
A General Purpose Data and Query Privacy Preserving Protocol for
Wireless Sensor Networks
|
Submitted to IEEE IoT Journal, 18 pages, 16 figures
| null | null | null |
cs.CR cs.NI
|
http://creativecommons.org/licenses/by/4.0/
|
Wireless Sensor Networks (WSNs) are composed of a large number of spatially
distributed devices equipped with sensing technology and interlinked via radio
signaling. A WSN deployed for monitoring purposes can provide a ubiquitous view
over the monitored environment. However, the management of collected data is
very resource-consuming and raises security and privacy issues. In this paper,
we propose a privacy preserving protocol for collecting aggregated data from
WSNs. The protocol relies on the Onion Routing technique to provide uniformly
distributed network traffic and confine the knowledge a foreign actor can gain
from monitoring messages traveling the network. Our solution employs the
computing power of nodes in the network by conveying them general-purpose
computer code for in-situ processing and aggregation of data sourcing from
multiple sensor nodes. We complement our work with a simulation of the proposed
solution using the network simulator ns-3. Results of the simulation give an
overview of the scalability of the solution and highlight potential
constraints.
|
[
{
"created": "Mon, 29 Nov 2021 22:18:19 GMT",
"version": "v1"
}
] |
2021-12-01
|
[
[
"Hrovatin",
"Niki",
""
],
[
"Tošić",
"Aleksandar",
""
],
[
"Mrissa",
"Michael",
""
],
[
"Vičič",
"Jernej",
""
]
] |
Wireless Sensor Networks (WSNs) are composed of a large number of spatially distributed devices equipped with sensing technology and interlinked via radio signaling. A WSN deployed for monitoring purposes can provide a ubiquitous view over the monitored environment. However, the management of collected data is very resource-consuming and raises security and privacy issues. In this paper, we propose a privacy preserving protocol for collecting aggregated data from WSNs. The protocol relies on the Onion Routing technique to provide uniformly distributed network traffic and confine the knowledge a foreign actor can gain from monitoring messages traveling the network. Our solution employs the computing power of nodes in the network by conveying them general-purpose computer code for in-situ processing and aggregation of data sourcing from multiple sensor nodes. We complement our work with a simulation of the proposed solution using the network simulator ns-3. Results of the simulation give an overview of the scalability of the solution and highlight potential constraints.
|
2103.06819
|
Jieren Deng
|
Jieren Deng, Yijue Wang, Ji Li, Chao Shang, Hang Liu, Sanguthevar
Rajasekaran and Caiwen Ding
|
TAG: Gradient Attack on Transformer-based Language Models
|
Accepted to Findings of EMNLP 2021
| null | null | null |
cs.CR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Although federated learning has increasingly gained attention in terms of
effectively utilizing local devices for data privacy enhancement, recent
studies show that publicly shared gradients in the training process can reveal
the private training images (gradient leakage) to a third-party in computer
vision. We have, however, no systematic understanding of the gradient leakage
mechanism on the Transformer based language models. In this paper, as the first
attempt, we formulate the gradient attack problem on the Transformer-based
language models and propose a gradient attack algorithm, TAG, to reconstruct
the local training data. We develop a set of metrics to evaluate the
effectiveness of the proposed attack algorithm quantitatively. Experimental
results on Transformer, TinyBERT$_{4}$, TinyBERT$_{6}$, BERT$_{BASE}$, and
BERT$_{LARGE}$ using GLUE benchmark show that TAG works well on more weight
distributions in reconstructing training data and achieves 1.5$\times$ recover
rate and 2.5$\times$ ROUGE-2 over prior methods without the need of ground
truth label. TAG can obtain up to 90$\%$ data by attacking gradients in CoLA
dataset. In addition, TAG has a stronger adversary on large models, small
dictionary size, and small input length. We hope the proposed TAG will shed
some light on the privacy leakage problem in Transformer-based NLP models.
|
[
{
"created": "Thu, 11 Mar 2021 17:41:32 GMT",
"version": "v1"
},
{
"created": "Mon, 15 Mar 2021 03:08:57 GMT",
"version": "v2"
},
{
"created": "Tue, 16 Mar 2021 20:51:19 GMT",
"version": "v3"
},
{
"created": "Wed, 21 Apr 2021 04:04:18 GMT",
"version": "v4"
},
{
"created": "Fri, 10 Sep 2021 02:23:35 GMT",
"version": "v5"
},
{
"created": "Tue, 21 Sep 2021 17:58:26 GMT",
"version": "v6"
}
] |
2021-09-22
|
[
[
"Deng",
"Jieren",
""
],
[
"Wang",
"Yijue",
""
],
[
"Li",
"Ji",
""
],
[
"Shang",
"Chao",
""
],
[
"Liu",
"Hang",
""
],
[
"Rajasekaran",
"Sanguthevar",
""
],
[
"Ding",
"Caiwen",
""
]
] |
Although federated learning has increasingly gained attention in terms of effectively utilizing local devices for data privacy enhancement, recent studies show that publicly shared gradients in the training process can reveal the private training images (gradient leakage) to a third-party in computer vision. We have, however, no systematic understanding of the gradient leakage mechanism on the Transformer based language models. In this paper, as the first attempt, we formulate the gradient attack problem on the Transformer-based language models and propose a gradient attack algorithm, TAG, to reconstruct the local training data. We develop a set of metrics to evaluate the effectiveness of the proposed attack algorithm quantitatively. Experimental results on Transformer, TinyBERT$_{4}$, TinyBERT$_{6}$, BERT$_{BASE}$, and BERT$_{LARGE}$ using GLUE benchmark show that TAG works well on more weight distributions in reconstructing training data and achieves 1.5$\times$ recover rate and 2.5$\times$ ROUGE-2 over prior methods without the need of ground truth label. TAG can obtain up to 90$\%$ data by attacking gradients in CoLA dataset. In addition, TAG has a stronger adversary on large models, small dictionary size, and small input length. We hope the proposed TAG will shed some light on the privacy leakage problem in Transformer-based NLP models.
|
2307.05694
|
Dr. Mohammed Javed
|
Anurag Dhote and Mohammed Javed and David S Doermann
|
A Survey on Figure Classification Techniques in Scientific Documents
|
Some contents of this paper appears in the accepted paper - "A Survey
and Approach to Chart Classification" at 15th IAPR GREC 2023 at 17th ICDAR
2023, August 21-26, San Jose, USA. arXiv admin note: text overlap with
arXiv:2307.04147
| null | null | null |
cs.IR cs.CV cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
Figures visually represent an essential piece of information and provide an
effective means to communicate scientific facts. Recently there have been many
efforts toward extracting data directly from figures, specifically from tables,
diagrams, and plots, using different Artificial Intelligence and Machine
Learning techniques. This is because removing information from figures could
lead to deeper insights into the concepts highlighted in the scientific
documents. In this survey paper, we systematically categorize figures into five
classes - tables, photos, diagrams, maps, and plots, and subsequently present a
critical review of the existing methodologies and data sets that address the
problem of figure classification. Finally, we identify the current research
gaps and provide possible directions for further research on figure
classification.
|
[
{
"created": "Sun, 9 Jul 2023 10:55:11 GMT",
"version": "v1"
}
] |
2023-07-13
|
[
[
"Dhote",
"Anurag",
""
],
[
"Javed",
"Mohammed",
""
],
[
"Doermann",
"David S",
""
]
] |
Figures visually represent an essential piece of information and provide an effective means to communicate scientific facts. Recently there have been many efforts toward extracting data directly from figures, specifically from tables, diagrams, and plots, using different Artificial Intelligence and Machine Learning techniques. This is because removing information from figures could lead to deeper insights into the concepts highlighted in the scientific documents. In this survey paper, we systematically categorize figures into five classes - tables, photos, diagrams, maps, and plots, and subsequently present a critical review of the existing methodologies and data sets that address the problem of figure classification. Finally, we identify the current research gaps and provide possible directions for further research on figure classification.
|
2306.00148
|
Wei Xiao
|
Wei Xiao and Tsun-Hsuan Wang and Chuang Gan and Daniela Rus
|
SafeDiffuser: Safe Planning with Diffusion Probabilistic Models
|
19 pages, website: https://safediffuser.github.io/safediffuser/
| null | null | null |
cs.LG cs.RO cs.SY eess.SY
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Diffusion model-based approaches have shown promise in data-driven planning,
but there are no safety guarantees, thus making it hard to be applied for
safety-critical applications. To address these challenges, we propose a new
method, called SafeDiffuser, to ensure diffusion probabilistic models satisfy
specifications by using a class of control barrier functions. The key idea of
our approach is to embed the proposed finite-time diffusion invariance into the
denoising diffusion procedure, which enables trustworthy diffusion data
generation. Moreover, we demonstrate that our finite-time diffusion invariance
method through generative models not only maintains generalization performance
but also creates robustness in safe data generation. We test our method on a
series of safe planning tasks, including maze path generation, legged robot
locomotion, and 3D space manipulation, with results showing the advantages of
robustness and guarantees over vanilla diffusion models.
|
[
{
"created": "Wed, 31 May 2023 19:38:12 GMT",
"version": "v1"
}
] |
2023-06-02
|
[
[
"Xiao",
"Wei",
""
],
[
"Wang",
"Tsun-Hsuan",
""
],
[
"Gan",
"Chuang",
""
],
[
"Rus",
"Daniela",
""
]
] |
Diffusion model-based approaches have shown promise in data-driven planning, but there are no safety guarantees, thus making it hard to be applied for safety-critical applications. To address these challenges, we propose a new method, called SafeDiffuser, to ensure diffusion probabilistic models satisfy specifications by using a class of control barrier functions. The key idea of our approach is to embed the proposed finite-time diffusion invariance into the denoising diffusion procedure, which enables trustworthy diffusion data generation. Moreover, we demonstrate that our finite-time diffusion invariance method through generative models not only maintains generalization performance but also creates robustness in safe data generation. We test our method on a series of safe planning tasks, including maze path generation, legged robot locomotion, and 3D space manipulation, with results showing the advantages of robustness and guarantees over vanilla diffusion models.
|
2206.00229
|
Wisdom Agboh
|
Wisdom C. Agboh, Jeffrey Ichnowski, Ken Goldberg, Mehmet R. Dogar
|
Multi-Object Grasping in the Plane
|
Accepted to the International Symposium on Robotics Research (ISRR),
2022
| null | null | null |
cs.RO cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
We consider a novel problem where multiple rigid convex polygonal objects
rest in randomly placed positions and orientations on a planar surface visible
from an overhead camera. The objective is to efficiently grasp and transport
all objects into a bin using multi-object push-grasps, where multiple objects
are pushed together to facilitate multi-object grasping. We provide necessary
conditions for frictionless multi-object push-grasps and apply these to filter
inadmissible grasps in a novel multi-object grasp planner. We find that our
planner is 19 times faster than a Mujoco simulator baseline. We also propose a
picking algorithm that uses both single- and multi-object grasps to pick
objects. In physical grasping experiments comparing performance with a
single-object picking baseline, we find that the frictionless multi-object
grasping system achieves 13.6\% higher grasp success and is 59.9\% faster, from
212 PPH to 340 PPH. See
\url{https://sites.google.com/view/multi-object-grasping} for videos and code.
|
[
{
"created": "Wed, 1 Jun 2022 04:40:45 GMT",
"version": "v1"
},
{
"created": "Wed, 21 Sep 2022 16:51:42 GMT",
"version": "v2"
}
] |
2022-09-22
|
[
[
"Agboh",
"Wisdom C.",
""
],
[
"Ichnowski",
"Jeffrey",
""
],
[
"Goldberg",
"Ken",
""
],
[
"Dogar",
"Mehmet R.",
""
]
] |
We consider a novel problem where multiple rigid convex polygonal objects rest in randomly placed positions and orientations on a planar surface visible from an overhead camera. The objective is to efficiently grasp and transport all objects into a bin using multi-object push-grasps, where multiple objects are pushed together to facilitate multi-object grasping. We provide necessary conditions for frictionless multi-object push-grasps and apply these to filter inadmissible grasps in a novel multi-object grasp planner. We find that our planner is 19 times faster than a Mujoco simulator baseline. We also propose a picking algorithm that uses both single- and multi-object grasps to pick objects. In physical grasping experiments comparing performance with a single-object picking baseline, we find that the frictionless multi-object grasping system achieves 13.6\% higher grasp success and is 59.9\% faster, from 212 PPH to 340 PPH. See \url{https://sites.google.com/view/multi-object-grasping} for videos and code.
|
2106.03373
|
Yiding Liu Dr.
|
Yiding Liu, Guan Huang, Jiaxiang Liu, Weixue Lu, Suqi Cheng, Yukun Li,
Daiting Shi, Shuaiqiang Wang, Zhicong Cheng, Dawei Yin
|
Pre-trained Language Model for Web-scale Retrieval in Baidu Search
|
Accepted by KDD 2021
| null | null | null |
cs.IR
|
http://creativecommons.org/licenses/by/4.0/
|
Retrieval is a crucial stage in web search that identifies a small set of
query-relevant candidates from a billion-scale corpus. Discovering more
semantically-related candidates in the retrieval stage is very promising to
expose more high-quality results to the end users. However, it still remains
non-trivial challenges of building and deploying effective retrieval models for
semantic matching in real search engine. In this paper, we describe the
retrieval system that we developed and deployed in Baidu Search. The system
exploits the recent state-of-the-art Chinese pretrained language model, namely
Enhanced Representation through kNowledge IntEgration (ERNIE), which
facilitates the system with expressive semantic matching. In particular, we
developed an ERNIE-based retrieval model, which is equipped with 1) expressive
Transformer-based semantic encoders, and 2) a comprehensive multi-stage
training paradigm. More importantly, we present a practical system workflow for
deploying the model in web-scale retrieval. Eventually, the system is fully
deployed into production, where rigorous offline and online experiments were
conducted. The results show that the system can perform high-quality candidate
retrieval, especially for those tail queries with uncommon demands. Overall,
the new retrieval system facilitated by pretrained language model (i.e., ERNIE)
can largely improve the usability and applicability of our search engine.
|
[
{
"created": "Mon, 7 Jun 2021 06:55:45 GMT",
"version": "v1"
},
{
"created": "Fri, 25 Jun 2021 13:32:13 GMT",
"version": "v2"
},
{
"created": "Wed, 30 Jun 2021 05:38:58 GMT",
"version": "v3"
},
{
"created": "Sat, 16 Oct 2021 15:12:57 GMT",
"version": "v4"
}
] |
2021-10-19
|
[
[
"Liu",
"Yiding",
""
],
[
"Huang",
"Guan",
""
],
[
"Liu",
"Jiaxiang",
""
],
[
"Lu",
"Weixue",
""
],
[
"Cheng",
"Suqi",
""
],
[
"Li",
"Yukun",
""
],
[
"Shi",
"Daiting",
""
],
[
"Wang",
"Shuaiqiang",
""
],
[
"Cheng",
"Zhicong",
""
],
[
"Yin",
"Dawei",
""
]
] |
Retrieval is a crucial stage in web search that identifies a small set of query-relevant candidates from a billion-scale corpus. Discovering more semantically-related candidates in the retrieval stage is very promising to expose more high-quality results to the end users. However, it still remains non-trivial challenges of building and deploying effective retrieval models for semantic matching in real search engine. In this paper, we describe the retrieval system that we developed and deployed in Baidu Search. The system exploits the recent state-of-the-art Chinese pretrained language model, namely Enhanced Representation through kNowledge IntEgration (ERNIE), which facilitates the system with expressive semantic matching. In particular, we developed an ERNIE-based retrieval model, which is equipped with 1) expressive Transformer-based semantic encoders, and 2) a comprehensive multi-stage training paradigm. More importantly, we present a practical system workflow for deploying the model in web-scale retrieval. Eventually, the system is fully deployed into production, where rigorous offline and online experiments were conducted. The results show that the system can perform high-quality candidate retrieval, especially for those tail queries with uncommon demands. Overall, the new retrieval system facilitated by pretrained language model (i.e., ERNIE) can largely improve the usability and applicability of our search engine.
|
2310.17569
|
Xinghui Li Mr.
|
Xinghui Li, Jingyi Lu, Kai Han, Victor Prisacariu
|
SD4Match: Learning to Prompt Stable Diffusion Model for Semantic
Matching
|
Accepted to CVPR 2024. Project website:
https://sd4match.active.vision/
| null | null | null |
cs.CV cs.LG
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
In this paper, we address the challenge of matching semantically similar
keypoints across image pairs. Existing research indicates that the intermediate
output of the UNet within the Stable Diffusion (SD) can serve as robust image
feature maps for such a matching task. We demonstrate that by employing a basic
prompt tuning technique, the inherent potential of Stable Diffusion can be
harnessed, resulting in a significant enhancement in accuracy over previous
approaches. We further introduce a novel conditional prompting module that
conditions the prompt on the local details of the input image pairs, leading to
a further improvement in performance. We designate our approach as SD4Match,
short for Stable Diffusion for Semantic Matching. Comprehensive evaluations of
SD4Match on the PF-Pascal, PF-Willow, and SPair-71k datasets show that it sets
new benchmarks in accuracy across all these datasets. Particularly, SD4Match
outperforms the previous state-of-the-art by a margin of 12 percentage points
on the challenging SPair-71k dataset.
|
[
{
"created": "Thu, 26 Oct 2023 16:58:01 GMT",
"version": "v1"
},
{
"created": "Tue, 26 Mar 2024 11:52:23 GMT",
"version": "v2"
}
] |
2024-03-27
|
[
[
"Li",
"Xinghui",
""
],
[
"Lu",
"Jingyi",
""
],
[
"Han",
"Kai",
""
],
[
"Prisacariu",
"Victor",
""
]
] |
In this paper, we address the challenge of matching semantically similar keypoints across image pairs. Existing research indicates that the intermediate output of the UNet within the Stable Diffusion (SD) can serve as robust image feature maps for such a matching task. We demonstrate that by employing a basic prompt tuning technique, the inherent potential of Stable Diffusion can be harnessed, resulting in a significant enhancement in accuracy over previous approaches. We further introduce a novel conditional prompting module that conditions the prompt on the local details of the input image pairs, leading to a further improvement in performance. We designate our approach as SD4Match, short for Stable Diffusion for Semantic Matching. Comprehensive evaluations of SD4Match on the PF-Pascal, PF-Willow, and SPair-71k datasets show that it sets new benchmarks in accuracy across all these datasets. Particularly, SD4Match outperforms the previous state-of-the-art by a margin of 12 percentage points on the challenging SPair-71k dataset.
|
2103.12328
|
Kazuma Kobayashi
|
Kazuma Kobayashi, Ryuichiro Hataya, Yusuke Kurose, Mototaka Miyake,
Masamichi Takahashi, Akiko Nakagawa, Tatsuya Harada, Ryuji Hamamoto
|
Decomposing Normal and Abnormal Features of Medical Images into Discrete
Latent Codes for Content-Based Image Retrieval
| null | null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In medical imaging, the characteristics purely derived from a disease should
reflect the extent to which abnormal findings deviate from the normal features.
Indeed, physicians often need corresponding images without abnormal findings of
interest or, conversely, images that contain similar abnormal findings
regardless of normal anatomical context. This is called comparative diagnostic
reading of medical images, which is essential for a correct diagnosis. To
support comparative diagnostic reading, content-based image retrieval (CBIR),
which can selectively utilize normal and abnormal features in medical images as
two separable semantic components, will be useful. Therefore, we propose a
neural network architecture to decompose the semantic components of medical
images into two latent codes: normal anatomy code and abnormal anatomy code.
The normal anatomy code represents normal anatomies that should have existed if
the sample is healthy, whereas the abnormal anatomy code attributes to abnormal
changes that reflect deviation from the normal baseline. These latent codes are
discretized through vector quantization to enable binary hashing, which can
reduce the computational burden at the time of similarity search. By
calculating the similarity based on either normal or abnormal anatomy codes or
the combination of the two codes, our algorithm can retrieve images according
to the selected semantic component from a dataset consisting of brain magnetic
resonance images of gliomas. Our CBIR system qualitatively and quantitatively
achieves remarkable results.
|
[
{
"created": "Tue, 23 Mar 2021 05:53:53 GMT",
"version": "v1"
}
] |
2021-03-24
|
[
[
"Kobayashi",
"Kazuma",
""
],
[
"Hataya",
"Ryuichiro",
""
],
[
"Kurose",
"Yusuke",
""
],
[
"Miyake",
"Mototaka",
""
],
[
"Takahashi",
"Masamichi",
""
],
[
"Nakagawa",
"Akiko",
""
],
[
"Harada",
"Tatsuya",
""
],
[
"Hamamoto",
"Ryuji",
""
]
] |
In medical imaging, the characteristics purely derived from a disease should reflect the extent to which abnormal findings deviate from the normal features. Indeed, physicians often need corresponding images without abnormal findings of interest or, conversely, images that contain similar abnormal findings regardless of normal anatomical context. This is called comparative diagnostic reading of medical images, which is essential for a correct diagnosis. To support comparative diagnostic reading, content-based image retrieval (CBIR), which can selectively utilize normal and abnormal features in medical images as two separable semantic components, will be useful. Therefore, we propose a neural network architecture to decompose the semantic components of medical images into two latent codes: normal anatomy code and abnormal anatomy code. The normal anatomy code represents normal anatomies that should have existed if the sample is healthy, whereas the abnormal anatomy code attributes to abnormal changes that reflect deviation from the normal baseline. These latent codes are discretized through vector quantization to enable binary hashing, which can reduce the computational burden at the time of similarity search. By calculating the similarity based on either normal or abnormal anatomy codes or the combination of the two codes, our algorithm can retrieve images according to the selected semantic component from a dataset consisting of brain magnetic resonance images of gliomas. Our CBIR system qualitatively and quantitatively achieves remarkable results.
|
2102.09202
|
Emir Demirel
|
Emir Demirel, Sven Ahlb\"ack, Simon Dixon
|
Low Resource Audio-to-Lyrics Alignment From Polyphonic Music Recordings
| null | null | null | null |
cs.SD eess.AS
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Lyrics alignment in long music recordings can be memory exhaustive when
performed in a single pass. In this study, we present a novel method that
performs audio-to-lyrics alignment with a low memory consumption footprint
regardless of the duration of the music recording. The proposed system first
spots the anchoring words within the audio signal. With respect to these
anchors, the recording is then segmented and a second-pass alignment is
performed to obtain the word timings. We show that our audio-to-lyrics
alignment system performs competitively with the state-of-the-art, while
requiring much less computational resources. In addition, we utilise our lyrics
alignment system to segment the music recordings into sentence-level chunks.
Notably on the segmented recordings, we report the lyrics transcription scores
on a number of benchmark test sets. Finally, our experiments highlight the
importance of the source separation step for good performance on the
transcription and alignment tasks. For reproducibility, we publicly share our
code with the research community.
|
[
{
"created": "Thu, 18 Feb 2021 07:54:56 GMT",
"version": "v1"
}
] |
2021-02-19
|
[
[
"Demirel",
"Emir",
""
],
[
"Ahlbäck",
"Sven",
""
],
[
"Dixon",
"Simon",
""
]
] |
Lyrics alignment in long music recordings can be memory exhaustive when performed in a single pass. In this study, we present a novel method that performs audio-to-lyrics alignment with a low memory consumption footprint regardless of the duration of the music recording. The proposed system first spots the anchoring words within the audio signal. With respect to these anchors, the recording is then segmented and a second-pass alignment is performed to obtain the word timings. We show that our audio-to-lyrics alignment system performs competitively with the state-of-the-art, while requiring much less computational resources. In addition, we utilise our lyrics alignment system to segment the music recordings into sentence-level chunks. Notably on the segmented recordings, we report the lyrics transcription scores on a number of benchmark test sets. Finally, our experiments highlight the importance of the source separation step for good performance on the transcription and alignment tasks. For reproducibility, we publicly share our code with the research community.
|
1401.2483
|
Andino Maseleno
|
Andino Maseleno and Md. Mahmud Hasan
|
Dempster-Shafer Theory for Move Prediction in Start Kicking of The
Bicycle Kick of Sepak Takraw Game
|
Middle-East Journal of Scientific Research, Vol. 16, No. 7, 2013.
ISSN 1990-9233, pp. 896 - 903
| null | null | null |
cs.AI
|
http://creativecommons.org/licenses/by/3.0/
|
This paper presents Dempster-Shafer theory for move prediction in start
kicking of the bicycle kick of sepak takraw game. Sepak takraw is a highly
complex net-barrier kicking sport that involves dazzling displays of quick
reflexes, acrobatic twists, turns and swerves of the agile human body movement.
A Bicycle kick or Scissor kick is a physical move made by throwing the body up
into the air, making a shearing movement with the legs to get one leg in front
of the other without holding on to the ground. Specifically, this paper
considers bicycle kick of sepak takraw game in start kicking of the ball with
uncertainty where player has different awareness regarding the contingencies.
We have chosen Dempster-Shafer theory because the advantages of the
Dempster-Shafer theory which include the ability to model information in a
flexible way without requiring a probability to be assigned to each element in
a set, providing a convenient and simple mechanism for combining two or more
pieces of evidence under certain conditions, it can model ignorance explicitly,
rejection of the law of additivity for belief in disjoint propositions.
|
[
{
"created": "Fri, 10 Jan 2014 23:48:40 GMT",
"version": "v1"
}
] |
2014-01-14
|
[
[
"Maseleno",
"Andino",
""
],
[
"Hasan",
"Md. Mahmud",
""
]
] |
This paper presents Dempster-Shafer theory for move prediction in start kicking of the bicycle kick of sepak takraw game. Sepak takraw is a highly complex net-barrier kicking sport that involves dazzling displays of quick reflexes, acrobatic twists, turns and swerves of the agile human body movement. A Bicycle kick or Scissor kick is a physical move made by throwing the body up into the air, making a shearing movement with the legs to get one leg in front of the other without holding on to the ground. Specifically, this paper considers bicycle kick of sepak takraw game in start kicking of the ball with uncertainty where player has different awareness regarding the contingencies. We have chosen Dempster-Shafer theory because the advantages of the Dempster-Shafer theory which include the ability to model information in a flexible way without requiring a probability to be assigned to each element in a set, providing a convenient and simple mechanism for combining two or more pieces of evidence under certain conditions, it can model ignorance explicitly, rejection of the law of additivity for belief in disjoint propositions.
|
1906.08462
|
Chongyi Li
|
Chongyi Li, Runmin Cong, Junhui Hou, Sanyi Zhang, Yue Qian, Sam Kwong
|
Nested Network with Two-Stream Pyramid for Salient Object Detection in
Optical Remote Sensing Images
|
11 pages, 8 figures, has been accepted by TGRS
| null |
10.1109/TGRS.2019.2925070
| null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Arising from the various object types and scales, diverse imaging
orientations, and cluttered backgrounds in optical remote sensing image (RSI),
it is difficult to directly extend the success of salient object detection for
nature scene image to the optical RSI. In this paper, we propose an end-to-end
deep network called LV-Net based on the shape of network architecture, which
detects salient objects from optical RSIs in a purely data-driven fashion. The
proposed LV-Net consists of two key modules, i.e., a two-stream pyramid module
(L-shaped module) and an encoder-decoder module with nested connections
(V-shaped module). Specifically, the L-shaped module extracts a set of
complementary information hierarchically by using a two-stream pyramid
structure, which is beneficial to perceiving the diverse scales and local
details of salient objects. The V-shaped module gradually integrates encoder
detail features with decoder semantic features through nested connections,
which aims at suppressing the cluttered backgrounds and highlighting the
salient objects. In addition, we construct the first publicly available optical
RSI dataset for salient object detection, including 800 images with varying
spatial resolutions, diverse saliency types, and pixel-wise ground truth.
Experiments on this benchmark dataset demonstrate that the proposed method
outperforms the state-of-the-art salient object detection methods both
qualitatively and quantitatively.
|
[
{
"created": "Thu, 20 Jun 2019 06:57:13 GMT",
"version": "v1"
}
] |
2020-01-08
|
[
[
"Li",
"Chongyi",
""
],
[
"Cong",
"Runmin",
""
],
[
"Hou",
"Junhui",
""
],
[
"Zhang",
"Sanyi",
""
],
[
"Qian",
"Yue",
""
],
[
"Kwong",
"Sam",
""
]
] |
Arising from the various object types and scales, diverse imaging orientations, and cluttered backgrounds in optical remote sensing image (RSI), it is difficult to directly extend the success of salient object detection for nature scene image to the optical RSI. In this paper, we propose an end-to-end deep network called LV-Net based on the shape of network architecture, which detects salient objects from optical RSIs in a purely data-driven fashion. The proposed LV-Net consists of two key modules, i.e., a two-stream pyramid module (L-shaped module) and an encoder-decoder module with nested connections (V-shaped module). Specifically, the L-shaped module extracts a set of complementary information hierarchically by using a two-stream pyramid structure, which is beneficial to perceiving the diverse scales and local details of salient objects. The V-shaped module gradually integrates encoder detail features with decoder semantic features through nested connections, which aims at suppressing the cluttered backgrounds and highlighting the salient objects. In addition, we construct the first publicly available optical RSI dataset for salient object detection, including 800 images with varying spatial resolutions, diverse saliency types, and pixel-wise ground truth. Experiments on this benchmark dataset demonstrate that the proposed method outperforms the state-of-the-art salient object detection methods both qualitatively and quantitatively.
|
1902.08646
|
Fabio Kepler
|
F\'abio Kepler, Jonay Tr\'enous, Marcos Treviso, Miguel Vera, Andr\'e
F. T. Martins
|
OpenKiwi: An Open Source Framework for Quality Estimation
|
Published at the Annual Meeting of the Association for Computational
Linguistics (ACL) 2019: System Demonstrations
(https://aclweb.org/anthology/papers/P/P19/P19-3020/)
| null | null | null |
cs.CL cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We introduce OpenKiwi, a PyTorch-based open source framework for translation
quality estimation. OpenKiwi supports training and testing of word-level and
sentence-level quality estimation systems, implementing the winning systems of
the WMT 2015-18 quality estimation campaigns. We benchmark OpenKiwi on two
datasets from WMT 2018 (English-German SMT and NMT), yielding state-of-the-art
performance on the word-level tasks and near state-of-the-art in the
sentence-level tasks.
|
[
{
"created": "Fri, 22 Feb 2019 19:27:45 GMT",
"version": "v1"
},
{
"created": "Mon, 26 Aug 2019 15:07:52 GMT",
"version": "v2"
}
] |
2019-08-27
|
[
[
"Kepler",
"Fábio",
""
],
[
"Trénous",
"Jonay",
""
],
[
"Treviso",
"Marcos",
""
],
[
"Vera",
"Miguel",
""
],
[
"Martins",
"André F. T.",
""
]
] |
We introduce OpenKiwi, a PyTorch-based open source framework for translation quality estimation. OpenKiwi supports training and testing of word-level and sentence-level quality estimation systems, implementing the winning systems of the WMT 2015-18 quality estimation campaigns. We benchmark OpenKiwi on two datasets from WMT 2018 (English-German SMT and NMT), yielding state-of-the-art performance on the word-level tasks and near state-of-the-art in the sentence-level tasks.
|
2104.08542
|
Huifeng Guo
|
Huifeng Guo, Wei Guo, Yong Gao, Ruiming Tang, Xiuqiang He, Wenzhi Liu
|
ScaleFreeCTR: MixCache-based Distributed Training System for CTR Models
with Huge Embedding Table
|
10 pages
| null | null | null |
cs.IR cs.AI cs.DC cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Because of the superior feature representation ability of deep learning,
various deep Click-Through Rate (CTR) models are deployed in the commercial
systems by industrial companies. To achieve better performance, it is necessary
to train the deep CTR models on huge volume of training data efficiently, which
makes speeding up the training process an essential problem. Different from the
models with dense training data, the training data for CTR models is usually
high-dimensional and sparse. To transform the high-dimensional sparse input
into low-dimensional dense real-value vectors, almost all deep CTR models adopt
the embedding layer, which easily reaches hundreds of GB or even TB. Since a
single GPU cannot afford to accommodate all the embedding parameters, when
performing distributed training, it is not reasonable to conduct the
data-parallelism only. Therefore, existing distributed training platforms for
recommendation adopt model-parallelism. Specifically, they use CPU (Host)
memory of servers to maintain and update the embedding parameters and utilize
GPU worker to conduct forward and backward computations. Unfortunately, these
platforms suffer from two bottlenecks: (1) the latency of pull \& push
operations between Host and GPU; (2) parameters update and synchronization in
the CPU servers. To address such bottlenecks, in this paper, we propose the
ScaleFreeCTR: a MixCache-based distributed training system for CTR models.
Specifically, in SFCTR, we also store huge embedding table in CPU but utilize
GPU instead of CPU to conduct embedding synchronization efficiently. To reduce
the latency of data transfer between both GPU-Host and GPU-GPU, the MixCache
mechanism and Virtual Sparse Id operation are proposed. Comprehensive
experiments and ablation studies are conducted to demonstrate the effectiveness
and efficiency of SFCTR.
|
[
{
"created": "Sat, 17 Apr 2021 13:36:19 GMT",
"version": "v1"
},
{
"created": "Tue, 11 May 2021 14:11:46 GMT",
"version": "v2"
}
] |
2021-05-12
|
[
[
"Guo",
"Huifeng",
""
],
[
"Guo",
"Wei",
""
],
[
"Gao",
"Yong",
""
],
[
"Tang",
"Ruiming",
""
],
[
"He",
"Xiuqiang",
""
],
[
"Liu",
"Wenzhi",
""
]
] |
Because of the superior feature representation ability of deep learning, various deep Click-Through Rate (CTR) models are deployed in the commercial systems by industrial companies. To achieve better performance, it is necessary to train the deep CTR models on huge volume of training data efficiently, which makes speeding up the training process an essential problem. Different from the models with dense training data, the training data for CTR models is usually high-dimensional and sparse. To transform the high-dimensional sparse input into low-dimensional dense real-value vectors, almost all deep CTR models adopt the embedding layer, which easily reaches hundreds of GB or even TB. Since a single GPU cannot afford to accommodate all the embedding parameters, when performing distributed training, it is not reasonable to conduct the data-parallelism only. Therefore, existing distributed training platforms for recommendation adopt model-parallelism. Specifically, they use CPU (Host) memory of servers to maintain and update the embedding parameters and utilize GPU worker to conduct forward and backward computations. Unfortunately, these platforms suffer from two bottlenecks: (1) the latency of pull \& push operations between Host and GPU; (2) parameters update and synchronization in the CPU servers. To address such bottlenecks, in this paper, we propose the ScaleFreeCTR: a MixCache-based distributed training system for CTR models. Specifically, in SFCTR, we also store huge embedding table in CPU but utilize GPU instead of CPU to conduct embedding synchronization efficiently. To reduce the latency of data transfer between both GPU-Host and GPU-GPU, the MixCache mechanism and Virtual Sparse Id operation are proposed. Comprehensive experiments and ablation studies are conducted to demonstrate the effectiveness and efficiency of SFCTR.
|
2401.15595
|
Manas Mhasakar
|
Manas Mhasakar, Shikhar Sharma, Apurv Mehra, Utkarsh Venaik, Ujjwal
Singhal, Dhruv Kumar, Kashish Mittal
|
Comuniqa : Exploring Large Language Models for improving speaking skills
|
Accepted at 7th ACM SIGCAS/SIGCHI Conference of Computing and
Sustainable Societies : ACM COMPASS 2024
| null | null | null |
cs.HC cs.CY
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this paper, we investigate the potential of Large Language Models (LLMs)
to improve English speaking skills. This is particularly relevant in countries
like India, where English is crucial for academic, professional, and personal
communication but remains a non-native language for many. Traditional methods
for enhancing speaking skills often rely on human experts, which can be limited
in terms of scalability, accessibility, and affordability. Recent advancements
in Artificial Intelligence (AI) offer promising solutions to overcome these
limitations.
We propose Comuniqa, a novel LLM-based system designed to enhance English
speaking skills. We adopt a human-centric evaluation approach, comparing
Comuniqa with the feedback and instructions provided by human experts. In our
evaluation, we divide the participants in three groups: those who use LLM-based
system for improving speaking skills, those guided by human experts for the
same task and those who utilize both the LLM-based system as well as the human
experts. Using surveys, interviews, and actual study sessions, we provide a
detailed perspective on the effectiveness of different learning modalities. Our
preliminary findings suggest that while LLM-based systems have commendable
accuracy, they lack human-level cognitive capabilities, both in terms of
accuracy and empathy. Nevertheless, Comuniqa represents a significant step
towards achieving Sustainable Development Goal 4: Quality Education by
providing a valuable learning tool for individuals who may not have access to
human experts for improving their speaking skills.
|
[
{
"created": "Sun, 28 Jan 2024 07:37:33 GMT",
"version": "v1"
},
{
"created": "Wed, 3 Apr 2024 14:33:10 GMT",
"version": "v2"
},
{
"created": "Tue, 14 May 2024 04:34:20 GMT",
"version": "v3"
}
] |
2024-05-15
|
[
[
"Mhasakar",
"Manas",
""
],
[
"Sharma",
"Shikhar",
""
],
[
"Mehra",
"Apurv",
""
],
[
"Venaik",
"Utkarsh",
""
],
[
"Singhal",
"Ujjwal",
""
],
[
"Kumar",
"Dhruv",
""
],
[
"Mittal",
"Kashish",
""
]
] |
In this paper, we investigate the potential of Large Language Models (LLMs) to improve English speaking skills. This is particularly relevant in countries like India, where English is crucial for academic, professional, and personal communication but remains a non-native language for many. Traditional methods for enhancing speaking skills often rely on human experts, which can be limited in terms of scalability, accessibility, and affordability. Recent advancements in Artificial Intelligence (AI) offer promising solutions to overcome these limitations. We propose Comuniqa, a novel LLM-based system designed to enhance English speaking skills. We adopt a human-centric evaluation approach, comparing Comuniqa with the feedback and instructions provided by human experts. In our evaluation, we divide the participants in three groups: those who use LLM-based system for improving speaking skills, those guided by human experts for the same task and those who utilize both the LLM-based system as well as the human experts. Using surveys, interviews, and actual study sessions, we provide a detailed perspective on the effectiveness of different learning modalities. Our preliminary findings suggest that while LLM-based systems have commendable accuracy, they lack human-level cognitive capabilities, both in terms of accuracy and empathy. Nevertheless, Comuniqa represents a significant step towards achieving Sustainable Development Goal 4: Quality Education by providing a valuable learning tool for individuals who may not have access to human experts for improving their speaking skills.
|
2110.02514
|
Woojoo Kim
|
Woojoo Kim, Shuping Xiong
|
ViewfinderVR: Configurable Viewfinder for Selection of Distant Objects
in VR
| null | null |
10.1007/s10055-022-00649-z
| null |
cs.HC
|
http://creativecommons.org/licenses/by/4.0/
|
Selection is one of the fundamental user interactions in virtual reality (VR)
and 3D user interaction, and raycasting has been one of the most popular object
selection techniques in VR. However, the selection of small or distant objects
through raycasting has been known to be difficult. To overcome this limitation,
this study proposed a new technique called ViewfinderVR for improved selection
of distant objects in VR, utilizing a virtual viewfinder panel with a modern
adaptation of the through-the-lens metaphor. ViewfinderVR enables faster and
more accurate target selection by allowing customization of the interaction
space projected onto a virtual panel within reach, and users can select objects
reflected on the panel with either ray-based or touch interaction. Experimental
results of Fitts' law-based tests with 20 participants showed that ViewfinderVR
outperformed traditional raycasting in terms of task performance (movement
time, error rate, and throughput) and perceived workload (NASA-TLX ratings),
where touch interaction was superior to ray-based interaction. The associated
user behavior was also recorded and analyzed to understand the underlying
reasons for the improved task performance and reduced workload. The proposed
technique can be used in VR applications to enhance the selection of distant
objects.
|
[
{
"created": "Wed, 6 Oct 2021 05:35:04 GMT",
"version": "v1"
}
] |
2022-06-07
|
[
[
"Kim",
"Woojoo",
""
],
[
"Xiong",
"Shuping",
""
]
] |
Selection is one of the fundamental user interactions in virtual reality (VR) and 3D user interaction, and raycasting has been one of the most popular object selection techniques in VR. However, the selection of small or distant objects through raycasting has been known to be difficult. To overcome this limitation, this study proposed a new technique called ViewfinderVR for improved selection of distant objects in VR, utilizing a virtual viewfinder panel with a modern adaptation of the through-the-lens metaphor. ViewfinderVR enables faster and more accurate target selection by allowing customization of the interaction space projected onto a virtual panel within reach, and users can select objects reflected on the panel with either ray-based or touch interaction. Experimental results of Fitts' law-based tests with 20 participants showed that ViewfinderVR outperformed traditional raycasting in terms of task performance (movement time, error rate, and throughput) and perceived workload (NASA-TLX ratings), where touch interaction was superior to ray-based interaction. The associated user behavior was also recorded and analyzed to understand the underlying reasons for the improved task performance and reduced workload. The proposed technique can be used in VR applications to enhance the selection of distant objects.
|
2107.06511
|
Dingcheng Yang
|
Dingcheng Yang, Wenjian Yu, Yuanbo Guo, Wenjie Liang
|
CNN-Cap: Effective Convolutional Neural Network Based Capacitance Models
for Full-Chip Parasitic Extraction
|
9 pages, 13 figures. Accepted at 2021 International Conference On
Computer Aided Design (ICCAD)
| null | null | null |
cs.LG cs.AR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Accurate capacitance extraction is becoming more important for designing
integrated circuits under advanced process technology. The pattern matching
based full-chip extraction methodology delivers fast computational speed, but
suffers from large error, and tedious efforts on building capacitance models of
the increasing structure patterns. In this work, we propose an effective method
for building convolutional neural network (CNN) based capacitance models
(called CNN-Cap) for two-dimensional (2-D) structures in full-chip capacitance
extraction. With a novel grid-based data representation, the proposed method is
able to model the pattern with a variable number of conductors, so that largely
reduce the number of patterns. Based on the ability of ResNet architecture on
capturing spatial information and the proposed training skills, the obtained
CNN-Cap exhibits much better performance over the multilayer perception neural
network based capacitance model while being more versatile. Extensive
experiments on a 55nm and a 15nm process technologies have demonstrated that
the error of total capacitance produced with CNN-Cap is always within 1.3% and
the error of produced coupling capacitance is less than 10% in over 99.5%
probability. CNN-Cap runs more than 4000X faster than 2-D field solver on a GPU
server, while it consumes negligible memory compared to the look-up table based
capacitance model.
|
[
{
"created": "Wed, 14 Jul 2021 07:14:35 GMT",
"version": "v1"
}
] |
2021-07-15
|
[
[
"Yang",
"Dingcheng",
""
],
[
"Yu",
"Wenjian",
""
],
[
"Guo",
"Yuanbo",
""
],
[
"Liang",
"Wenjie",
""
]
] |
Accurate capacitance extraction is becoming more important for designing integrated circuits under advanced process technology. The pattern matching based full-chip extraction methodology delivers fast computational speed, but suffers from large error, and tedious efforts on building capacitance models of the increasing structure patterns. In this work, we propose an effective method for building convolutional neural network (CNN) based capacitance models (called CNN-Cap) for two-dimensional (2-D) structures in full-chip capacitance extraction. With a novel grid-based data representation, the proposed method is able to model the pattern with a variable number of conductors, so that largely reduce the number of patterns. Based on the ability of ResNet architecture on capturing spatial information and the proposed training skills, the obtained CNN-Cap exhibits much better performance over the multilayer perception neural network based capacitance model while being more versatile. Extensive experiments on a 55nm and a 15nm process technologies have demonstrated that the error of total capacitance produced with CNN-Cap is always within 1.3% and the error of produced coupling capacitance is less than 10% in over 99.5% probability. CNN-Cap runs more than 4000X faster than 2-D field solver on a GPU server, while it consumes negligible memory compared to the look-up table based capacitance model.
|
2311.00434
|
Shintaro Shiba
|
Shintaro Shiba, Friedhelm Hamann, Yoshimitsu Aoki, Guillermo Gallego
|
Event-based Background-Oriented Schlieren
|
Accepted at IEEE T-PAMI
|
IEEE Transactions on Pattern Analysis and Machine Intelligence,
Oct. 2023
|
10.1109/TPAMI.2023.3328188
| null |
cs.CV eess.IV
|
http://creativecommons.org/licenses/by/4.0/
|
Schlieren imaging is an optical technique to observe the flow of transparent
media, such as air or water, without any particle seeding. However,
conventional frame-based techniques require both high spatial and temporal
resolution cameras, which impose bright illumination and expensive computation
limitations. Event cameras offer potential advantages (high dynamic range, high
temporal resolution, and data efficiency) to overcome such limitations due to
their bio-inspired sensing principle. This paper presents a novel technique for
perceiving air convection using events and frames by providing the first
theoretical analysis that connects event data and schlieren. We formulate the
problem as a variational optimization one combining the linearized event
generation model with a physically-motivated parameterization that estimates
the temporal derivative of the air density. The experiments with accurately
aligned frame- and event camera data reveal that the proposed method enables
event cameras to obtain on par results with existing frame-based optical flow
techniques. Moreover, the proposed method works under dark conditions where
frame-based schlieren fails, and also enables slow-motion analysis by
leveraging the event camera's advantages. Our work pioneers and opens a new
stack of event camera applications, as we publish the source code as well as
the first schlieren dataset with high-quality frame and event data.
https://github.com/tub-rip/event_based_bos
|
[
{
"created": "Wed, 1 Nov 2023 10:57:20 GMT",
"version": "v1"
}
] |
2024-03-05
|
[
[
"Shiba",
"Shintaro",
""
],
[
"Hamann",
"Friedhelm",
""
],
[
"Aoki",
"Yoshimitsu",
""
],
[
"Gallego",
"Guillermo",
""
]
] |
Schlieren imaging is an optical technique to observe the flow of transparent media, such as air or water, without any particle seeding. However, conventional frame-based techniques require both high spatial and temporal resolution cameras, which impose bright illumination and expensive computation limitations. Event cameras offer potential advantages (high dynamic range, high temporal resolution, and data efficiency) to overcome such limitations due to their bio-inspired sensing principle. This paper presents a novel technique for perceiving air convection using events and frames by providing the first theoretical analysis that connects event data and schlieren. We formulate the problem as a variational optimization one combining the linearized event generation model with a physically-motivated parameterization that estimates the temporal derivative of the air density. The experiments with accurately aligned frame- and event camera data reveal that the proposed method enables event cameras to obtain on par results with existing frame-based optical flow techniques. Moreover, the proposed method works under dark conditions where frame-based schlieren fails, and also enables slow-motion analysis by leveraging the event camera's advantages. Our work pioneers and opens a new stack of event camera applications, as we publish the source code as well as the first schlieren dataset with high-quality frame and event data. https://github.com/tub-rip/event_based_bos
|
1906.01622
|
Mozhi Zhang
|
Mozhi Zhang, Keyulu Xu, Ken-ichi Kawarabayashi, Stefanie Jegelka,
Jordan Boyd-Graber
|
Are Girls Neko or Sh\=ojo? Cross-Lingual Alignment of Non-Isomorphic
Embeddings with Iterative Normalization
|
ACL 2019
| null | null | null |
cs.CL cs.AI cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Cross-lingual word embeddings (CLWE) underlie many multilingual natural
language processing systems, often through orthogonal transformations of
pre-trained monolingual embeddings. However, orthogonal mapping only works on
language pairs whose embeddings are naturally isomorphic. For non-isomorphic
pairs, our method (Iterative Normalization) transforms monolingual embeddings
to make orthogonal alignment easier by simultaneously enforcing that (1)
individual word vectors are unit length, and (2) each language's average vector
is zero. Iterative Normalization consistently improves word translation
accuracy of three CLWE methods, with the largest improvement observed on
English-Japanese (from 2% to 44% test accuracy).
|
[
{
"created": "Tue, 4 Jun 2019 17:56:22 GMT",
"version": "v1"
},
{
"created": "Wed, 5 Jun 2019 01:34:19 GMT",
"version": "v2"
},
{
"created": "Mon, 11 Nov 2019 07:36:47 GMT",
"version": "v3"
}
] |
2019-11-12
|
[
[
"Zhang",
"Mozhi",
""
],
[
"Xu",
"Keyulu",
""
],
[
"Kawarabayashi",
"Ken-ichi",
""
],
[
"Jegelka",
"Stefanie",
""
],
[
"Boyd-Graber",
"Jordan",
""
]
] |
Cross-lingual word embeddings (CLWE) underlie many multilingual natural language processing systems, often through orthogonal transformations of pre-trained monolingual embeddings. However, orthogonal mapping only works on language pairs whose embeddings are naturally isomorphic. For non-isomorphic pairs, our method (Iterative Normalization) transforms monolingual embeddings to make orthogonal alignment easier by simultaneously enforcing that (1) individual word vectors are unit length, and (2) each language's average vector is zero. Iterative Normalization consistently improves word translation accuracy of three CLWE methods, with the largest improvement observed on English-Japanese (from 2% to 44% test accuracy).
|
2305.04539
|
Masato Uchida
|
Kota Kawamoto and Masato Uchida
|
Q&A Label Learning
|
46 pages, 5 figures
| null | null | null |
cs.LG stat.ML
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Assigning labels to instances is crucial for supervised machine learning. In
this paper, we proposed a novel annotation method called Q&A labeling, which
involves a question generator that asks questions about the labels of the
instances to be assigned, and an annotator who answers the questions and
assigns the corresponding labels to the instances. We derived a generative
model of labels assigned according to two different Q&A labeling procedures
that differ in the way questions are asked and answered. We showed that, in
both procedures, the derived model is partially consistent with that assumed in
previous studies. The main distinction of this study from previous studies lies
in the fact that the label generative model was not assumed, but rather derived
based on the definition of a specific annotation method, Q&A labeling. We also
derived a loss function to evaluate the classification risk of ordinary
supervised machine learning using instances assigned Q&A labels and evaluated
the upper bound of the classification error. The results indicate statistical
consistency in learning with Q&A labels.
|
[
{
"created": "Mon, 8 May 2023 08:22:18 GMT",
"version": "v1"
}
] |
2023-05-09
|
[
[
"Kawamoto",
"Kota",
""
],
[
"Uchida",
"Masato",
""
]
] |
Assigning labels to instances is crucial for supervised machine learning. In this paper, we proposed a novel annotation method called Q&A labeling, which involves a question generator that asks questions about the labels of the instances to be assigned, and an annotator who answers the questions and assigns the corresponding labels to the instances. We derived a generative model of labels assigned according to two different Q&A labeling procedures that differ in the way questions are asked and answered. We showed that, in both procedures, the derived model is partially consistent with that assumed in previous studies. The main distinction of this study from previous studies lies in the fact that the label generative model was not assumed, but rather derived based on the definition of a specific annotation method, Q&A labeling. We also derived a loss function to evaluate the classification risk of ordinary supervised machine learning using instances assigned Q&A labels and evaluated the upper bound of the classification error. The results indicate statistical consistency in learning with Q&A labels.
|
1904.09823
|
Yingchao Feng
|
Yingchao Feng, Wenhui Diao, Zhonghan Chang, Menglong Yan, Xian Sun,
Xin Gao
|
Ship Instance Segmentation From Remote Sensing Images Using Sequence
Local Context Module
|
4 pages, 5 figures, IEEE Geoscience and Remote Sensing Society 2019
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The performance of object instance segmentation in remote sensing images has
been greatly improved through the introduction of many landmark frameworks
based on convolutional neural network. However, the object densely issue still
affects the accuracy of such segmentation frameworks. Objects of the same class
are easily confused, which is most likely due to the close docking between
objects. We think context information is critical to address this issue. So, we
propose a novel framework called SLCMASK-Net, in which a sequence local context
module (SLC) is introduced to avoid confusion between objects of the same
class. The SLC module applies a sequence of dilation convolution blocks to
progressively learn multi-scale context information in the mask branch.
Besides, we try to add SLC module to different locations in our framework and
experiment with the effect of different parameter settings. Comparative
experiments are conducted on remote sensing images acquired by QuickBird with a
resolution of $0.5m-1m$ and the results show that the proposed method achieves
state-of-the-art performance.
|
[
{
"created": "Mon, 22 Apr 2019 12:33:06 GMT",
"version": "v1"
}
] |
2019-04-23
|
[
[
"Feng",
"Yingchao",
""
],
[
"Diao",
"Wenhui",
""
],
[
"Chang",
"Zhonghan",
""
],
[
"Yan",
"Menglong",
""
],
[
"Sun",
"Xian",
""
],
[
"Gao",
"Xin",
""
]
] |
The performance of object instance segmentation in remote sensing images has been greatly improved through the introduction of many landmark frameworks based on convolutional neural network. However, the object densely issue still affects the accuracy of such segmentation frameworks. Objects of the same class are easily confused, which is most likely due to the close docking between objects. We think context information is critical to address this issue. So, we propose a novel framework called SLCMASK-Net, in which a sequence local context module (SLC) is introduced to avoid confusion between objects of the same class. The SLC module applies a sequence of dilation convolution blocks to progressively learn multi-scale context information in the mask branch. Besides, we try to add SLC module to different locations in our framework and experiment with the effect of different parameter settings. Comparative experiments are conducted on remote sensing images acquired by QuickBird with a resolution of $0.5m-1m$ and the results show that the proposed method achieves state-of-the-art performance.
|
2105.02905
|
Roberto Metere
|
Roberto Metere, Myriam Neaimeh, Charles Morisset, Carsten Maple,
Xavier Bellekens, Ricardo M. Czekster
|
Securing the Electric Vehicle Charging Infrastructure
|
42 pages, white paper
| null | null | null |
cs.CR cs.NI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Electric Vehicles (EVs) can help alleviate our reliance on fossil fuels for
transport and electricity systems. However, charging millions of EV batteries
requires management to prevent overloading the electricity grid and minimise
costly upgrades that are ultimately paid for by consumers.
Managed chargers, such as Vehicle-to-Grid (V2G) chargers, allow control over
the time, speed and direction of charging. Such control assists in balancing
electricity supply and demand across a green electricity system and could
reduce costs for consumers.
Smart and V2G chargers connect EVs to the power grid using a charging device
which includes a data connection to exchange information and control commands
between various entities in the EV ecosystem. This introduces data privacy
concerns and is a potential target for cyber-security attacks. Therefore, the
implementation of a secure system is crucial to permit both consumers and
electricity system operators to trust smart charging and V2G.
In principle, we already have the technology needed for a connected EV
charging infrastructure to be securely enabled, borrowing best practices from
the Internet and industrial control systems. We must properly adapt the
security technology to take into account the challenges peculiar to the EV
charging infrastructure. Challenges go beyond technical considerations and
other issues arise such as balancing trade-offs between security and other
desirable qualities such as interoperability, scalability, crypto-agility,
affordability and energy efficiency.
This document reviews security and privacy topics relevant to the EV charging
ecosystem with a focus on smart charging and V2G.
|
[
{
"created": "Thu, 6 May 2021 18:10:42 GMT",
"version": "v1"
},
{
"created": "Mon, 4 Apr 2022 16:03:07 GMT",
"version": "v2"
},
{
"created": "Wed, 6 Jul 2022 09:54:31 GMT",
"version": "v3"
}
] |
2022-07-07
|
[
[
"Metere",
"Roberto",
""
],
[
"Neaimeh",
"Myriam",
""
],
[
"Morisset",
"Charles",
""
],
[
"Maple",
"Carsten",
""
],
[
"Bellekens",
"Xavier",
""
],
[
"Czekster",
"Ricardo M.",
""
]
] |
Electric Vehicles (EVs) can help alleviate our reliance on fossil fuels for transport and electricity systems. However, charging millions of EV batteries requires management to prevent overloading the electricity grid and minimise costly upgrades that are ultimately paid for by consumers. Managed chargers, such as Vehicle-to-Grid (V2G) chargers, allow control over the time, speed and direction of charging. Such control assists in balancing electricity supply and demand across a green electricity system and could reduce costs for consumers. Smart and V2G chargers connect EVs to the power grid using a charging device which includes a data connection to exchange information and control commands between various entities in the EV ecosystem. This introduces data privacy concerns and is a potential target for cyber-security attacks. Therefore, the implementation of a secure system is crucial to permit both consumers and electricity system operators to trust smart charging and V2G. In principle, we already have the technology needed for a connected EV charging infrastructure to be securely enabled, borrowing best practices from the Internet and industrial control systems. We must properly adapt the security technology to take into account the challenges peculiar to the EV charging infrastructure. Challenges go beyond technical considerations and other issues arise such as balancing trade-offs between security and other desirable qualities such as interoperability, scalability, crypto-agility, affordability and energy efficiency. This document reviews security and privacy topics relevant to the EV charging ecosystem with a focus on smart charging and V2G.
|
2209.13514
|
Hang Zhou
|
Zhiliang Xu, Hang Zhou, Zhibin Hong, Ziwei Liu, Jiaming Liu, Zhizhi
Guo, Junyu Han, Jingtuo Liu, Errui Ding, Jingdong Wang
|
StyleSwap: Style-Based Generator Empowers Robust Face Swapping
|
Accepted to ECCV 2022. Demo videos and code can be found at
https://hangz-nju-cuhk.github.io/projects/StyleSwap
| null | null | null |
cs.CV cs.GR
|
http://creativecommons.org/licenses/by/4.0/
|
Numerous attempts have been made to the task of person-agnostic face swapping
given its wide applications. While existing methods mostly rely on tedious
network and loss designs, they still struggle in the information balancing
between the source and target faces, and tend to produce visible artifacts. In
this work, we introduce a concise and effective framework named StyleSwap. Our
core idea is to leverage a style-based generator to empower high-fidelity and
robust face swapping, thus the generator's advantage can be adopted for
optimizing identity similarity. We identify that with only minimal
modifications, a StyleGAN2 architecture can successfully handle the desired
information from both source and target. Additionally, inspired by the ToRGB
layers, a Swapping-Driven Mask Branch is further devised to improve information
blending. Furthermore, the advantage of StyleGAN inversion can be adopted.
Particularly, a Swapping-Guided ID Inversion strategy is proposed to optimize
identity similarity. Extensive experiments validate that our framework
generates high-quality face swapping results that outperform state-of-the-art
methods both qualitatively and quantitatively.
|
[
{
"created": "Tue, 27 Sep 2022 16:35:16 GMT",
"version": "v1"
}
] |
2022-09-28
|
[
[
"Xu",
"Zhiliang",
""
],
[
"Zhou",
"Hang",
""
],
[
"Hong",
"Zhibin",
""
],
[
"Liu",
"Ziwei",
""
],
[
"Liu",
"Jiaming",
""
],
[
"Guo",
"Zhizhi",
""
],
[
"Han",
"Junyu",
""
],
[
"Liu",
"Jingtuo",
""
],
[
"Ding",
"Errui",
""
],
[
"Wang",
"Jingdong",
""
]
] |
Numerous attempts have been made to the task of person-agnostic face swapping given its wide applications. While existing methods mostly rely on tedious network and loss designs, they still struggle in the information balancing between the source and target faces, and tend to produce visible artifacts. In this work, we introduce a concise and effective framework named StyleSwap. Our core idea is to leverage a style-based generator to empower high-fidelity and robust face swapping, thus the generator's advantage can be adopted for optimizing identity similarity. We identify that with only minimal modifications, a StyleGAN2 architecture can successfully handle the desired information from both source and target. Additionally, inspired by the ToRGB layers, a Swapping-Driven Mask Branch is further devised to improve information blending. Furthermore, the advantage of StyleGAN inversion can be adopted. Particularly, a Swapping-Guided ID Inversion strategy is proposed to optimize identity similarity. Extensive experiments validate that our framework generates high-quality face swapping results that outperform state-of-the-art methods both qualitatively and quantitatively.
|
2311.12312
|
Evgenia (Eugenia) Ternovska
|
Eugenia Ternovska
|
Promise Algebra: An Algebraic Model of Non-Deterministic Computations
|
34 pages, 6 figures
| null | null | null |
cs.LO cs.DB
|
http://creativecommons.org/licenses/by/4.0/
|
Our goal is to define an algebraic language for reasoning about
non-deterministic computations. Towards this goal, we introduce an algebra of
string-to-string transductions. Specifically, it is an algebra of partial
functions on words over the alphabet of relational $\tau$-structures over the
same domain. The algebra has a two-level syntax, and thus, two parameters to
control its expressive power. The top level defines algebraic expressions, and
the bottom level specifies atomic transitions. History-dependent Choice
functions resolve atomic non-determinism, and make general relations
functional. Equivalence classes of such functions serve as certificates for
computational problems specified by algebraic terms. The algebra has an
equivalent syntax in the form of a Dynamic Logic, where terms describing
computational processes or programs appear inside the modalities. We define a
simple secondary logic for representing atomic transitions, which is a
modification of conjunctive queries. With this logic, the algebra can represent
both reachability and counting examples, which is not possible in Datalog. We
analyze the data complexity of this logic, measured in the size of the input
structure, and show that a restricted fragment of the logic captures the
complexity class NP. The logic can be viewed as a database query language,
where atomic propagations are separated from control.
|
[
{
"created": "Tue, 21 Nov 2023 03:10:43 GMT",
"version": "v1"
}
] |
2023-11-22
|
[
[
"Ternovska",
"Eugenia",
""
]
] |
Our goal is to define an algebraic language for reasoning about non-deterministic computations. Towards this goal, we introduce an algebra of string-to-string transductions. Specifically, it is an algebra of partial functions on words over the alphabet of relational $\tau$-structures over the same domain. The algebra has a two-level syntax, and thus, two parameters to control its expressive power. The top level defines algebraic expressions, and the bottom level specifies atomic transitions. History-dependent Choice functions resolve atomic non-determinism, and make general relations functional. Equivalence classes of such functions serve as certificates for computational problems specified by algebraic terms. The algebra has an equivalent syntax in the form of a Dynamic Logic, where terms describing computational processes or programs appear inside the modalities. We define a simple secondary logic for representing atomic transitions, which is a modification of conjunctive queries. With this logic, the algebra can represent both reachability and counting examples, which is not possible in Datalog. We analyze the data complexity of this logic, measured in the size of the input structure, and show that a restricted fragment of the logic captures the complexity class NP. The logic can be viewed as a database query language, where atomic propagations are separated from control.
|
2103.12452
|
Rianne De Heide
|
Rianne de Heide and James Cheshire and Pierre M\'enard and Alexandra
Carpentier
|
Bandits with many optimal arms
|
Substantial rewrite and added experiments. Accepted for NeurIPS 2021
| null | null | null |
cs.LG stat.ML
|
http://creativecommons.org/licenses/by/4.0/
|
We consider a stochastic bandit problem with a possibly infinite number of
arms. We write $p^*$ for the proportion of optimal arms and $\Delta$ for the
minimal mean-gap between optimal and sub-optimal arms. We characterize the
optimal learning rates both in the cumulative regret setting, and in the
best-arm identification setting in terms of the problem parameters $T$ (the
budget), $p^*$ and $\Delta$. For the objective of minimizing the cumulative
regret, we provide a lower bound of order $\Omega(\log(T)/(p^*\Delta))$ and a
UCB-style algorithm with matching upper bound up to a factor of
$\log(1/\Delta)$. Our algorithm needs $p^*$ to calibrate its parameters, and we
prove that this knowledge is necessary, since adapting to $p^*$ in this setting
is impossible. For best-arm identification we also provide a lower bound of
order $\Omega(\exp(-cT\Delta^2 p^*))$ on the probability of outputting a
sub-optimal arm where $c>0$ is an absolute constant. We also provide an
elimination algorithm with an upper bound matching the lower bound up to a
factor of order $\log(T)$ in the exponential, and that does not need $p^*$ or
$\Delta$ as parameter. Our results apply directly to the three related problems
of competing against the $j$-th best arm, identifying an $\epsilon$ good arm,
and finding an arm with mean larger than a quantile of a known order.
|
[
{
"created": "Tue, 23 Mar 2021 11:02:31 GMT",
"version": "v1"
},
{
"created": "Fri, 5 Nov 2021 08:25:11 GMT",
"version": "v2"
}
] |
2021-11-08
|
[
[
"de Heide",
"Rianne",
""
],
[
"Cheshire",
"James",
""
],
[
"Ménard",
"Pierre",
""
],
[
"Carpentier",
"Alexandra",
""
]
] |
We consider a stochastic bandit problem with a possibly infinite number of arms. We write $p^*$ for the proportion of optimal arms and $\Delta$ for the minimal mean-gap between optimal and sub-optimal arms. We characterize the optimal learning rates both in the cumulative regret setting, and in the best-arm identification setting in terms of the problem parameters $T$ (the budget), $p^*$ and $\Delta$. For the objective of minimizing the cumulative regret, we provide a lower bound of order $\Omega(\log(T)/(p^*\Delta))$ and a UCB-style algorithm with matching upper bound up to a factor of $\log(1/\Delta)$. Our algorithm needs $p^*$ to calibrate its parameters, and we prove that this knowledge is necessary, since adapting to $p^*$ in this setting is impossible. For best-arm identification we also provide a lower bound of order $\Omega(\exp(-cT\Delta^2 p^*))$ on the probability of outputting a sub-optimal arm where $c>0$ is an absolute constant. We also provide an elimination algorithm with an upper bound matching the lower bound up to a factor of order $\log(T)$ in the exponential, and that does not need $p^*$ or $\Delta$ as parameter. Our results apply directly to the three related problems of competing against the $j$-th best arm, identifying an $\epsilon$ good arm, and finding an arm with mean larger than a quantile of a known order.
|
1810.03993
|
Margaret Mitchell
|
Margaret Mitchell and Simone Wu and Andrew Zaldivar and Parker Barnes
and Lucy Vasserman and Ben Hutchinson and Elena Spitzer and Inioluwa Deborah
Raji and Timnit Gebru
|
Model Cards for Model Reporting
| null |
FAT* '19: Conference on Fairness, Accountability, and
Transparency, January 29--31, 2019, Atlanta, GA, USA
|
10.1145/3287560.3287596
| null |
cs.LG cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Trained machine learning models are increasingly used to perform high-impact
tasks in areas such as law enforcement, medicine, education, and employment. In
order to clarify the intended use cases of machine learning models and minimize
their usage in contexts for which they are not well suited, we recommend that
released models be accompanied by documentation detailing their performance
characteristics. In this paper, we propose a framework that we call model
cards, to encourage such transparent model reporting. Model cards are short
documents accompanying trained machine learning models that provide benchmarked
evaluation in a variety of conditions, such as across different cultural,
demographic, or phenotypic groups (e.g., race, geographic location, sex,
Fitzpatrick skin type) and intersectional groups (e.g., age and race, or sex
and Fitzpatrick skin type) that are relevant to the intended application
domains. Model cards also disclose the context in which models are intended to
be used, details of the performance evaluation procedures, and other relevant
information. While we focus primarily on human-centered machine learning models
in the application fields of computer vision and natural language processing,
this framework can be used to document any trained machine learning model. To
solidify the concept, we provide cards for two supervised models: One trained
to detect smiling faces in images, and one trained to detect toxic comments in
text. We propose model cards as a step towards the responsible democratization
of machine learning and related AI technology, increasing transparency into how
well AI technology works. We hope this work encourages those releasing trained
machine learning models to accompany model releases with similar detailed
evaluation numbers and other relevant documentation.
|
[
{
"created": "Fri, 5 Oct 2018 22:33:43 GMT",
"version": "v1"
},
{
"created": "Mon, 14 Jan 2019 20:25:27 GMT",
"version": "v2"
}
] |
2019-01-16
|
[
[
"Mitchell",
"Margaret",
""
],
[
"Wu",
"Simone",
""
],
[
"Zaldivar",
"Andrew",
""
],
[
"Barnes",
"Parker",
""
],
[
"Vasserman",
"Lucy",
""
],
[
"Hutchinson",
"Ben",
""
],
[
"Spitzer",
"Elena",
""
],
[
"Raji",
"Inioluwa Deborah",
""
],
[
"Gebru",
"Timnit",
""
]
] |
Trained machine learning models are increasingly used to perform high-impact tasks in areas such as law enforcement, medicine, education, and employment. In order to clarify the intended use cases of machine learning models and minimize their usage in contexts for which they are not well suited, we recommend that released models be accompanied by documentation detailing their performance characteristics. In this paper, we propose a framework that we call model cards, to encourage such transparent model reporting. Model cards are short documents accompanying trained machine learning models that provide benchmarked evaluation in a variety of conditions, such as across different cultural, demographic, or phenotypic groups (e.g., race, geographic location, sex, Fitzpatrick skin type) and intersectional groups (e.g., age and race, or sex and Fitzpatrick skin type) that are relevant to the intended application domains. Model cards also disclose the context in which models are intended to be used, details of the performance evaluation procedures, and other relevant information. While we focus primarily on human-centered machine learning models in the application fields of computer vision and natural language processing, this framework can be used to document any trained machine learning model. To solidify the concept, we provide cards for two supervised models: One trained to detect smiling faces in images, and one trained to detect toxic comments in text. We propose model cards as a step towards the responsible democratization of machine learning and related AI technology, increasing transparency into how well AI technology works. We hope this work encourages those releasing trained machine learning models to accompany model releases with similar detailed evaluation numbers and other relevant documentation.
|
1702.06943
|
Fengan Li
|
Fengan Li, Lingjiao Chen, Yijing Zeng, Arun Kumar, Jeffrey F.
Naughton, Jignesh M. Patel, Xi Wu
|
Tuple-oriented Compression for Large-scale Mini-batch Stochastic
Gradient Descent
|
Accepted to Sigmod 2019
| null |
10.1145/3299869.3300070
| null |
cs.LG cs.DB stat.ML
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Data compression is a popular technique for improving the efficiency of data
processing workloads such as SQL queries and more recently, machine learning
(ML) with classical batch gradient methods. But the efficacy of such ideas for
mini-batch stochastic gradient descent (MGD), arguably the workhorse algorithm
of modern ML, is an open question. MGD's unique data access pattern renders
prior art, including those designed for batch gradient methods, less effective.
We fill this crucial research gap by proposing a new lossless compression
scheme we call tuple-oriented compression (TOC) that is inspired by an unlikely
source, the string/text compression scheme Lempel-Ziv-Welch, but tailored to
MGD in a way that preserves tuple boundaries within mini-batches. We then
present a suite of novel compressed matrix operation execution techniques
tailored to the TOC compression scheme that operate directly over the
compressed data representation and avoid decompression overheads. An extensive
empirical evaluation with real-world datasets shows that TOC consistently
achieves substantial compression ratios by up to 51x and reduces runtimes for
MGD workloads by up to 10.2x in popular ML systems.
|
[
{
"created": "Wed, 22 Feb 2017 18:58:25 GMT",
"version": "v1"
},
{
"created": "Wed, 1 Mar 2017 05:43:41 GMT",
"version": "v2"
},
{
"created": "Sun, 20 Jan 2019 05:13:18 GMT",
"version": "v3"
}
] |
2019-01-23
|
[
[
"Li",
"Fengan",
""
],
[
"Chen",
"Lingjiao",
""
],
[
"Zeng",
"Yijing",
""
],
[
"Kumar",
"Arun",
""
],
[
"Naughton",
"Jeffrey F.",
""
],
[
"Patel",
"Jignesh M.",
""
],
[
"Wu",
"Xi",
""
]
] |
Data compression is a popular technique for improving the efficiency of data processing workloads such as SQL queries and more recently, machine learning (ML) with classical batch gradient methods. But the efficacy of such ideas for mini-batch stochastic gradient descent (MGD), arguably the workhorse algorithm of modern ML, is an open question. MGD's unique data access pattern renders prior art, including those designed for batch gradient methods, less effective. We fill this crucial research gap by proposing a new lossless compression scheme we call tuple-oriented compression (TOC) that is inspired by an unlikely source, the string/text compression scheme Lempel-Ziv-Welch, but tailored to MGD in a way that preserves tuple boundaries within mini-batches. We then present a suite of novel compressed matrix operation execution techniques tailored to the TOC compression scheme that operate directly over the compressed data representation and avoid decompression overheads. An extensive empirical evaluation with real-world datasets shows that TOC consistently achieves substantial compression ratios by up to 51x and reduces runtimes for MGD workloads by up to 10.2x in popular ML systems.
|
1009.3527
|
Darren Strash
|
Michael T. Goodrich and Darren Strash
|
Priority Range Trees
|
12 pages, 3 figures. To appear at 21st International Symposium on
Algorithms and Computation (ISAAC 2010)
| null | null | null |
cs.CG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We describe a data structure, called a priority range tree, which
accommodates fast orthogonal range reporting queries on prioritized points. Let
$S$ be a set of $n$ points in the plane, where each point $p$ in $S$ is
assigned a weight $w(p)$ that is polynomial in $n$, and define the rank of $p$
to be $r(p)=\lfloor \log w(p) \rfloor$. Then the priority range tree can be
used to report all points in a three- or four-sided query range $R$ with rank
at least $\lfloor \log w \rfloor$ in time $O(\log W/w + k)$, and report $k$
highest-rank points in $R$ in time $O(\log\log n + \log W/w' + k)$, where
$W=\sum_{p\in S}{w(p)}$, $w'$ is the smallest weight of any point reported, and
$k$ is the output size. All times assume the standard RAM model of computation.
If the query range of interest is three sided, then the priority range tree
occupies $O(n)$ space, otherwise $O(n\log n)$ space is used to answer
four-sided queries. These queries are motivated by the Weber--Fechner Law,
which states that humans perceive and interpret data on a logarithmic scale.
|
[
{
"created": "Sat, 18 Sep 2010 01:00:33 GMT",
"version": "v1"
}
] |
2010-09-21
|
[
[
"Goodrich",
"Michael T.",
""
],
[
"Strash",
"Darren",
""
]
] |
We describe a data structure, called a priority range tree, which accommodates fast orthogonal range reporting queries on prioritized points. Let $S$ be a set of $n$ points in the plane, where each point $p$ in $S$ is assigned a weight $w(p)$ that is polynomial in $n$, and define the rank of $p$ to be $r(p)=\lfloor \log w(p) \rfloor$. Then the priority range tree can be used to report all points in a three- or four-sided query range $R$ with rank at least $\lfloor \log w \rfloor$ in time $O(\log W/w + k)$, and report $k$ highest-rank points in $R$ in time $O(\log\log n + \log W/w' + k)$, where $W=\sum_{p\in S}{w(p)}$, $w'$ is the smallest weight of any point reported, and $k$ is the output size. All times assume the standard RAM model of computation. If the query range of interest is three sided, then the priority range tree occupies $O(n)$ space, otherwise $O(n\log n)$ space is used to answer four-sided queries. These queries are motivated by the Weber--Fechner Law, which states that humans perceive and interpret data on a logarithmic scale.
|
1911.12507
|
Thuong Nguyen Canh
|
Thuong, Nguyen Canh, Chien, Trinh Van
|
Error Resilient Deep Compressive Sensing
|
4 pages, 2 figures
| null | null | null |
cs.CV cs.LG eess.IV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Compressive sensing (CS) is an emerging sampling technology that enables
reconstructing signals from a subset of measurements and even corrupted
measurements. Deep learning-based compressive sensing (DCS) has improved CS
performance while maintaining a fast reconstruction but requires a training
network for each measurement rate. Also, concerning the transmission scheme of
measurement lost, DCS cannot recover the original signal. Thereby, it fails to
maintain the error-resilient property. In this work, we proposed a robust deep
reconstruction network to preserve the error-resilient property under the
assumption of random measurement lost. Measurement lost layer is proposed to
simulate the measurement lost in an end-to-end framework.
|
[
{
"created": "Thu, 28 Nov 2019 03:16:39 GMT",
"version": "v1"
}
] |
2019-12-02
|
[
[
"Thuong",
"",
""
],
[
"Canh",
"Nguyen",
""
],
[
"Chien",
"",
""
],
[
"Van",
"Trinh",
""
]
] |
Compressive sensing (CS) is an emerging sampling technology that enables reconstructing signals from a subset of measurements and even corrupted measurements. Deep learning-based compressive sensing (DCS) has improved CS performance while maintaining a fast reconstruction but requires a training network for each measurement rate. Also, concerning the transmission scheme of measurement lost, DCS cannot recover the original signal. Thereby, it fails to maintain the error-resilient property. In this work, we proposed a robust deep reconstruction network to preserve the error-resilient property under the assumption of random measurement lost. Measurement lost layer is proposed to simulate the measurement lost in an end-to-end framework.
|
2204.09617
|
Zheng Chen
|
Zheng Chen, Durgakant Pushp, Lantao Liu
|
CALI: Coarse-to-Fine ALIgnments Based Unsupervised Domain Adaptation of
Traversability Prediction for Deployable Autonomous Navigation
| null | null | null | null |
cs.RO
|
http://creativecommons.org/licenses/by/4.0/
|
Traversability prediction is a fundamental perception capability for
autonomous navigation. The diversity of data in different domains imposes
significant gaps to the prediction performance of the perception model. In this
work, we make efforts to reduce the gaps by proposing a novel coarse-to-fine
unsupervised domain adaptation (UDA) model - CALI. Our aim is to transfer the
perception model with high data efficiency, eliminate the prohibitively
expensive data labeling, and improve the generalization capability during the
adaptation from easy-to-obtain source domains to various challenging target
domains. We prove that a combination of a coarse alignment and a fine alignment
can be beneficial to each other and further design a first-coarse-then-fine
alignment process. This proposed work bridges theoretical analyses and
algorithm designs, leading to an efficient UDA model with easy and stable
training. We show the advantages of our proposed model over multiple baselines
in several challenging domain adaptation setups. To further validate the
effectiveness of our model, we then combine our perception model with a visual
planner to build a navigation system and show the high reliability of our model
in complex natural environments where no labeled data is available.
|
[
{
"created": "Wed, 20 Apr 2022 16:52:43 GMT",
"version": "v1"
}
] |
2022-04-21
|
[
[
"Chen",
"Zheng",
""
],
[
"Pushp",
"Durgakant",
""
],
[
"Liu",
"Lantao",
""
]
] |
Traversability prediction is a fundamental perception capability for autonomous navigation. The diversity of data in different domains imposes significant gaps to the prediction performance of the perception model. In this work, we make efforts to reduce the gaps by proposing a novel coarse-to-fine unsupervised domain adaptation (UDA) model - CALI. Our aim is to transfer the perception model with high data efficiency, eliminate the prohibitively expensive data labeling, and improve the generalization capability during the adaptation from easy-to-obtain source domains to various challenging target domains. We prove that a combination of a coarse alignment and a fine alignment can be beneficial to each other and further design a first-coarse-then-fine alignment process. This proposed work bridges theoretical analyses and algorithm designs, leading to an efficient UDA model with easy and stable training. We show the advantages of our proposed model over multiple baselines in several challenging domain adaptation setups. To further validate the effectiveness of our model, we then combine our perception model with a visual planner to build a navigation system and show the high reliability of our model in complex natural environments where no labeled data is available.
|
1211.5221
|
Reza Malekian Ph.D.
|
Reza Malekian and Abdul Hanan Abdullah
|
Traffic Engineering Based on Effective Envelope Algorithm on Novel
Resource Reservation Method over Mobile Internet Protocol Version 6
|
International Journal of Innovative Computing, Information and
Control, 2012
| null | null | null |
cs.NI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The first decade of the 21st century has seen tremendous improvements in
mobile internet and its technologies. The high traffic volume of services such
as video conference and other real-time traffic applications are imposing a
great challenge on networks. In the meantime, demand for the use of mobile
devices in computation and communication such as smart phones, personal digital
assistants, and mobile-enabled laptops has grown rapidly. These services have
driven the demand for increasing and guaranteing bandwidth requirements in the
network. A direction of this paper is in the case of resource reservation
protocol (RSVP) over mobile IPv6 networks. There are numbers of proposed
solutions for RSVP and quality of service provision over mobile IPv6 networks,
but most of them using advanced resource reservation. In this paper, we propose
a mathematical model to determine maximum end-to-end delay bound through
intermediate routers along the network. These bounds are sent back to the home
agent for further processing. Once the home agent receives maximum end-to-end
delay bounds, it calculates cumulative bound and compares this bound with the
desired application end-to-end delay bound to make final decision on resource
reservation. This approach improves network resource utilization.
|
[
{
"created": "Thu, 22 Nov 2012 07:38:59 GMT",
"version": "v1"
}
] |
2012-11-26
|
[
[
"Malekian",
"Reza",
""
],
[
"Abdullah",
"Abdul Hanan",
""
]
] |
The first decade of the 21st century has seen tremendous improvements in mobile internet and its technologies. The high traffic volume of services such as video conference and other real-time traffic applications are imposing a great challenge on networks. In the meantime, demand for the use of mobile devices in computation and communication such as smart phones, personal digital assistants, and mobile-enabled laptops has grown rapidly. These services have driven the demand for increasing and guaranteing bandwidth requirements in the network. A direction of this paper is in the case of resource reservation protocol (RSVP) over mobile IPv6 networks. There are numbers of proposed solutions for RSVP and quality of service provision over mobile IPv6 networks, but most of them using advanced resource reservation. In this paper, we propose a mathematical model to determine maximum end-to-end delay bound through intermediate routers along the network. These bounds are sent back to the home agent for further processing. Once the home agent receives maximum end-to-end delay bounds, it calculates cumulative bound and compares this bound with the desired application end-to-end delay bound to make final decision on resource reservation. This approach improves network resource utilization.
|
1311.1712
|
Shaoshi Yang Dr.
|
Shaoshi Yang, Li Wang, Tiejun Lv and Lajos Hanzo
|
Approximate Bayesian Probabilistic-Data-Association-Aided Iterative
Detection for MIMO Systems Using Arbitrary M-ary Modulation
|
13 pages, 14 figures, 1 table, published in IEEE Transactions on
Vehicular Technology, vol. 62, no. 3, pp. 1228-1240, March, 2013
|
IEEE Transactions on Vehicular Technology, vol. 62, no. 3, pp.
1228-1240, March, 2013
|
10.1109/TVT.2012.2227863
| null |
cs.IT math.IT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this paper, the issue of designing an iterative-detection-and-decoding
(IDD)-aided receiver, relying on the low-complexity probabilistic data
association (PDA) method, is addressed for turbo-coded
multiple-input-multiple-output (MIMO) systems using general M-ary modulations.
We demonstrate that the classic candidate-search-aided bit-based extrinsic
log-likelihood ratio (LLR) calculation method is not applicable to the family
of PDA-based detectors. Additionally, we reveal that, in contrast to the
interpretation in the existing literature, the output symbol probabilities of
existing PDA algorithms are not the true a posteriori probabilities (APPs) but,
rather, the normalized symbol likelihoods. Therefore, the classic relationship,
where the extrinsic LLRs are given by subtracting the a priori LLRs from the a
posteriori LLRs, does not hold for the existing PDA-based detectors. Motivated
by these revelations, we conceive a new approximate Bayesian-theorem-based
logarithmic-domain PDA (AB-Log-PDA) method and unveil the technique of
calculating bit-based extrinsic LLRs for the AB-Log-PDA, which facilitates the
employment of the AB-Log-PDA in a simplified IDD receiver structure.
Additionally, we demonstrate that we may dispense with inner iterations within
the AB-Log-PDA in the context of IDD receivers. Our complexity analysis and
numerical results recorded for Nakagami-m fading channels demonstrate that the
proposed AB-Log-PDA-based IDD scheme is capable of achieving a performance
comparable with that of the optimal maximum a posteriori (MAP)-detector-based
IDD receiver, while imposing significantly lower computational complexity in
the scenarios considered.
|
[
{
"created": "Thu, 7 Nov 2013 15:26:34 GMT",
"version": "v1"
}
] |
2013-11-08
|
[
[
"Yang",
"Shaoshi",
""
],
[
"Wang",
"Li",
""
],
[
"Lv",
"Tiejun",
""
],
[
"Hanzo",
"Lajos",
""
]
] |
In this paper, the issue of designing an iterative-detection-and-decoding (IDD)-aided receiver, relying on the low-complexity probabilistic data association (PDA) method, is addressed for turbo-coded multiple-input-multiple-output (MIMO) systems using general M-ary modulations. We demonstrate that the classic candidate-search-aided bit-based extrinsic log-likelihood ratio (LLR) calculation method is not applicable to the family of PDA-based detectors. Additionally, we reveal that, in contrast to the interpretation in the existing literature, the output symbol probabilities of existing PDA algorithms are not the true a posteriori probabilities (APPs) but, rather, the normalized symbol likelihoods. Therefore, the classic relationship, where the extrinsic LLRs are given by subtracting the a priori LLRs from the a posteriori LLRs, does not hold for the existing PDA-based detectors. Motivated by these revelations, we conceive a new approximate Bayesian-theorem-based logarithmic-domain PDA (AB-Log-PDA) method and unveil the technique of calculating bit-based extrinsic LLRs for the AB-Log-PDA, which facilitates the employment of the AB-Log-PDA in a simplified IDD receiver structure. Additionally, we demonstrate that we may dispense with inner iterations within the AB-Log-PDA in the context of IDD receivers. Our complexity analysis and numerical results recorded for Nakagami-m fading channels demonstrate that the proposed AB-Log-PDA-based IDD scheme is capable of achieving a performance comparable with that of the optimal maximum a posteriori (MAP)-detector-based IDD receiver, while imposing significantly lower computational complexity in the scenarios considered.
|
2006.00954
|
Gianni Franchi
|
Gianni Franchi, Andrei Bursuc, Emanuel Aldea, Severine Dubuisson,
Isabelle Bloch
|
One Versus all for deep Neural Network Incertitude (OVNNI)
quantification
| null | null | null | null |
cs.CV cs.LG stat.ML
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Deep neural networks (DNNs) are powerful learning models yet their results
are not always reliable. This is due to the fact that modern DNNs are usually
uncalibrated and we cannot characterize their epistemic uncertainty. In this
work, we propose a new technique to quantify the epistemic uncertainty of data
easily. This method consists in mixing the predictions of an ensemble of DNNs
trained to classify One class vs All the other classes (OVA) with predictions
from a standard DNN trained to perform All vs All (AVA) classification. On the
one hand, the adjustment provided by the AVA DNN to the score of the base
classifiers allows for a more fine-grained inter-class separation. On the other
hand, the two types of classifiers enforce mutually their detection of
out-of-distribution (OOD) samples, circumventing entirely the requirement of
using such samples during training. Our method achieves state of the art
performance in quantifying OOD data across multiple datasets and architectures
while requiring little hyper-parameter tuning.
|
[
{
"created": "Mon, 1 Jun 2020 14:06:12 GMT",
"version": "v1"
}
] |
2020-06-03
|
[
[
"Franchi",
"Gianni",
""
],
[
"Bursuc",
"Andrei",
""
],
[
"Aldea",
"Emanuel",
""
],
[
"Dubuisson",
"Severine",
""
],
[
"Bloch",
"Isabelle",
""
]
] |
Deep neural networks (DNNs) are powerful learning models yet their results are not always reliable. This is due to the fact that modern DNNs are usually uncalibrated and we cannot characterize their epistemic uncertainty. In this work, we propose a new technique to quantify the epistemic uncertainty of data easily. This method consists in mixing the predictions of an ensemble of DNNs trained to classify One class vs All the other classes (OVA) with predictions from a standard DNN trained to perform All vs All (AVA) classification. On the one hand, the adjustment provided by the AVA DNN to the score of the base classifiers allows for a more fine-grained inter-class separation. On the other hand, the two types of classifiers enforce mutually their detection of out-of-distribution (OOD) samples, circumventing entirely the requirement of using such samples during training. Our method achieves state of the art performance in quantifying OOD data across multiple datasets and architectures while requiring little hyper-parameter tuning.
|
1205.7044
|
Negin Golrezaei
|
Negin Golrezaei, Alexandros G. Dimakis, Andreas F. Molisch
|
Wireless Device-to-Device Communications with Distributed Caching
|
to appear in ISIT 2012
| null | null | null |
cs.IT cs.NI math.IT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We introduce a novel wireless device-to-device (D2D) collaboration
architecture that exploits distributed storage of popular content to enable
frequency reuse. We identify a fundamental conflict between collaboration
distance and interference and show how to optimize the transmission power to
maximize frequency reuse. Our analysis depends on the user content request
statistics which are modeled by a Zipf distribution. Our main result is a
closed form expression of the optimal collaboration distance as a function of
the content reuse distribution parameters. We show that if the Zipf exponent of
the content reuse distribution is greater than 1, it is possible to have a
number of D2D interference-free collaboration pairs that scales linearly in the
number of nodes. If the Zipf exponent is smaller than 1, we identify the best
possible scaling in the number of D2D collaborating links. Surprisingly, a very
simple distributed caching policy achieves the optimal scaling behavior and
therefore there is no need to centrally coordinate what each node is caching.
|
[
{
"created": "Thu, 31 May 2012 17:02:31 GMT",
"version": "v1"
}
] |
2012-06-01
|
[
[
"Golrezaei",
"Negin",
""
],
[
"Dimakis",
"Alexandros G.",
""
],
[
"Molisch",
"Andreas F.",
""
]
] |
We introduce a novel wireless device-to-device (D2D) collaboration architecture that exploits distributed storage of popular content to enable frequency reuse. We identify a fundamental conflict between collaboration distance and interference and show how to optimize the transmission power to maximize frequency reuse. Our analysis depends on the user content request statistics which are modeled by a Zipf distribution. Our main result is a closed form expression of the optimal collaboration distance as a function of the content reuse distribution parameters. We show that if the Zipf exponent of the content reuse distribution is greater than 1, it is possible to have a number of D2D interference-free collaboration pairs that scales linearly in the number of nodes. If the Zipf exponent is smaller than 1, we identify the best possible scaling in the number of D2D collaborating links. Surprisingly, a very simple distributed caching policy achieves the optimal scaling behavior and therefore there is no need to centrally coordinate what each node is caching.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.