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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2304.14268
|
Colin Cleveland Mr
|
Colin Cleveland, Chin-Yen Lee, Shen-Fu Tsai, Wei-Hsuan Yu, Hsuan-Wei
Lee
|
Graphlet and Orbit Computation on Heterogeneous Graphs
|
13 pages, 7 figures
| null | null | null |
cs.SI physics.data-an
|
http://creativecommons.org/licenses/by/4.0/
|
Many applications, ranging from natural to social sciences, rely on graphlet
analysis for the intuitive and meaningful characterization of networks
employing micro-level structures as building blocks. However, it has not been
thoroughly explored in heterogeneous graphs, which comprise various types of
nodes and edges. Finding graphlets and orbits for heterogeneous graphs is
difficult because of the heterogeneity and abundance of semantic information.
We consider heterogeneous graphs, which can be treated as colored graphs. By
applying the canonical label technique, we determine the graph isomorphism
problem with multiple states on nodes and edges. With minimal parameters, we
build all non-isomorphic graphs and associated orbits. We provide a Python
package that can be used to generate orbits for colored directed graphs and
determine the frequency of orbit occurrence. Finally, we provide four examples
to illustrate the use of the Python package.
|
[
{
"created": "Wed, 26 Apr 2023 13:16:22 GMT",
"version": "v1"
},
{
"created": "Fri, 28 Apr 2023 20:16:51 GMT",
"version": "v2"
},
{
"created": "Mon, 5 Jun 2023 13:52:26 GMT",
"version": "v3"
}
] |
2023-06-06
|
[
[
"Cleveland",
"Colin",
""
],
[
"Lee",
"Chin-Yen",
""
],
[
"Tsai",
"Shen-Fu",
""
],
[
"Yu",
"Wei-Hsuan",
""
],
[
"Lee",
"Hsuan-Wei",
""
]
] |
Many applications, ranging from natural to social sciences, rely on graphlet analysis for the intuitive and meaningful characterization of networks employing micro-level structures as building blocks. However, it has not been thoroughly explored in heterogeneous graphs, which comprise various types of nodes and edges. Finding graphlets and orbits for heterogeneous graphs is difficult because of the heterogeneity and abundance of semantic information. We consider heterogeneous graphs, which can be treated as colored graphs. By applying the canonical label technique, we determine the graph isomorphism problem with multiple states on nodes and edges. With minimal parameters, we build all non-isomorphic graphs and associated orbits. We provide a Python package that can be used to generate orbits for colored directed graphs and determine the frequency of orbit occurrence. Finally, we provide four examples to illustrate the use of the Python package.
|
2208.10265
|
Soumyabrata Dev
|
Jiantao Wu, Fabrizio Orlandi, Tarek AlSkaif, Declan O'Sullivan, and
Soumyabrata Dev
|
A semantic web approach to uplift decentralized household energy data
|
Published in Sustainable Energy, Grids and Networks (SEGAN) 2022
| null | null | null |
cs.AI cs.IR cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In a decentralized household energy system comprised of various devices such
as home appliances, electric vehicles, and solar panels, end-users are able to
dig deeper into the system's details and further achieve energy sustainability
if they are presented with data on the electric energy consumption and
production at the granularity of the device. However, many databases in this
field are siloed from other domains, including solely information pertaining to
energy. This may result in the loss of information (e.g. weather) on each
device's energy use. Meanwhile, a large number of these datasets have been
extensively used in computational modeling techniques such as machine learning
models. While such computational approaches achieve great accuracy and
performance by concentrating only on a local view of datasets, model
reliability cannot be guaranteed since such models are very vulnerable to data
input fluctuations when information omission is taken into account. This
article tackles the data isolation issue in the field of smart energy systems
by examining Semantic Web methods on top of a household energy system. We offer
an ontology-based approach for managing decentralized data at the device-level
resolution in a system. As a consequence, the scope of the data associated with
each device may easily be expanded in an interoperable manner throughout the
Web, and additional information, such as weather, can be obtained from the Web,
provided that the data is organized according to W3C standards.
|
[
{
"created": "Thu, 18 Aug 2022 17:21:18 GMT",
"version": "v1"
},
{
"created": "Fri, 26 Aug 2022 22:48:54 GMT",
"version": "v2"
}
] |
2022-08-30
|
[
[
"Wu",
"Jiantao",
""
],
[
"Orlandi",
"Fabrizio",
""
],
[
"AlSkaif",
"Tarek",
""
],
[
"O'Sullivan",
"Declan",
""
],
[
"Dev",
"Soumyabrata",
""
]
] |
In a decentralized household energy system comprised of various devices such as home appliances, electric vehicles, and solar panels, end-users are able to dig deeper into the system's details and further achieve energy sustainability if they are presented with data on the electric energy consumption and production at the granularity of the device. However, many databases in this field are siloed from other domains, including solely information pertaining to energy. This may result in the loss of information (e.g. weather) on each device's energy use. Meanwhile, a large number of these datasets have been extensively used in computational modeling techniques such as machine learning models. While such computational approaches achieve great accuracy and performance by concentrating only on a local view of datasets, model reliability cannot be guaranteed since such models are very vulnerable to data input fluctuations when information omission is taken into account. This article tackles the data isolation issue in the field of smart energy systems by examining Semantic Web methods on top of a household energy system. We offer an ontology-based approach for managing decentralized data at the device-level resolution in a system. As a consequence, the scope of the data associated with each device may easily be expanded in an interoperable manner throughout the Web, and additional information, such as weather, can be obtained from the Web, provided that the data is organized according to W3C standards.
|
2006.09736
|
Casper Hansen
|
Christian Hansen and Casper Hansen and Jakob Grue Simonsen and Birger
Larsen and Stephen Alstrup and Christina Lioma
|
Factuality Checking in News Headlines with Eye Tracking
|
Accepted to SIGIR 2020
| null |
10.1145/3397271.3401221
| null |
cs.HC cs.IR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We study whether it is possible to infer if a news headline is true or false
using only the movement of the human eyes when reading news headlines. Our
study with 55 participants who are eye-tracked when reading 108 news headlines
(72 true, 36 false) shows that false headlines receive statistically
significantly less visual attention than true headlines. We further build an
ensemble learner that predicts news headline factuality using only eye-tracking
measurements. Our model yields a mean AUC of 0.688 and is better at detecting
false than true headlines. Through a model analysis, we find that eye-tracking
25 users when reading 3-6 headlines is sufficient for our ensemble learner.
|
[
{
"created": "Wed, 17 Jun 2020 09:24:21 GMT",
"version": "v1"
}
] |
2020-06-18
|
[
[
"Hansen",
"Christian",
""
],
[
"Hansen",
"Casper",
""
],
[
"Simonsen",
"Jakob Grue",
""
],
[
"Larsen",
"Birger",
""
],
[
"Alstrup",
"Stephen",
""
],
[
"Lioma",
"Christina",
""
]
] |
We study whether it is possible to infer if a news headline is true or false using only the movement of the human eyes when reading news headlines. Our study with 55 participants who are eye-tracked when reading 108 news headlines (72 true, 36 false) shows that false headlines receive statistically significantly less visual attention than true headlines. We further build an ensemble learner that predicts news headline factuality using only eye-tracking measurements. Our model yields a mean AUC of 0.688 and is better at detecting false than true headlines. Through a model analysis, we find that eye-tracking 25 users when reading 3-6 headlines is sufficient for our ensemble learner.
|
2306.02648
|
Cosijopii Garc\'ia-Garc\'ia
|
Cosijopii Garcia-Garcia and Alicia Morales-Reyes and Hugo Jair
Escalante
|
Continuous Cartesian Genetic Programming based representation for
Multi-Objective Neural Architecture Search
| null | null | null | null |
cs.NE cs.CV
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
We propose a novel approach for the challenge of designing less complex yet
highly effective convolutional neural networks (CNNs) through the use of
cartesian genetic programming (CGP) for neural architecture search (NAS). Our
approach combines real-based and block-chained CNNs representations based on
CGP for optimization in the continuous domain using multi-objective
evolutionary algorithms (MOEAs). Two variants are introduced that differ in the
granularity of the search space they consider. The proposed CGP-NASV1 and
CGP-NASV2 algorithms were evaluated using the non-dominated sorting genetic
algorithm II (NSGA-II) on the CIFAR-10 and CIFAR-100 datasets. The empirical
analysis was extended to assess the crossover operator from differential
evolution (DE), the multi-objective evolutionary algorithm based on
decomposition (MOEA/D) and S metric selection evolutionary multi-objective
algorithm (SMS-EMOA) using the same representation. Experimental results
demonstrate that our approach is competitive with state-of-the-art proposals in
terms of classification performance and model complexity.
|
[
{
"created": "Mon, 5 Jun 2023 07:32:47 GMT",
"version": "v1"
}
] |
2023-06-06
|
[
[
"Garcia-Garcia",
"Cosijopii",
""
],
[
"Morales-Reyes",
"Alicia",
""
],
[
"Escalante",
"Hugo Jair",
""
]
] |
We propose a novel approach for the challenge of designing less complex yet highly effective convolutional neural networks (CNNs) through the use of cartesian genetic programming (CGP) for neural architecture search (NAS). Our approach combines real-based and block-chained CNNs representations based on CGP for optimization in the continuous domain using multi-objective evolutionary algorithms (MOEAs). Two variants are introduced that differ in the granularity of the search space they consider. The proposed CGP-NASV1 and CGP-NASV2 algorithms were evaluated using the non-dominated sorting genetic algorithm II (NSGA-II) on the CIFAR-10 and CIFAR-100 datasets. The empirical analysis was extended to assess the crossover operator from differential evolution (DE), the multi-objective evolutionary algorithm based on decomposition (MOEA/D) and S metric selection evolutionary multi-objective algorithm (SMS-EMOA) using the same representation. Experimental results demonstrate that our approach is competitive with state-of-the-art proposals in terms of classification performance and model complexity.
|
2011.11095
|
Gevorg Yeghikyan
|
Gevorg Yeghikyan
|
How will AI and automation transform society and cities?
|
8 pages, 1 figure
| null | null | null |
cs.CY
|
http://creativecommons.org/licenses/by/4.0/
|
Against the backdrop of rising anxiety and discussions on the impact of AI on
society, I explore in this article the structural possibilities of AI and
automation triggering a new social conflict between the current capitalist
elites and the emerging "creative class" (R&D scientists, engineers, business
developers, etc.), and how this conflict can produce social tensions and
transform urban space. By drawing insights from a structurally similar conflict
in 17-18th century Europe between the aristocracy and the emerging bourgeoisie,
the impact of this conflict on the social, spatial, and power landscapes in
cities of that time, as well as current trends in urban geography, this article
outlines the prospects of urban transformations under changing production and
consumption economies.
|
[
{
"created": "Sun, 22 Nov 2020 19:44:51 GMT",
"version": "v1"
}
] |
2020-11-24
|
[
[
"Yeghikyan",
"Gevorg",
""
]
] |
Against the backdrop of rising anxiety and discussions on the impact of AI on society, I explore in this article the structural possibilities of AI and automation triggering a new social conflict between the current capitalist elites and the emerging "creative class" (R&D scientists, engineers, business developers, etc.), and how this conflict can produce social tensions and transform urban space. By drawing insights from a structurally similar conflict in 17-18th century Europe between the aristocracy and the emerging bourgeoisie, the impact of this conflict on the social, spatial, and power landscapes in cities of that time, as well as current trends in urban geography, this article outlines the prospects of urban transformations under changing production and consumption economies.
|
2101.03477
|
Peter Washington
|
Peter Washington, Onur Cezmi Mutlu, Emilie Leblanc, Aaron Kline, Cathy
Hou, Brianna Chrisman, Nate Stockham, Kelley Paskov, Catalin Voss, Nick
Haber, Dennis Wall
|
Training Affective Computer Vision Models by Crowdsourcing Soft-Target
Labels
| null | null | null | null |
cs.CV cs.HC
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Emotion classifiers traditionally predict discrete emotions. However, emotion
expressions are often subjective, thus requiring a method to handle subjective
labels. We explore the use of crowdsourcing to acquire reliable soft-target
labels and evaluate an emotion detection classifier trained with these labels.
We center our study on the Child Affective Facial Expression (CAFE) dataset, a
gold standard collection of images depicting pediatric facial expressions along
with 100 human labels per image. To test the feasibility of crowdsourcing to
generate these labels, we used Microworkers to acquire labels for 207 CAFE
images. We evaluate both unfiltered workers as well as workers selected through
a short crowd filtration process. We then train two versions of a classifiers
on soft-target CAFE labels using the original 100 annotations provided with the
dataset: (1) a classifier trained with traditional one-hot encoded labels, and
(2) a classifier trained with vector labels representing the distribution of
CAFE annotator responses. We compare the resulting softmax output distributions
of the two classifiers with a 2-sample independent t-test of L1 distances
between the classifier's output probability distribution and the distribution
of human labels. While agreement with CAFE is weak for unfiltered crowd
workers, the filtered crowd agree with the CAFE labels 100% of the time for
many emotions. While the F1-score for a one-hot encoded classifier is much
higher (94.33% vs. 78.68%) with respect to the ground truth CAFE labels, the
output probability vector of the crowd-trained classifier more closely
resembles the distribution of human labels (t=3.2827, p=0.0014). Reporting an
emotion probability distribution that accounts for the subjectivity of human
interpretation. Crowdsourcing, including a sufficient filtering mechanism, is a
feasible solution for acquiring soft-target labels.
|
[
{
"created": "Sun, 10 Jan 2021 05:26:55 GMT",
"version": "v1"
},
{
"created": "Wed, 22 Sep 2021 23:12:50 GMT",
"version": "v2"
}
] |
2021-09-24
|
[
[
"Washington",
"Peter",
""
],
[
"Mutlu",
"Onur Cezmi",
""
],
[
"Leblanc",
"Emilie",
""
],
[
"Kline",
"Aaron",
""
],
[
"Hou",
"Cathy",
""
],
[
"Chrisman",
"Brianna",
""
],
[
"Stockham",
"Nate",
""
],
[
"Paskov",
"Kelley",
""
],
[
"Voss",
"Catalin",
""
],
[
"Haber",
"Nick",
""
],
[
"Wall",
"Dennis",
""
]
] |
Emotion classifiers traditionally predict discrete emotions. However, emotion expressions are often subjective, thus requiring a method to handle subjective labels. We explore the use of crowdsourcing to acquire reliable soft-target labels and evaluate an emotion detection classifier trained with these labels. We center our study on the Child Affective Facial Expression (CAFE) dataset, a gold standard collection of images depicting pediatric facial expressions along with 100 human labels per image. To test the feasibility of crowdsourcing to generate these labels, we used Microworkers to acquire labels for 207 CAFE images. We evaluate both unfiltered workers as well as workers selected through a short crowd filtration process. We then train two versions of a classifiers on soft-target CAFE labels using the original 100 annotations provided with the dataset: (1) a classifier trained with traditional one-hot encoded labels, and (2) a classifier trained with vector labels representing the distribution of CAFE annotator responses. We compare the resulting softmax output distributions of the two classifiers with a 2-sample independent t-test of L1 distances between the classifier's output probability distribution and the distribution of human labels. While agreement with CAFE is weak for unfiltered crowd workers, the filtered crowd agree with the CAFE labels 100% of the time for many emotions. While the F1-score for a one-hot encoded classifier is much higher (94.33% vs. 78.68%) with respect to the ground truth CAFE labels, the output probability vector of the crowd-trained classifier more closely resembles the distribution of human labels (t=3.2827, p=0.0014). Reporting an emotion probability distribution that accounts for the subjectivity of human interpretation. Crowdsourcing, including a sufficient filtering mechanism, is a feasible solution for acquiring soft-target labels.
|
1412.2342
|
Hayaru Shouno
|
Hayaru Shouno
|
Bayesian Image Restoration for Poisson Corrupted Image using a Latent
Variational Method with Gaussian MRF
|
9 pages, 6 figures, The of this manuscript is submitting to the
Information Processing Society of Japan(IPSJ), Transactions on Mathematical
Modeling and its Applications (TOM)
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We treat an image restoration problem with a Poisson noise chan- nel using a
Bayesian framework. The Poisson randomness might be appeared in observation of
low contrast object in the field of imaging. The noise observation is often
hard to treat in a theo- retical analysis. In our formulation, we interpret the
observation through the Poisson noise channel as a likelihood, and evaluate the
bound of it with a Gaussian function using a latent variable method. We then
introduce a Gaussian Markov random field (GMRF) as the prior for the Bayesian
approach, and derive the posterior as a Gaussian distribution. The latent
parameters in the likelihood and the hyperparameter in the GMRF prior could be
treated as hid- den parameters, so that, we propose an algorithm to infer them
in the expectation maximization (EM) framework using loopy belief
propagation(LBP). We confirm the ability of our algorithm in the computer
simulation, and compare it with the results of other im- age restoration
frameworks.
|
[
{
"created": "Sun, 7 Dec 2014 10:59:55 GMT",
"version": "v1"
}
] |
2014-12-09
|
[
[
"Shouno",
"Hayaru",
""
]
] |
We treat an image restoration problem with a Poisson noise chan- nel using a Bayesian framework. The Poisson randomness might be appeared in observation of low contrast object in the field of imaging. The noise observation is often hard to treat in a theo- retical analysis. In our formulation, we interpret the observation through the Poisson noise channel as a likelihood, and evaluate the bound of it with a Gaussian function using a latent variable method. We then introduce a Gaussian Markov random field (GMRF) as the prior for the Bayesian approach, and derive the posterior as a Gaussian distribution. The latent parameters in the likelihood and the hyperparameter in the GMRF prior could be treated as hid- den parameters, so that, we propose an algorithm to infer them in the expectation maximization (EM) framework using loopy belief propagation(LBP). We confirm the ability of our algorithm in the computer simulation, and compare it with the results of other im- age restoration frameworks.
|
2311.00278
|
Min Jae Jung
|
Min Jae Jung, Seung Dae Han and Joohee Kim
|
Re-Scoring Using Image-Language Similarity for Few-Shot Object Detection
|
19 pages, 11 figures
| null |
10.1016/j.cviu.2024.103956
| null |
cs.CV cs.AI
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Few-shot object detection, which focuses on detecting novel objects with few
labels, is an emerging challenge in the community. Recent studies show that
adapting a pre-trained model or modified loss function can improve performance.
In this paper, we explore leveraging the power of Contrastive Language-Image
Pre-training (CLIP) and hard negative classification loss in low data setting.
Specifically, we propose Re-scoring using Image-language Similarity for
Few-shot object detection (RISF) which extends Faster R-CNN by introducing
Calibration Module using CLIP (CM-CLIP) and Background Negative Re-scale Loss
(BNRL). The former adapts CLIP, which performs zero-shot classification, to
re-score the classification scores of a detector using image-class
similarities, the latter is modified classification loss considering the
punishment for fake backgrounds as well as confusing categories on a
generalized few-shot object detection dataset. Extensive experiments on MS-COCO
and PASCAL VOC show that the proposed RISF substantially outperforms the
state-of-the-art approaches. The code will be available.
|
[
{
"created": "Wed, 1 Nov 2023 04:04:34 GMT",
"version": "v1"
}
] |
2024-07-25
|
[
[
"Jung",
"Min Jae",
""
],
[
"Han",
"Seung Dae",
""
],
[
"Kim",
"Joohee",
""
]
] |
Few-shot object detection, which focuses on detecting novel objects with few labels, is an emerging challenge in the community. Recent studies show that adapting a pre-trained model or modified loss function can improve performance. In this paper, we explore leveraging the power of Contrastive Language-Image Pre-training (CLIP) and hard negative classification loss in low data setting. Specifically, we propose Re-scoring using Image-language Similarity for Few-shot object detection (RISF) which extends Faster R-CNN by introducing Calibration Module using CLIP (CM-CLIP) and Background Negative Re-scale Loss (BNRL). The former adapts CLIP, which performs zero-shot classification, to re-score the classification scores of a detector using image-class similarities, the latter is modified classification loss considering the punishment for fake backgrounds as well as confusing categories on a generalized few-shot object detection dataset. Extensive experiments on MS-COCO and PASCAL VOC show that the proposed RISF substantially outperforms the state-of-the-art approaches. The code will be available.
|
1509.02620
|
Dharmendra Dixit
|
Dharmendra Dixit and P. R. Sahu
|
Performance of QAM Schemes with Dual-Hop DF Relaying Systems over Mixed
$\eta$-$\mu$ and $\kappa$-$\mu$ Fading Channels
|
25
| null | null | null |
cs.IT math.IT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Performance of quadrature amplitude modulation (QAM) schemes is analyzed with
dual-hop decode-and-forward (DF) relaying systems over mixed $\eta$-$\mu$ and
$\kappa$-$\mu$ fading channels. Closed-form expressions are obtained for the
average symbol error rate (ASER) for general order rectangular QAM and cross
QAM schemes using moment generating function based approach. Derived
expressions are in the form of Lauricella's $(F_D^{(n)}(\cdot),
\Phi_1^{(n)}(\cdot))$ hypergeometric functions which can be numerically
evaluated using either integral or series representation. The obtained ASER
expressions include other mixed fading channel cases addressed in the
literature as special cases such as mixed Hoyt, and Rice fading, mixed
Nakagami-$m$, and Rice fading. We further obtain a simple expression for the
asymptotic ASER, which is useful to determine a factor governing the system
performance at high SNRs, i.e., the diversity order. Additionally, we analyze
the optimal power allocation, which provides a practical design rule to
optimally distribute the total transmission power between the source and the
relay to minimize the ASER. Extensive numerical and computer simulation results
are presented that confirm the accuracy of presented mathematical analysis.
|
[
{
"created": "Wed, 9 Sep 2015 03:14:41 GMT",
"version": "v1"
}
] |
2015-09-10
|
[
[
"Dixit",
"Dharmendra",
""
],
[
"Sahu",
"P. R.",
""
]
] |
Performance of quadrature amplitude modulation (QAM) schemes is analyzed with dual-hop decode-and-forward (DF) relaying systems over mixed $\eta$-$\mu$ and $\kappa$-$\mu$ fading channels. Closed-form expressions are obtained for the average symbol error rate (ASER) for general order rectangular QAM and cross QAM schemes using moment generating function based approach. Derived expressions are in the form of Lauricella's $(F_D^{(n)}(\cdot), \Phi_1^{(n)}(\cdot))$ hypergeometric functions which can be numerically evaluated using either integral or series representation. The obtained ASER expressions include other mixed fading channel cases addressed in the literature as special cases such as mixed Hoyt, and Rice fading, mixed Nakagami-$m$, and Rice fading. We further obtain a simple expression for the asymptotic ASER, which is useful to determine a factor governing the system performance at high SNRs, i.e., the diversity order. Additionally, we analyze the optimal power allocation, which provides a practical design rule to optimally distribute the total transmission power between the source and the relay to minimize the ASER. Extensive numerical and computer simulation results are presented that confirm the accuracy of presented mathematical analysis.
|
2205.00840
|
Volker Nannen
|
Volker Nannen and Damian Bover
|
Traction of Interlocking Spikes on a Granular Material
| null |
Earth and Space 2022, 18th Biennial International Conference on
Engineering, Science, Construction, and Operations in Challenging
Environments
|
10.1061/9780784484470.009
| null |
cs.RO cond-mat.soft
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The interlock drive system generates traction by inserting narrow articulated
spikes into the ground and by leveraging the soil's strength to resist
horizontal draft forces. The system promises high tractive performance in low
gravity environments where tires have little traction for lack of weight. At
Earth and Space 2021 we reported the performance of such spikes on a silty clay
loam, a cohesive soil. We found that in such soil, traction below a critical
depth is provided by a zone of lateral soil failure. We also found that the
articulation translates a horizontal draft force into a vertical penetration
force strong enough to penetrate a narrow spike to a depth where the soil can
sustain the draft force, in a self-regulating way. It is conceivable that a
granular material like regolith or sand with little to no cohesive strength
provides less vertical penetration resistance and less resistance to a
horizontal draft force than a cohesive soil, which leads to the question of
whether and how much tractive force an interlocking spike can generate on a
granular material. Here we report on field trials that study different spike
designs in dry and unsaturated moist sand. The results demonstrate that a loose
granular material requires larger spikes than a cohesive soil, that these
larger spikes penetrate dry and moist sand reliably, and that they promise good
tractive efficiency. The trials indicate that on sand, a larger spike diameter
can improve the pull/weight ratio without a loss of tractive performance.
|
[
{
"created": "Sat, 2 Apr 2022 22:07:10 GMT",
"version": "v1"
},
{
"created": "Fri, 21 Apr 2023 19:02:38 GMT",
"version": "v2"
}
] |
2023-04-25
|
[
[
"Nannen",
"Volker",
""
],
[
"Bover",
"Damian",
""
]
] |
The interlock drive system generates traction by inserting narrow articulated spikes into the ground and by leveraging the soil's strength to resist horizontal draft forces. The system promises high tractive performance in low gravity environments where tires have little traction for lack of weight. At Earth and Space 2021 we reported the performance of such spikes on a silty clay loam, a cohesive soil. We found that in such soil, traction below a critical depth is provided by a zone of lateral soil failure. We also found that the articulation translates a horizontal draft force into a vertical penetration force strong enough to penetrate a narrow spike to a depth where the soil can sustain the draft force, in a self-regulating way. It is conceivable that a granular material like regolith or sand with little to no cohesive strength provides less vertical penetration resistance and less resistance to a horizontal draft force than a cohesive soil, which leads to the question of whether and how much tractive force an interlocking spike can generate on a granular material. Here we report on field trials that study different spike designs in dry and unsaturated moist sand. The results demonstrate that a loose granular material requires larger spikes than a cohesive soil, that these larger spikes penetrate dry and moist sand reliably, and that they promise good tractive efficiency. The trials indicate that on sand, a larger spike diameter can improve the pull/weight ratio without a loss of tractive performance.
|
1304.4326
|
Himanshu Chauhan
|
Himanshu Chauhan, Vijay K. Garg, Aravind Natarajan, Neeraj Mittal
|
Distributed Abstraction Algorithm for Online Predicate Detection
| null | null | null | null |
cs.DC
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Analyzing a distributed computation is a hard problem in general due to the
combinatorial explosion in the size of the state-space with the number of
processes in the system. By abstracting the computation, unnecessary
explorations can be avoided. Computation slicing is an approach for abstracting
dis- tributed computations with respect to a given predicate. We focus on
regular predicates, a family of predicates that covers a large number of
commonly used predicates for runtime verification. The existing algorithms for
computation slicing are centralized in nature in which a single process is
responsible for computing the slice in either offline or online manner. In this
paper, we present a distributed online algorithm for computing the slice of a
distributed computation with respect to a regular predicate. Our algorithm
distributes the work and storage requirements across the system, thus reducing
the space and computation complexities per process. In addition, for
conjunctive predicates, our algorithm also reduces the message load per
process.
|
[
{
"created": "Tue, 16 Apr 2013 03:56:24 GMT",
"version": "v1"
},
{
"created": "Fri, 31 May 2013 04:03:28 GMT",
"version": "v2"
},
{
"created": "Tue, 4 Jun 2013 06:23:50 GMT",
"version": "v3"
}
] |
2013-06-05
|
[
[
"Chauhan",
"Himanshu",
""
],
[
"Garg",
"Vijay K.",
""
],
[
"Natarajan",
"Aravind",
""
],
[
"Mittal",
"Neeraj",
""
]
] |
Analyzing a distributed computation is a hard problem in general due to the combinatorial explosion in the size of the state-space with the number of processes in the system. By abstracting the computation, unnecessary explorations can be avoided. Computation slicing is an approach for abstracting dis- tributed computations with respect to a given predicate. We focus on regular predicates, a family of predicates that covers a large number of commonly used predicates for runtime verification. The existing algorithms for computation slicing are centralized in nature in which a single process is responsible for computing the slice in either offline or online manner. In this paper, we present a distributed online algorithm for computing the slice of a distributed computation with respect to a regular predicate. Our algorithm distributes the work and storage requirements across the system, thus reducing the space and computation complexities per process. In addition, for conjunctive predicates, our algorithm also reduces the message load per process.
|
2011.00717
|
Hongyuan Mei
|
Hongyuan Mei, Tom Wan, Jason Eisner
|
Noise-Contrastive Estimation for Multivariate Point Processes
|
NeurIPS 2020 camera-ready
| null | null | null |
cs.LG stat.ML
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The log-likelihood of a generative model often involves both positive and
negative terms. For a temporal multivariate point process, the negative term
sums over all the possible event types at each time and also integrates over
all the possible times. As a result, maximum likelihood estimation is
expensive. We show how to instead apply a version of noise-contrastive
estimation---a general parameter estimation method with a less expensive
stochastic objective. Our specific instantiation of this general idea works out
in an interestingly non-trivial way and has provable guarantees for its
optimality, consistency and efficiency. On several synthetic and real-world
datasets, our method shows benefits: for the model to achieve the same level of
log-likelihood on held-out data, our method needs considerably fewer function
evaluations and less wall-clock time.
|
[
{
"created": "Mon, 2 Nov 2020 04:09:33 GMT",
"version": "v1"
}
] |
2020-11-03
|
[
[
"Mei",
"Hongyuan",
""
],
[
"Wan",
"Tom",
""
],
[
"Eisner",
"Jason",
""
]
] |
The log-likelihood of a generative model often involves both positive and negative terms. For a temporal multivariate point process, the negative term sums over all the possible event types at each time and also integrates over all the possible times. As a result, maximum likelihood estimation is expensive. We show how to instead apply a version of noise-contrastive estimation---a general parameter estimation method with a less expensive stochastic objective. Our specific instantiation of this general idea works out in an interestingly non-trivial way and has provable guarantees for its optimality, consistency and efficiency. On several synthetic and real-world datasets, our method shows benefits: for the model to achieve the same level of log-likelihood on held-out data, our method needs considerably fewer function evaluations and less wall-clock time.
|
1810.13320
|
Longyue Wang
|
Baosong Yang, Longyue Wang, Derek F. Wong, Lidia S. Chao, Zhaopeng Tu
|
Convolutional Self-Attention Network
|
The least version of this paper has been uploaded to another link:
arXiv:1904.03107
| null | null | null |
cs.CL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Self-attention network (SAN) has recently attracted increasing interest due
to its fully parallelized computation and flexibility in modeling dependencies.
It can be further enhanced with multi-headed attention mechanism by allowing
the model to jointly attend to information from different representation
subspaces at different positions (Vaswani et al., 2017). In this work, we
propose a novel convolutional self-attention network (CSAN), which offers SAN
the abilities to 1) capture neighboring dependencies, and 2) model the
interaction between multiple attention heads. Experimental results on WMT14
English-to-German translation task demonstrate that the proposed approach
outperforms both the strong Transformer baseline and other existing works on
enhancing the locality of SAN. Comparing with previous work, our model does not
introduce any new parameters.
|
[
{
"created": "Wed, 31 Oct 2018 14:58:30 GMT",
"version": "v1"
},
{
"created": "Mon, 8 Apr 2019 09:15:30 GMT",
"version": "v2"
}
] |
2019-04-09
|
[
[
"Yang",
"Baosong",
""
],
[
"Wang",
"Longyue",
""
],
[
"Wong",
"Derek F.",
""
],
[
"Chao",
"Lidia S.",
""
],
[
"Tu",
"Zhaopeng",
""
]
] |
Self-attention network (SAN) has recently attracted increasing interest due to its fully parallelized computation and flexibility in modeling dependencies. It can be further enhanced with multi-headed attention mechanism by allowing the model to jointly attend to information from different representation subspaces at different positions (Vaswani et al., 2017). In this work, we propose a novel convolutional self-attention network (CSAN), which offers SAN the abilities to 1) capture neighboring dependencies, and 2) model the interaction between multiple attention heads. Experimental results on WMT14 English-to-German translation task demonstrate that the proposed approach outperforms both the strong Transformer baseline and other existing works on enhancing the locality of SAN. Comparing with previous work, our model does not introduce any new parameters.
|
1607.05671
|
Shankara Narayanan Krishna
|
S Akshay, Patricia Bouyer, Shankara Narayanan Krishna, Lakshmi Manasa,
Ashutosh Trivedi
|
Stochastic Timed Games Revisited
| null | null | null | null |
cs.LO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Stochastic timed games (STGs), introduced by Bouyer and Forejt, naturally
generalize both continuous-time Markov chains and timed automata by providing a
partition of the locations between those controlled by two players (Player Box
and Player Diamond) with competing objectives and those governed by stochastic
laws. Depending on the number of players---$2$, $1$, or $0$---subclasses of
stochastic timed games are often classified as $2\frac{1}{2}$-player,
$1\frac{1}{2}$-player, and $\frac{1}{2}$-player games where the $\frac{1}{2}$
symbolizes the presence of the stochastic "nature" player. For STGs with
reachability objectives it is known that $1\frac{1}{2}$-player one-clock STGs
are decidable for qualitative objectives, and that $2\frac{1}{2}$-player
three-clock STGs are undecidable for quantitative reachability objectives. This
paper further refines the gap in this decidability spectrum. We show that
quantitative reachability objectives are already undecidable for $1\frac{1}{2}$
player four-clock STGs, and even under the time-bounded restriction for
$2\frac{1}{2}$-player five-clock STGs. We also obtain a class of
$1\frac{1}{2}$, $2\frac{1}{2}$ player STGs for which the quantitative
reachability problem is decidable.
|
[
{
"created": "Tue, 19 Jul 2016 17:27:14 GMT",
"version": "v1"
}
] |
2016-07-20
|
[
[
"Akshay",
"S",
""
],
[
"Bouyer",
"Patricia",
""
],
[
"Krishna",
"Shankara Narayanan",
""
],
[
"Manasa",
"Lakshmi",
""
],
[
"Trivedi",
"Ashutosh",
""
]
] |
Stochastic timed games (STGs), introduced by Bouyer and Forejt, naturally generalize both continuous-time Markov chains and timed automata by providing a partition of the locations between those controlled by two players (Player Box and Player Diamond) with competing objectives and those governed by stochastic laws. Depending on the number of players---$2$, $1$, or $0$---subclasses of stochastic timed games are often classified as $2\frac{1}{2}$-player, $1\frac{1}{2}$-player, and $\frac{1}{2}$-player games where the $\frac{1}{2}$ symbolizes the presence of the stochastic "nature" player. For STGs with reachability objectives it is known that $1\frac{1}{2}$-player one-clock STGs are decidable for qualitative objectives, and that $2\frac{1}{2}$-player three-clock STGs are undecidable for quantitative reachability objectives. This paper further refines the gap in this decidability spectrum. We show that quantitative reachability objectives are already undecidable for $1\frac{1}{2}$ player four-clock STGs, and even under the time-bounded restriction for $2\frac{1}{2}$-player five-clock STGs. We also obtain a class of $1\frac{1}{2}$, $2\frac{1}{2}$ player STGs for which the quantitative reachability problem is decidable.
|
1805.11598
|
Phoebe Mulcaire
|
Phoebe Mulcaire, Swabha Swayamdipta, Noah Smith
|
Polyglot Semantic Role Labeling
|
To appear at ACL 2018
| null | null | null |
cs.CL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Previous approaches to multilingual semantic dependency parsing treat
languages independently, without exploiting the similarities between semantic
structures across languages. We experiment with a new approach where we combine
resources from a pair of languages in the CoNLL 2009 shared task to build a
polyglot semantic role labeler. Notwithstanding the absence of parallel data,
and the dissimilarity in annotations between languages, our approach results in
an improvement in SRL performance on multiple languages over a monolingual
baseline. Analysis of the polyglot model shows it to be advantageous in
lower-resource settings.
|
[
{
"created": "Tue, 29 May 2018 17:29:55 GMT",
"version": "v1"
}
] |
2018-05-30
|
[
[
"Mulcaire",
"Phoebe",
""
],
[
"Swayamdipta",
"Swabha",
""
],
[
"Smith",
"Noah",
""
]
] |
Previous approaches to multilingual semantic dependency parsing treat languages independently, without exploiting the similarities between semantic structures across languages. We experiment with a new approach where we combine resources from a pair of languages in the CoNLL 2009 shared task to build a polyglot semantic role labeler. Notwithstanding the absence of parallel data, and the dissimilarity in annotations between languages, our approach results in an improvement in SRL performance on multiple languages over a monolingual baseline. Analysis of the polyglot model shows it to be advantageous in lower-resource settings.
|
2104.12141
|
Abhinandan Nath
|
Abhinandan Nath
|
Coresets for $k$-median clustering under Fr\'{e}chet and Hausdorff
distances
| null | null | null | null |
cs.CG
|
http://creativecommons.org/licenses/by/4.0/
|
We give algorithms for computing coresets for $(1+\varepsilon)$-approximate
$k$-median clustering of polygonal curves (under the discrete and continuous
Fr\'{e}chet distance) and point sets (under the Hausdorff distance), when the
cluster centers are restricted to be of low complexity. Ours is the first such
result, where the size of the coreset is independent of the number of input
curves/point sets to be clustered (although it still depends on the maximum
complexity of each input object). Specifically, the size of the coreset is
$\Theta\left(\frac{k^3lm^{\delta}d}{\varepsilon^2}\log\left(
\frac{kl}{\varepsilon}\right)\right)$ for any $\delta > 0$, where $d$ is the
ambient dimension, $m$ is the maximum number of points in an input curve/point
set, and $l$ is the maximum number of points allowed in a cluster center. We
formally characterize a general condition on the restricted space of cluster
centers -- this helps us to generalize and apply the importance sampling
framework, that was used by Langberg and Schulman for computing coresets for
$k$-median clustering of $d$-dimensional points on normed spaces in
$\mathbb{R}^d$, to the problem of clustering curves and point sets using the
Fr\'{e}chet and Hausdorff metrics. Roughly, the condition places an upper bound
on the number of different combinations of metric balls that the restricted
space of cluster centers can hit. We also derive lower bounds on the size of
the coreset, given the restriction that the coreset must be a subset of the
input objects.
|
[
{
"created": "Sun, 25 Apr 2021 12:27:05 GMT",
"version": "v1"
}
] |
2021-04-27
|
[
[
"Nath",
"Abhinandan",
""
]
] |
We give algorithms for computing coresets for $(1+\varepsilon)$-approximate $k$-median clustering of polygonal curves (under the discrete and continuous Fr\'{e}chet distance) and point sets (under the Hausdorff distance), when the cluster centers are restricted to be of low complexity. Ours is the first such result, where the size of the coreset is independent of the number of input curves/point sets to be clustered (although it still depends on the maximum complexity of each input object). Specifically, the size of the coreset is $\Theta\left(\frac{k^3lm^{\delta}d}{\varepsilon^2}\log\left( \frac{kl}{\varepsilon}\right)\right)$ for any $\delta > 0$, where $d$ is the ambient dimension, $m$ is the maximum number of points in an input curve/point set, and $l$ is the maximum number of points allowed in a cluster center. We formally characterize a general condition on the restricted space of cluster centers -- this helps us to generalize and apply the importance sampling framework, that was used by Langberg and Schulman for computing coresets for $k$-median clustering of $d$-dimensional points on normed spaces in $\mathbb{R}^d$, to the problem of clustering curves and point sets using the Fr\'{e}chet and Hausdorff metrics. Roughly, the condition places an upper bound on the number of different combinations of metric balls that the restricted space of cluster centers can hit. We also derive lower bounds on the size of the coreset, given the restriction that the coreset must be a subset of the input objects.
|
2404.03164
|
Haonan Zhang
|
Haonan Zhang, Dongxia Wang, Zhu Sun, Yanhui Li, Youcheng Sun, Huizhi
Liang, Wenhai Wang
|
Does Knowledge Graph Really Matter for Recommender Systems?
| null | null | null | null |
cs.IR cs.AI cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Recommender systems (RSs) are designed to provide personalized
recommendations to users. Recently, knowledge graphs (KGs) have been widely
introduced in RSs to improve recommendation accuracy. In this study, however,
we demonstrate that RSs do not necessarily perform worse even if the KG is
downgraded to the user-item interaction graph only (or removed). We propose an
evaluation framework KG4RecEval to systematically evaluate how much a KG
contributes to the recommendation accuracy of a KG-based RS, using our defined
metric KGER (KG utilization efficiency in recommendation). We consider the
scenarios where knowledge in a KG gets completely removed, randomly distorted
and decreased, and also where recommendations are for cold-start users. Our
extensive experiments on four commonly used datasets and a number of
state-of-the-art KG-based RSs reveal that: to remove, randomly distort or
decrease knowledge does not necessarily decrease recommendation accuracy, even
for cold-start users. These findings inspire us to rethink how to better
utilize knowledge from existing KGs, whereby we discuss and provide insights
into what characteristics of datasets and KG-based RSs may help improve KG
utilization efficiency.
|
[
{
"created": "Thu, 4 Apr 2024 02:32:58 GMT",
"version": "v1"
}
] |
2024-04-05
|
[
[
"Zhang",
"Haonan",
""
],
[
"Wang",
"Dongxia",
""
],
[
"Sun",
"Zhu",
""
],
[
"Li",
"Yanhui",
""
],
[
"Sun",
"Youcheng",
""
],
[
"Liang",
"Huizhi",
""
],
[
"Wang",
"Wenhai",
""
]
] |
Recommender systems (RSs) are designed to provide personalized recommendations to users. Recently, knowledge graphs (KGs) have been widely introduced in RSs to improve recommendation accuracy. In this study, however, we demonstrate that RSs do not necessarily perform worse even if the KG is downgraded to the user-item interaction graph only (or removed). We propose an evaluation framework KG4RecEval to systematically evaluate how much a KG contributes to the recommendation accuracy of a KG-based RS, using our defined metric KGER (KG utilization efficiency in recommendation). We consider the scenarios where knowledge in a KG gets completely removed, randomly distorted and decreased, and also where recommendations are for cold-start users. Our extensive experiments on four commonly used datasets and a number of state-of-the-art KG-based RSs reveal that: to remove, randomly distort or decrease knowledge does not necessarily decrease recommendation accuracy, even for cold-start users. These findings inspire us to rethink how to better utilize knowledge from existing KGs, whereby we discuss and provide insights into what characteristics of datasets and KG-based RSs may help improve KG utilization efficiency.
|
2312.03558
|
Wenhui Wang
|
Wenhui Wang, Shuming Ma, Hanwen Xu, Naoto Usuyama, Jiayu Ding, Hoifung
Poon, Furu Wei
|
When an Image is Worth 1,024 x 1,024 Words: A Case Study in
Computational Pathology
| null | null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
This technical report presents LongViT, a vision Transformer that can process
gigapixel images in an end-to-end manner. Specifically, we split the gigapixel
image into a sequence of millions of patches and project them linearly into
embeddings. LongNet is then employed to model the extremely long sequence,
generating representations that capture both short-range and long-range
dependencies. The linear computation complexity of LongNet, along with its
distributed algorithm, enables us to overcome the constraints of both
computation and memory. We apply LongViT in the field of computational
pathology, aiming for cancer diagnosis and prognosis within gigapixel
whole-slide images. Experimental results demonstrate that LongViT effectively
encodes gigapixel images and outperforms previous state-of-the-art methods on
cancer subtyping and survival prediction. Code and models will be available at
https://aka.ms/LongViT.
|
[
{
"created": "Wed, 6 Dec 2023 15:40:28 GMT",
"version": "v1"
}
] |
2023-12-07
|
[
[
"Wang",
"Wenhui",
""
],
[
"Ma",
"Shuming",
""
],
[
"Xu",
"Hanwen",
""
],
[
"Usuyama",
"Naoto",
""
],
[
"Ding",
"Jiayu",
""
],
[
"Poon",
"Hoifung",
""
],
[
"Wei",
"Furu",
""
]
] |
This technical report presents LongViT, a vision Transformer that can process gigapixel images in an end-to-end manner. Specifically, we split the gigapixel image into a sequence of millions of patches and project them linearly into embeddings. LongNet is then employed to model the extremely long sequence, generating representations that capture both short-range and long-range dependencies. The linear computation complexity of LongNet, along with its distributed algorithm, enables us to overcome the constraints of both computation and memory. We apply LongViT in the field of computational pathology, aiming for cancer diagnosis and prognosis within gigapixel whole-slide images. Experimental results demonstrate that LongViT effectively encodes gigapixel images and outperforms previous state-of-the-art methods on cancer subtyping and survival prediction. Code and models will be available at https://aka.ms/LongViT.
|
2310.11644
|
Morteza Fayazi
|
Serafina Kamp, Morteza Fayazi, Zineb Benameur-El, Shuyan Yu, Ronald
Dreslinski
|
Open Information Extraction: A Review of Baseline Techniques,
Approaches, and Applications
|
15 pages, 9 figures
| null | null | null |
cs.IR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
With the abundant amount of available online and offline text data, there
arises a crucial need to extract the relation between phrases and summarize the
main content of each document in a few words. For this purpose, there have been
many studies recently in Open Information Extraction (OIE). OIE improves upon
relation extraction techniques by analyzing relations across different domains
and avoids requiring hand-labeling pre-specified relations in sentences. This
paper surveys recent approaches of OIE and its applications on Knowledge Graph
(KG), text summarization, and Question Answering (QA). Moreover, the paper
describes OIE basis methods in relation extraction. It briefly discusses the
main approaches and the pros and cons of each method. Finally, it gives an
overview about challenges, open issues, and future work opportunities for OIE,
relation extraction, and OIE applications.
|
[
{
"created": "Wed, 18 Oct 2023 01:06:01 GMT",
"version": "v1"
}
] |
2023-10-19
|
[
[
"Kamp",
"Serafina",
""
],
[
"Fayazi",
"Morteza",
""
],
[
"Benameur-El",
"Zineb",
""
],
[
"Yu",
"Shuyan",
""
],
[
"Dreslinski",
"Ronald",
""
]
] |
With the abundant amount of available online and offline text data, there arises a crucial need to extract the relation between phrases and summarize the main content of each document in a few words. For this purpose, there have been many studies recently in Open Information Extraction (OIE). OIE improves upon relation extraction techniques by analyzing relations across different domains and avoids requiring hand-labeling pre-specified relations in sentences. This paper surveys recent approaches of OIE and its applications on Knowledge Graph (KG), text summarization, and Question Answering (QA). Moreover, the paper describes OIE basis methods in relation extraction. It briefly discusses the main approaches and the pros and cons of each method. Finally, it gives an overview about challenges, open issues, and future work opportunities for OIE, relation extraction, and OIE applications.
|
2310.18042
|
Alberto Sonnino
|
Sam Blackshear, Andrey Chursin, George Danezis, Anastasios Kichidis,
Lefteris Kokoris-Kogias, Xun Li, Mark Logan, Ashok Menon, Todd Nowacki,
Alberto Sonnino, Brandon Williams, Lu Zhang
|
Sui Lutris: A Blockchain Combining Broadcast and Consensus
| null | null | null | null |
cs.DC cs.CR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Sui Lutris is the first smart-contract platform to sustainably achieve
sub-second finality. It achieves this significant decrease by employing
consensusless agreement not only for simple payments but for a large variety of
transactions. Unlike prior work, Sui Lutris neither compromises expressiveness
nor throughput and can run perpetually without restarts. Sui Lutris achieves
this by safely integrating consensuless agreement with a high-throughput
consensus protocol that is invoked out of the critical finality path but
ensures that when a transaction is at risk of inconsistent concurrent accesses,
its settlement is delayed until the total ordering is resolved. Building such a
hybrid architecture is especially delicate during reconfiguration events, where
the system needs to preserve the safety of the consensusless path without
compromising the long-term liveness of potentially misconfigured clients. We
thus develop a novel reconfiguration protocol, the first to provably show the
safe and efficient reconfiguration of a consensusless blockchain. Sui Lutris is
currently running in production and underpins the Sui smart-contract platform.
Combined with the use of Objects instead of accounts it enables the safe
execution of smart contracts that expose objects as a first-class resource. In
our experiments Sui Lutris achieves latency lower than 0.5 seconds for
throughput up to 5,000 certificates per second (150k ops/s with transaction
blocks), compared to the state-of-the-art real-world consensus latencies of 3
seconds. Furthermore, it gracefully handles validators crash-recovery and does
not suffer visible performance degradation during reconfiguration.
|
[
{
"created": "Fri, 27 Oct 2023 10:40:11 GMT",
"version": "v1"
},
{
"created": "Wed, 1 May 2024 13:14:03 GMT",
"version": "v2"
},
{
"created": "Mon, 6 May 2024 10:50:44 GMT",
"version": "v3"
},
{
"created": "Mon, 12 Aug 2024 08:09:19 GMT",
"version": "v4"
}
] |
2024-08-13
|
[
[
"Blackshear",
"Sam",
""
],
[
"Chursin",
"Andrey",
""
],
[
"Danezis",
"George",
""
],
[
"Kichidis",
"Anastasios",
""
],
[
"Kokoris-Kogias",
"Lefteris",
""
],
[
"Li",
"Xun",
""
],
[
"Logan",
"Mark",
""
],
[
"Menon",
"Ashok",
""
],
[
"Nowacki",
"Todd",
""
],
[
"Sonnino",
"Alberto",
""
],
[
"Williams",
"Brandon",
""
],
[
"Zhang",
"Lu",
""
]
] |
Sui Lutris is the first smart-contract platform to sustainably achieve sub-second finality. It achieves this significant decrease by employing consensusless agreement not only for simple payments but for a large variety of transactions. Unlike prior work, Sui Lutris neither compromises expressiveness nor throughput and can run perpetually without restarts. Sui Lutris achieves this by safely integrating consensuless agreement with a high-throughput consensus protocol that is invoked out of the critical finality path but ensures that when a transaction is at risk of inconsistent concurrent accesses, its settlement is delayed until the total ordering is resolved. Building such a hybrid architecture is especially delicate during reconfiguration events, where the system needs to preserve the safety of the consensusless path without compromising the long-term liveness of potentially misconfigured clients. We thus develop a novel reconfiguration protocol, the first to provably show the safe and efficient reconfiguration of a consensusless blockchain. Sui Lutris is currently running in production and underpins the Sui smart-contract platform. Combined with the use of Objects instead of accounts it enables the safe execution of smart contracts that expose objects as a first-class resource. In our experiments Sui Lutris achieves latency lower than 0.5 seconds for throughput up to 5,000 certificates per second (150k ops/s with transaction blocks), compared to the state-of-the-art real-world consensus latencies of 3 seconds. Furthermore, it gracefully handles validators crash-recovery and does not suffer visible performance degradation during reconfiguration.
|
2211.00806
|
Hamid Hosseinianfar
|
Hamid Hosseinianfar, Hami Rabbani, and Maite Brandt-Pearce
|
Optical Channel Impulse Response-Based Localization Using An Artificial
Neural Network
| null | null | null | null |
cs.IT cs.LG eess.SP math.IT
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Visible light positioning has the potential to yield sub-centimeter accuracy
in indoor environments, yet conventional received signal strength (RSS)-based
localization algorithms cannot achieve this because their performance degrades
from optical multipath reflection. However, this part of the optical received
signal is deterministic due to the often static and predictable nature of the
optical wireless channel. In this paper, the performance of optical channel
impulse response (OCIR)-based localization is studied using an artificial
neural network (ANN) to map embedded features of the OCIR to the user
equipment's location. Numerical results show that OCIR-based localization
outperforms conventional RSS techniques by two orders of magnitude using only
two photodetectors as anchor points. The ANN technique can take advantage of
multipath features in a wide range of scenarios, from using only the DC value
to relying on high-resolution time sampling that can result in sub-centimeter
accuracy.
|
[
{
"created": "Wed, 2 Nov 2022 00:54:18 GMT",
"version": "v1"
},
{
"created": "Fri, 4 Nov 2022 18:59:10 GMT",
"version": "v2"
}
] |
2022-11-08
|
[
[
"Hosseinianfar",
"Hamid",
""
],
[
"Rabbani",
"Hami",
""
],
[
"Brandt-Pearce",
"Maite",
""
]
] |
Visible light positioning has the potential to yield sub-centimeter accuracy in indoor environments, yet conventional received signal strength (RSS)-based localization algorithms cannot achieve this because their performance degrades from optical multipath reflection. However, this part of the optical received signal is deterministic due to the often static and predictable nature of the optical wireless channel. In this paper, the performance of optical channel impulse response (OCIR)-based localization is studied using an artificial neural network (ANN) to map embedded features of the OCIR to the user equipment's location. Numerical results show that OCIR-based localization outperforms conventional RSS techniques by two orders of magnitude using only two photodetectors as anchor points. The ANN technique can take advantage of multipath features in a wide range of scenarios, from using only the DC value to relying on high-resolution time sampling that can result in sub-centimeter accuracy.
|
2303.15114
|
Aidana Massalimova
|
Aidana Massalimova, Maikel Timmermans, Nicola Cavalcanti, Daniel
Suter, Matthias Seibold, Fabio Carrillo, Christoph J. Laux, Reto Sutter,
Mazda Farshad, Kathleen Denis, Philipp F\"urnstahl
|
Automatic breach detection during spine pedicle drilling based on
vibroacoustic sensing
| null | null | null | null |
cs.CV
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Pedicle drilling is a complex and critical spinal surgery task. Detecting
breach or penetration of the surgical tool to the cortical wall during
pilot-hole drilling is essential to avoid damage to vital anatomical structures
adjacent to the pedicle, such as the spinal cord, blood vessels, and nerves.
Currently, the guidance of pedicle drilling is done using image-guided methods
that are radiation intensive and limited to the preoperative information. This
work proposes a new radiation-free breach detection algorithm leveraging a
non-visual sensor setup in combination with deep learning approach. Multiple
vibroacoustic sensors, such as a contact microphone, a free-field microphone, a
tri-axial accelerometer, a uni-axial accelerometer, and an optical tracking
system were integrated into the setup. Data were collected on four cadaveric
human spines, ranging from L5 to T10. An experienced spine surgeon drilled the
pedicles relying on optical navigation. A new automatic labeling method based
on the tracking data was introduced. Labeled data was subsequently fed to the
network in mel-spectrograms, classifying the data into breach and non-breach.
Different sensor types, sensor positioning, and their combinations were
evaluated. The best results in breach recall for individual sensors could be
achieved using contact microphones attached to the dorsal skin (85.8\%) and
uni-axial accelerometers clamped to the spinous process of the drilled vertebra
(81.0\%). The best-performing data fusion model combined the latter two sensors
with a breach recall of 98\%. The proposed method shows the great potential of
non-visual sensor fusion for avoiding screw misplacement and accidental bone
breaches during pedicle drilling and could be extended to further surgical
applications.
|
[
{
"created": "Mon, 27 Mar 2023 11:32:14 GMT",
"version": "v1"
}
] |
2023-03-28
|
[
[
"Massalimova",
"Aidana",
""
],
[
"Timmermans",
"Maikel",
""
],
[
"Cavalcanti",
"Nicola",
""
],
[
"Suter",
"Daniel",
""
],
[
"Seibold",
"Matthias",
""
],
[
"Carrillo",
"Fabio",
""
],
[
"Laux",
"Christoph J.",
""
],
[
"Sutter",
"Reto",
""
],
[
"Farshad",
"Mazda",
""
],
[
"Denis",
"Kathleen",
""
],
[
"Fürnstahl",
"Philipp",
""
]
] |
Pedicle drilling is a complex and critical spinal surgery task. Detecting breach or penetration of the surgical tool to the cortical wall during pilot-hole drilling is essential to avoid damage to vital anatomical structures adjacent to the pedicle, such as the spinal cord, blood vessels, and nerves. Currently, the guidance of pedicle drilling is done using image-guided methods that are radiation intensive and limited to the preoperative information. This work proposes a new radiation-free breach detection algorithm leveraging a non-visual sensor setup in combination with deep learning approach. Multiple vibroacoustic sensors, such as a contact microphone, a free-field microphone, a tri-axial accelerometer, a uni-axial accelerometer, and an optical tracking system were integrated into the setup. Data were collected on four cadaveric human spines, ranging from L5 to T10. An experienced spine surgeon drilled the pedicles relying on optical navigation. A new automatic labeling method based on the tracking data was introduced. Labeled data was subsequently fed to the network in mel-spectrograms, classifying the data into breach and non-breach. Different sensor types, sensor positioning, and their combinations were evaluated. The best results in breach recall for individual sensors could be achieved using contact microphones attached to the dorsal skin (85.8\%) and uni-axial accelerometers clamped to the spinous process of the drilled vertebra (81.0\%). The best-performing data fusion model combined the latter two sensors with a breach recall of 98\%. The proposed method shows the great potential of non-visual sensor fusion for avoiding screw misplacement and accidental bone breaches during pedicle drilling and could be extended to further surgical applications.
|
2302.08005
|
Hongzheng Chen
|
Hongzheng Chen, Cody Hao Yu, Shuai Zheng, Zhen Zhang, Zhiru Zhang,
Yida Wang
|
Slapo: A Schedule Language for Progressive Optimization of Large Deep
Learning Model Training
|
Accepted to ASPLOS'24
| null | null | null |
cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
Recent years have seen an increase in the development of large deep learning
(DL) models, which makes training efficiency crucial. Common practice is
struggling with the trade-off between usability and performance. On one hand,
DL frameworks such as PyTorch use dynamic graphs to facilitate model developers
at a price of sub-optimal model training performance. On the other hand,
practitioners propose various approaches to improving the training efficiency
by sacrificing some of the flexibility, ranging from making the graph static
for more thorough optimization (e.g., XLA) to customizing optimization towards
large-scale distributed training (e.g., DeepSpeed and Megatron-LM). In this
paper, we aim to address the tension between usability and training efficiency
through separation of concerns. Inspired by DL compilers that decouple the
platform-specific optimizations of a tensor-level operator from its arithmetic
definition, this paper proposes a schedule language, Slapo, to decouple model
execution from definition. Specifically, Slapo works on a PyTorch model and
uses a set of schedule primitives to convert the model for common model
training optimizations such as high-performance kernels, effective 3D
parallelism, and efficient activation checkpointing. Compared to existing
optimization solutions, Slapo progressively optimizes the model "as-needed"
through high-level primitives, and thus preserving programmability and
debuggability for users to a large extent. Our evaluation results show that by
scheduling the existing hand-crafted optimizations in a systematic way using
Slapo, we are able to improve training throughput by up to 2.92x on a single
machine with 8 NVIDIA V100 GPUs, and by up to 1.41x on multiple machines with
up to 64 GPUs, when compared to the out-of-the-box performance of DeepSpeed and
Megatron-LM.
|
[
{
"created": "Thu, 16 Feb 2023 00:34:53 GMT",
"version": "v1"
},
{
"created": "Sat, 23 Dec 2023 03:52:35 GMT",
"version": "v2"
}
] |
2023-12-27
|
[
[
"Chen",
"Hongzheng",
""
],
[
"Yu",
"Cody Hao",
""
],
[
"Zheng",
"Shuai",
""
],
[
"Zhang",
"Zhen",
""
],
[
"Zhang",
"Zhiru",
""
],
[
"Wang",
"Yida",
""
]
] |
Recent years have seen an increase in the development of large deep learning (DL) models, which makes training efficiency crucial. Common practice is struggling with the trade-off between usability and performance. On one hand, DL frameworks such as PyTorch use dynamic graphs to facilitate model developers at a price of sub-optimal model training performance. On the other hand, practitioners propose various approaches to improving the training efficiency by sacrificing some of the flexibility, ranging from making the graph static for more thorough optimization (e.g., XLA) to customizing optimization towards large-scale distributed training (e.g., DeepSpeed and Megatron-LM). In this paper, we aim to address the tension between usability and training efficiency through separation of concerns. Inspired by DL compilers that decouple the platform-specific optimizations of a tensor-level operator from its arithmetic definition, this paper proposes a schedule language, Slapo, to decouple model execution from definition. Specifically, Slapo works on a PyTorch model and uses a set of schedule primitives to convert the model for common model training optimizations such as high-performance kernels, effective 3D parallelism, and efficient activation checkpointing. Compared to existing optimization solutions, Slapo progressively optimizes the model "as-needed" through high-level primitives, and thus preserving programmability and debuggability for users to a large extent. Our evaluation results show that by scheduling the existing hand-crafted optimizations in a systematic way using Slapo, we are able to improve training throughput by up to 2.92x on a single machine with 8 NVIDIA V100 GPUs, and by up to 1.41x on multiple machines with up to 64 GPUs, when compared to the out-of-the-box performance of DeepSpeed and Megatron-LM.
|
1305.2006
|
Jierui Xie
|
Jierui Xie, Mingming Chen, Boleslaw K. Szymanski
|
LabelRankT: Incremental Community Detection in Dynamic Networks via
Label Propagation
|
DyNetMM 2013, New York, USA (conjunction with SIGMOD/PODS 2013)
|
Proc. DyNetMM 2013 at SIGMOD/PODS 2013, New York, NY, 2013
| null | null |
cs.SI physics.soc-ph
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
An increasingly important challenge in network analysis is efficient
detection and tracking of communities in dynamic networks for which changes
arrive as a stream. There is a need for algorithms that can incrementally
update and monitor communities whose evolution generates huge realtime data
streams, such as the Internet or on-line social networks. In this paper, we
propose LabelRankT, an online distributed algorithm for detection of
communities in large-scale dynamic networks through stabilized label
propagation. Results of tests on real-world networks demonstrate that
LabelRankT has much lower computational costs than other algorithms. It also
improves the quality of the detected communities compared to dynamic detection
methods and matches the quality achieved by static detection approaches. Unlike
most of other algorithms which apply only to binary networks, LabelRankT works
on weighted and directed networks, which provides a flexible and promising
solution for real-world applications.
|
[
{
"created": "Thu, 9 May 2013 04:01:46 GMT",
"version": "v1"
},
{
"created": "Sun, 12 May 2013 18:41:13 GMT",
"version": "v2"
}
] |
2013-05-15
|
[
[
"Xie",
"Jierui",
""
],
[
"Chen",
"Mingming",
""
],
[
"Szymanski",
"Boleslaw K.",
""
]
] |
An increasingly important challenge in network analysis is efficient detection and tracking of communities in dynamic networks for which changes arrive as a stream. There is a need for algorithms that can incrementally update and monitor communities whose evolution generates huge realtime data streams, such as the Internet or on-line social networks. In this paper, we propose LabelRankT, an online distributed algorithm for detection of communities in large-scale dynamic networks through stabilized label propagation. Results of tests on real-world networks demonstrate that LabelRankT has much lower computational costs than other algorithms. It also improves the quality of the detected communities compared to dynamic detection methods and matches the quality achieved by static detection approaches. Unlike most of other algorithms which apply only to binary networks, LabelRankT works on weighted and directed networks, which provides a flexible and promising solution for real-world applications.
|
1602.07285
|
Rohit Singh
|
Rohit Singh, Armando Solar-Lezama
|
Automatic Generation of Formula Simplifiers based on Conditional Rewrite
Rules
|
Submitted for peer reviewed conference
| null | null | null |
cs.PL
|
http://creativecommons.org/licenses/by/4.0/
|
This paper addresses the problem of creating simplifiers for logic formulas
based on conditional term rewriting. In particular, the paper focuses on a
program synthesis application where formula simplifications have been shown to
have a significant impact. We show that by combining machine learning
techniques with constraint-based synthesis, it is possible to synthesize a
formula simplifier fully automatically from a corpus of representative
problems, making it possible to create formula simplifiers tailored to specific
problem domains. We demonstrate the benefits of our approach for synthesis
benchmarks from the SyGuS competition and automated grading.
|
[
{
"created": "Tue, 23 Feb 2016 20:09:33 GMT",
"version": "v1"
}
] |
2016-02-24
|
[
[
"Singh",
"Rohit",
""
],
[
"Solar-Lezama",
"Armando",
""
]
] |
This paper addresses the problem of creating simplifiers for logic formulas based on conditional term rewriting. In particular, the paper focuses on a program synthesis application where formula simplifications have been shown to have a significant impact. We show that by combining machine learning techniques with constraint-based synthesis, it is possible to synthesize a formula simplifier fully automatically from a corpus of representative problems, making it possible to create formula simplifiers tailored to specific problem domains. We demonstrate the benefits of our approach for synthesis benchmarks from the SyGuS competition and automated grading.
|
2301.04205
|
Benjamin Mikek
|
Saksham Goel, Benjamin Mikek, Jehad Aly, Venkat Arun, Ahmed Saeed,
Aditya Akella
|
A Performance Verification Methodology for Resource Allocation
Heuristics
|
12 pages, 11 figures
| null | null | null |
cs.LO
|
http://creativecommons.org/licenses/by/4.0/
|
Performance verification is a nascent but promising tool for understanding
the performance and limitations of heuristics under realistic assumptions.
Bespoke performance verification tools have already demonstrated their value in
settings like congestion control and packet scheduling. In this paper, we aim
to emphasize the broad applicability and utility of performance verification.
To that end, we highlight the design principles of performance verification.
Then, we leverage that understanding to develop a set of easy-to-follow
guidelines that are applicable to a wide range of resource allocation
heuristics. In particular, we introduce Virelay, a framework that enables
heuristic designers to express the behavior of their algorithms and their
assumptions about the system in an environment that resembles a discrete-event
simulator. We demonstrate the utility and ease-of-use of Virelay by applying it
to six diverse case studies. We produce bounds on the performance of classical
algorithms, work stealing and SRPT scheduling, under practical assumptions. We
demonstrate Virelay's expressiveness by capturing existing models for
congestion control and packet scheduling, and we verify the observation that
TCP unfairness can cause some ML training workloads to spontaneously converge
to a state of high network utilization. Finally, we use Virelay to identify two
bugs in the Linux CFS load balancer.
|
[
{
"created": "Tue, 10 Jan 2023 20:46:20 GMT",
"version": "v1"
},
{
"created": "Wed, 28 Feb 2024 15:15:36 GMT",
"version": "v2"
}
] |
2024-02-29
|
[
[
"Goel",
"Saksham",
""
],
[
"Mikek",
"Benjamin",
""
],
[
"Aly",
"Jehad",
""
],
[
"Arun",
"Venkat",
""
],
[
"Saeed",
"Ahmed",
""
],
[
"Akella",
"Aditya",
""
]
] |
Performance verification is a nascent but promising tool for understanding the performance and limitations of heuristics under realistic assumptions. Bespoke performance verification tools have already demonstrated their value in settings like congestion control and packet scheduling. In this paper, we aim to emphasize the broad applicability and utility of performance verification. To that end, we highlight the design principles of performance verification. Then, we leverage that understanding to develop a set of easy-to-follow guidelines that are applicable to a wide range of resource allocation heuristics. In particular, we introduce Virelay, a framework that enables heuristic designers to express the behavior of their algorithms and their assumptions about the system in an environment that resembles a discrete-event simulator. We demonstrate the utility and ease-of-use of Virelay by applying it to six diverse case studies. We produce bounds on the performance of classical algorithms, work stealing and SRPT scheduling, under practical assumptions. We demonstrate Virelay's expressiveness by capturing existing models for congestion control and packet scheduling, and we verify the observation that TCP unfairness can cause some ML training workloads to spontaneously converge to a state of high network utilization. Finally, we use Virelay to identify two bugs in the Linux CFS load balancer.
|
1404.2772
|
Ravi Ranjan
|
Ravi Ranjan and G. Sahoo
|
A New Clustering Approach for Anomaly Intrusion Detection
|
10 pages with 3 figures,2 Tables This paper explains about clustering
methodology used in Data Mining field for Intrusion Detection in the area of
Network Security
|
International Journal of Data Mining & Knowledge Management
Process (IJDKP),ISSN:2230-9608[Online],2231-007X[Print] Vol.4, No.2, March
2014, page(s): 29-38
|
10.5121/ijdkp.2014.4203
| null |
cs.DC cs.CR cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Recent advances in technology have made our work easier compare to earlier
times. Computer network is growing day by day but while discussing about the
security of computers and networks it has always been a major concerns for
organizations varying from smaller to larger enterprises. It is true that
organizations are aware of the possible threats and attacks so they always
prepare for the safer side but due to some loopholes attackers are able to make
attacks. Intrusion detection is one of the major fields of research and
researchers are trying to find new algorithms for detecting intrusions.
Clustering techniques of data mining is an interested area of research for
detecting possible intrusions and attacks. This paper presents a new clustering
approach for anomaly intrusion detection by using the approach of K-medoids
method of clustering and its certain modifications. The proposed algorithm is
able to achieve high detection rate and overcomes the disadvantages of K-means
algorithm.
|
[
{
"created": "Thu, 10 Apr 2014 11:22:17 GMT",
"version": "v1"
}
] |
2014-04-11
|
[
[
"Ranjan",
"Ravi",
""
],
[
"Sahoo",
"G.",
""
]
] |
Recent advances in technology have made our work easier compare to earlier times. Computer network is growing day by day but while discussing about the security of computers and networks it has always been a major concerns for organizations varying from smaller to larger enterprises. It is true that organizations are aware of the possible threats and attacks so they always prepare for the safer side but due to some loopholes attackers are able to make attacks. Intrusion detection is one of the major fields of research and researchers are trying to find new algorithms for detecting intrusions. Clustering techniques of data mining is an interested area of research for detecting possible intrusions and attacks. This paper presents a new clustering approach for anomaly intrusion detection by using the approach of K-medoids method of clustering and its certain modifications. The proposed algorithm is able to achieve high detection rate and overcomes the disadvantages of K-means algorithm.
|
1305.4583
|
Xin Zhao
|
Xin Zhao
|
Parallel Coordinates Guided High Dimensional Transfer Function Design
|
6 pages, 5 figures. This paper has been withdrawn by the author due
to publication
| null | null | null |
cs.GR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
High-dimensional transfer function design is widely used to provide
appropriate data classification for direct volume rendering of various
datasets. However, its design is a complicated task. Parallel coordinate plot
(PCP), as a powerful visualization tool, can efficiently display
high-dimensional geometry and accurately analyze multivariate data. In this
paper, we propose to combine parallel coordinates with dimensional reduction
methods to guide high-dimensional transfer function design. Our pipeline has
two major advantages: (1) combine and display extracted high-dimensional
features in parameter space; and (2) select appropriate high-dimensional
parameters, with the help of dimensional reduction methods, to obtain
sophisticated data classification as transfer function for volume rendering. In
order to efficiently design high-dimensional transfer functions, the
combination of both parallel coordinate components and dimension reduction
results is necessary to generate final visualization results. We demonstrate
the capability of our method for direct volume rendering using various CT and
MRI datasets.
|
[
{
"created": "Mon, 20 May 2013 17:27:29 GMT",
"version": "v1"
},
{
"created": "Sun, 3 Nov 2013 21:39:13 GMT",
"version": "v2"
}
] |
2013-11-05
|
[
[
"Zhao",
"Xin",
""
]
] |
High-dimensional transfer function design is widely used to provide appropriate data classification for direct volume rendering of various datasets. However, its design is a complicated task. Parallel coordinate plot (PCP), as a powerful visualization tool, can efficiently display high-dimensional geometry and accurately analyze multivariate data. In this paper, we propose to combine parallel coordinates with dimensional reduction methods to guide high-dimensional transfer function design. Our pipeline has two major advantages: (1) combine and display extracted high-dimensional features in parameter space; and (2) select appropriate high-dimensional parameters, with the help of dimensional reduction methods, to obtain sophisticated data classification as transfer function for volume rendering. In order to efficiently design high-dimensional transfer functions, the combination of both parallel coordinate components and dimension reduction results is necessary to generate final visualization results. We demonstrate the capability of our method for direct volume rendering using various CT and MRI datasets.
|
1202.4626
|
Avraham N. Trahtman
|
A. N. Trahtman
|
The \v{C}erny conjecture
|
14 pages, 11 Lemmas, most of which are considered trivial by various
reviewers. Everything goes to that the main result is also trivial. And the
author himself is inclined to admit it
| null | null | null |
cs.DM
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
A word $w$ of letters on edges of underlying graph $\Gamma$ of deterministic
finite automaton (DFA) is called synchronizing if $w$ sends all states of the
automaton to a unique state. J. \v{C}erny discovered in 1964 a sequence of
$n$-state complete DFA possessing a minimal synchronizing word of length
$(n-1)^2$. The hypothesis, well known today as the \v{C}erny conjecture, claims
that it is also precise upper bound on the length of such a word for a complete
DFA. The hypothesis was formulated in 1966 by Starke. The problem has motivated
great and constantly growing number of investigations and generalizations. To
prove the conjecture, we use algebra w on a special class of row monomial
matrices (one unit and rest zeros in every row), induced by words in the
alphabet of labels on edges. These matrices generate a space with respect to
the mentioned operation. The proof is based on connection between length of
words $u$ and dimension of the space generated by solutions $L_x$ of matrix
equation $M_uL_x=M_s$ for synchronizing word $s$, as well as on the relation
between ranks of $M_u$ and $L_x$.
|
[
{
"created": "Tue, 21 Feb 2012 12:50:14 GMT",
"version": "v1"
},
{
"created": "Mon, 14 Jun 2021 15:24:13 GMT",
"version": "v10"
},
{
"created": "Tue, 18 Jan 2022 11:16:53 GMT",
"version": "v11"
},
{
"created": "Sat, 25 Feb 2012 09:42:30 GMT",
"version": "v2"
},
{
"created": "Wed, 29 Feb 2012 08:58:28 GMT",
"version": "v3"
},
{
"created": "Mon, 19 Aug 2013 18:54:12 GMT",
"version": "v4"
},
{
"created": "Thu, 29 Aug 2013 06:51:30 GMT",
"version": "v5"
},
{
"created": "Thu, 17 Oct 2013 07:22:11 GMT",
"version": "v6"
},
{
"created": "Thu, 20 Mar 2014 13:29:06 GMT",
"version": "v7"
},
{
"created": "Fri, 16 Sep 2016 14:55:56 GMT",
"version": "v8"
},
{
"created": "Tue, 4 Jul 2017 10:30:27 GMT",
"version": "v9"
}
] |
2022-01-19
|
[
[
"Trahtman",
"A. N.",
""
]
] |
A word $w$ of letters on edges of underlying graph $\Gamma$ of deterministic finite automaton (DFA) is called synchronizing if $w$ sends all states of the automaton to a unique state. J. \v{C}erny discovered in 1964 a sequence of $n$-state complete DFA possessing a minimal synchronizing word of length $(n-1)^2$. The hypothesis, well known today as the \v{C}erny conjecture, claims that it is also precise upper bound on the length of such a word for a complete DFA. The hypothesis was formulated in 1966 by Starke. The problem has motivated great and constantly growing number of investigations and generalizations. To prove the conjecture, we use algebra w on a special class of row monomial matrices (one unit and rest zeros in every row), induced by words in the alphabet of labels on edges. These matrices generate a space with respect to the mentioned operation. The proof is based on connection between length of words $u$ and dimension of the space generated by solutions $L_x$ of matrix equation $M_uL_x=M_s$ for synchronizing word $s$, as well as on the relation between ranks of $M_u$ and $L_x$.
|
2101.01677
|
Vitor Guizilini
|
Rares Ambrus, Vitor Guizilini, Naveen Kuppuswamy, Andrew Beaulieu,
Adrien Gaidon, Alex Alspach
|
Monocular Depth Estimation for Soft Visuotactile Sensors
| null | null | null | null |
cs.RO cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
Fluid-filled soft visuotactile sensors such as the Soft-bubbles alleviate key
challenges for robust manipulation, as they enable reliable grasps along with
the ability to obtain high-resolution sensory feedback on contact geometry and
forces. Although they are simple in construction, their utility has been
limited due to size constraints introduced by enclosed custom IR/depth imaging
sensors to directly measure surface deformations. Towards mitigating this
limitation, we investigate the application of state-of-the-art monocular depth
estimation to infer dense internal (tactile) depth maps directly from the
internal single small IR imaging sensor. Through real-world experiments, we
show that deep networks typically used for long-range depth estimation (1-100m)
can be effectively trained for precise predictions at a much shorter range
(1-100mm) inside a mostly textureless deformable fluid-filled sensor. We
propose a simple supervised learning process to train an object-agnostic
network requiring less than 10 random poses in contact for less than 10 seconds
for a small set of diverse objects (mug, wine glass, box, and fingers in our
experiments). We show that our approach is sample-efficient, accurate, and
generalizes across different objects and sensor configurations unseen at
training time. Finally, we discuss the implications of our approach for the
design of soft visuotactile sensors and grippers.
|
[
{
"created": "Tue, 5 Jan 2021 17:51:11 GMT",
"version": "v1"
}
] |
2021-01-06
|
[
[
"Ambrus",
"Rares",
""
],
[
"Guizilini",
"Vitor",
""
],
[
"Kuppuswamy",
"Naveen",
""
],
[
"Beaulieu",
"Andrew",
""
],
[
"Gaidon",
"Adrien",
""
],
[
"Alspach",
"Alex",
""
]
] |
Fluid-filled soft visuotactile sensors such as the Soft-bubbles alleviate key challenges for robust manipulation, as they enable reliable grasps along with the ability to obtain high-resolution sensory feedback on contact geometry and forces. Although they are simple in construction, their utility has been limited due to size constraints introduced by enclosed custom IR/depth imaging sensors to directly measure surface deformations. Towards mitigating this limitation, we investigate the application of state-of-the-art monocular depth estimation to infer dense internal (tactile) depth maps directly from the internal single small IR imaging sensor. Through real-world experiments, we show that deep networks typically used for long-range depth estimation (1-100m) can be effectively trained for precise predictions at a much shorter range (1-100mm) inside a mostly textureless deformable fluid-filled sensor. We propose a simple supervised learning process to train an object-agnostic network requiring less than 10 random poses in contact for less than 10 seconds for a small set of diverse objects (mug, wine glass, box, and fingers in our experiments). We show that our approach is sample-efficient, accurate, and generalizes across different objects and sensor configurations unseen at training time. Finally, we discuss the implications of our approach for the design of soft visuotactile sensors and grippers.
|
2308.16248
|
Danai Korre
|
Danai Korre and Andrew Sherlock
|
Augmented Reality in Higher Education: a Case Study in Medical Education
|
4 pages, 2 figures, 9th International Conference of the Immersive
Learning Research Network (iLRN2023)
| null | null | null |
cs.HC cs.ET
|
http://creativecommons.org/licenses/by-sa/4.0/
|
During lockdown, we piloted a variety of augmented reality (AR) experiences
in collaboration with subject matter experts from different fields aiming at
creating remote teaching and training experiences. In this paper, we present a
case study on how AR can be used as a teaching aid for medical education with
pertinent focus on remote and social distanced learning. We describe the
process of creating an AR experience that can enhance the knowledge and
understanding of anatomy for medical students. The Anatomy Experience is an AR
enhanced learning experience developed in collaboration with the Medical School
of the University of Edinburgh aiming to assist medical students understand the
complex geometry of different parts of the human body. After conducting a focus
group study with medical students, trainees, and trainers, we received very
positive feedback on the Anatomy Experience and its effects on understanding
anatomy, enriching the learning process, and using it as a tool for anatomy
teaching.
|
[
{
"created": "Wed, 30 Aug 2023 18:11:58 GMT",
"version": "v1"
}
] |
2023-09-14
|
[
[
"Korre",
"Danai",
""
],
[
"Sherlock",
"Andrew",
""
]
] |
During lockdown, we piloted a variety of augmented reality (AR) experiences in collaboration with subject matter experts from different fields aiming at creating remote teaching and training experiences. In this paper, we present a case study on how AR can be used as a teaching aid for medical education with pertinent focus on remote and social distanced learning. We describe the process of creating an AR experience that can enhance the knowledge and understanding of anatomy for medical students. The Anatomy Experience is an AR enhanced learning experience developed in collaboration with the Medical School of the University of Edinburgh aiming to assist medical students understand the complex geometry of different parts of the human body. After conducting a focus group study with medical students, trainees, and trainers, we received very positive feedback on the Anatomy Experience and its effects on understanding anatomy, enriching the learning process, and using it as a tool for anatomy teaching.
|
2004.04446
|
Yuqing Wang
|
Yuqing Wang, Zhaoliang Xu, Hao Shen, Baoshan Cheng, Lirong Yang
|
CenterMask: single shot instance segmentation with point representation
|
To appear at CVPR 2020
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this paper, we propose a single-shot instance segmentation method, which
is simple, fast and accurate. There are two main challenges for one-stage
instance segmentation: object instances differentiation and pixel-wise feature
alignment. Accordingly, we decompose the instance segmentation into two
parallel subtasks: Local Shape prediction that separates instances even in
overlapping conditions, and Global Saliency generation that segments the whole
image in a pixel-to-pixel manner. The outputs of the two branches are assembled
to form the final instance masks. To realize that, the local shape information
is adopted from the representation of object center points. Totally trained
from scratch and without any bells and whistles, the proposed CenterMask
achieves 34.5 mask AP with a speed of 12.3 fps, using a single-model with
single-scale training/testing on the challenging COCO dataset. The accuracy is
higher than all other one-stage instance segmentation methods except the 5
times slower TensorMask, which shows the effectiveness of CenterMask. Besides,
our method can be easily embedded to other one-stage object detectors such as
FCOS and performs well, showing the generalization of CenterMask.
|
[
{
"created": "Thu, 9 Apr 2020 09:35:15 GMT",
"version": "v1"
},
{
"created": "Sat, 11 Apr 2020 05:12:10 GMT",
"version": "v2"
}
] |
2020-04-14
|
[
[
"Wang",
"Yuqing",
""
],
[
"Xu",
"Zhaoliang",
""
],
[
"Shen",
"Hao",
""
],
[
"Cheng",
"Baoshan",
""
],
[
"Yang",
"Lirong",
""
]
] |
In this paper, we propose a single-shot instance segmentation method, which is simple, fast and accurate. There are two main challenges for one-stage instance segmentation: object instances differentiation and pixel-wise feature alignment. Accordingly, we decompose the instance segmentation into two parallel subtasks: Local Shape prediction that separates instances even in overlapping conditions, and Global Saliency generation that segments the whole image in a pixel-to-pixel manner. The outputs of the two branches are assembled to form the final instance masks. To realize that, the local shape information is adopted from the representation of object center points. Totally trained from scratch and without any bells and whistles, the proposed CenterMask achieves 34.5 mask AP with a speed of 12.3 fps, using a single-model with single-scale training/testing on the challenging COCO dataset. The accuracy is higher than all other one-stage instance segmentation methods except the 5 times slower TensorMask, which shows the effectiveness of CenterMask. Besides, our method can be easily embedded to other one-stage object detectors such as FCOS and performs well, showing the generalization of CenterMask.
|
1809.05353
|
Diego Rodriguez
|
Diego Rodriguez, Corbin Cogswell, Seongyong Koo, and Sven Behnke
|
Transferring Grasping Skills to Novel Instances by Latent Space
Non-Rigid Registration
|
In Proceedings of IEEE International Conference on Robotics and
Automation (ICRA), Brisbane, Australia, May 2018
| null | null | null |
cs.RO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Robots acting in open environments need to be able to handle novel objects.
Based on the observation that objects within a category are often similar in
their shapes and usage, we propose an approach for transferring grasping skills
from known instances to novel instances of an object category. Correspondences
between the instances are established by means of a non-rigid registration
method that combines the Coherent Point Drift approach with subspace methods.
The known object instances are modeled using a canonical shape and a
transformation which deforms it to match the instance shape. The principle axes
of variation of these deformations define a low-dimensional latent space. New
instances can be generated through interpolation and extrapolation in this
shape space. For inferring the shape parameters of an unknown instance, an
energy function expressed in terms of the latent variables is minimized. Due to
the class-level knowledge of the object, our method is able to complete novel
shapes from partial views. Control poses for generating grasping motions are
transferred efficiently to novel instances by the estimated non-rigid
transformation.
|
[
{
"created": "Fri, 14 Sep 2018 11:06:58 GMT",
"version": "v1"
}
] |
2018-09-17
|
[
[
"Rodriguez",
"Diego",
""
],
[
"Cogswell",
"Corbin",
""
],
[
"Koo",
"Seongyong",
""
],
[
"Behnke",
"Sven",
""
]
] |
Robots acting in open environments need to be able to handle novel objects. Based on the observation that objects within a category are often similar in their shapes and usage, we propose an approach for transferring grasping skills from known instances to novel instances of an object category. Correspondences between the instances are established by means of a non-rigid registration method that combines the Coherent Point Drift approach with subspace methods. The known object instances are modeled using a canonical shape and a transformation which deforms it to match the instance shape. The principle axes of variation of these deformations define a low-dimensional latent space. New instances can be generated through interpolation and extrapolation in this shape space. For inferring the shape parameters of an unknown instance, an energy function expressed in terms of the latent variables is minimized. Due to the class-level knowledge of the object, our method is able to complete novel shapes from partial views. Control poses for generating grasping motions are transferred efficiently to novel instances by the estimated non-rigid transformation.
|
1403.3905
|
Michael Hemmer
|
Francisc Bungiu and Michael Hemmer and John Hershberger and Kan Huang
and Alexander Kr\"oller
|
Efficient Computation of Visibility Polygons
| null | null | null | null |
cs.CG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Determining visibility in planar polygons and arrangements is an important
subroutine for many algorithms in computational geometry. In this paper, we
report on new implementations, and corresponding experimental evaluations, for
two established and one novel algorithm for computing visibility polygons.
These algorithms will be released to the public shortly, as a new package for
the Computational Geometry Algorithms Library (CGAL).
|
[
{
"created": "Sun, 16 Mar 2014 11:07:49 GMT",
"version": "v1"
}
] |
2014-03-18
|
[
[
"Bungiu",
"Francisc",
""
],
[
"Hemmer",
"Michael",
""
],
[
"Hershberger",
"John",
""
],
[
"Huang",
"Kan",
""
],
[
"Kröller",
"Alexander",
""
]
] |
Determining visibility in planar polygons and arrangements is an important subroutine for many algorithms in computational geometry. In this paper, we report on new implementations, and corresponding experimental evaluations, for two established and one novel algorithm for computing visibility polygons. These algorithms will be released to the public shortly, as a new package for the Computational Geometry Algorithms Library (CGAL).
|
1606.07383
|
Soheil Feizi
|
Soheil Feizi, Muriel Medard, Gerald Quon, Manolis Kellis and Ken Duffy
|
Network Infusion to Infer Information Sources in Networks
|
21 pages, 13 figures
| null | null | null |
cs.SI physics.soc-ph
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Several significant models have been developed that enable the study of
diffusion of signals across biological, social and engineered networks. Within
these established frameworks, the inverse problem of identifying the source of
the propagated signal is challenging, owing to the numerous alternative
possibilities for signal progression through the network. In real world
networks, the challenge of determining sources is compounded as the true
propagation dynamics are typically unknown, and when they have been directly
measured, they rarely conform to the assumptions of any of the well-studied
models. In this paper we introduce a method called Network Infusion (NI) that
has been designed to circumvent these issues, making source inference practical
for large, complex real world networks. The key idea is that to infer the
source node in the network, full characterization of diffusion dynamics, in
many cases, may not be necessary. This objective is achieved by creating a
diffusion kernel that well-approximates standard diffusion models, but lends
itself to inversion, by design, via likelihood maximization or error
minimization. We apply NI for both single-source and multi-source diffusion,
for both single-snapshot and multi-snapshot observations, and for both
homogeneous and heterogeneous diffusion setups. We prove the mean-field
optimality of NI for different scenarios, and demonstrate its effectiveness
over several synthetic networks. Moreover, we apply NI to a real-data
application, identifying news sources in the Digg social network, and
demonstrate the effectiveness of NI compared to existing methods. Finally, we
propose an integrative source inference framework that combines NI with a
distance centrality-based method, which leads to a robust performance in cases
where the underlying dynamics are unknown.
|
[
{
"created": "Thu, 23 Jun 2016 17:45:23 GMT",
"version": "v1"
}
] |
2016-06-24
|
[
[
"Feizi",
"Soheil",
""
],
[
"Medard",
"Muriel",
""
],
[
"Quon",
"Gerald",
""
],
[
"Kellis",
"Manolis",
""
],
[
"Duffy",
"Ken",
""
]
] |
Several significant models have been developed that enable the study of diffusion of signals across biological, social and engineered networks. Within these established frameworks, the inverse problem of identifying the source of the propagated signal is challenging, owing to the numerous alternative possibilities for signal progression through the network. In real world networks, the challenge of determining sources is compounded as the true propagation dynamics are typically unknown, and when they have been directly measured, they rarely conform to the assumptions of any of the well-studied models. In this paper we introduce a method called Network Infusion (NI) that has been designed to circumvent these issues, making source inference practical for large, complex real world networks. The key idea is that to infer the source node in the network, full characterization of diffusion dynamics, in many cases, may not be necessary. This objective is achieved by creating a diffusion kernel that well-approximates standard diffusion models, but lends itself to inversion, by design, via likelihood maximization or error minimization. We apply NI for both single-source and multi-source diffusion, for both single-snapshot and multi-snapshot observations, and for both homogeneous and heterogeneous diffusion setups. We prove the mean-field optimality of NI for different scenarios, and demonstrate its effectiveness over several synthetic networks. Moreover, we apply NI to a real-data application, identifying news sources in the Digg social network, and demonstrate the effectiveness of NI compared to existing methods. Finally, we propose an integrative source inference framework that combines NI with a distance centrality-based method, which leads to a robust performance in cases where the underlying dynamics are unknown.
|
2211.08273
|
Adolf Kamuzora Mr
|
Adolf Kamuzora, Wadie Skaf, Ermiyas Birihanu, Jiyan Mahmud, P\'eter
Kiss, Tam\'as Jursonovics, Peter Pogrzeba, Imre Lend\'ak and Tom\'a\v{s}
Horv\'ath
|
Matrix Factorization for Cache Optimization in Content Delivery Networks
(CDN)
| null |
22nd Industrial Conference on Data Mining 2022, New York, USA
Proceedings P. 1-10
| null | null |
cs.NI
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Content delivery networks (CDNs) are key components of high throughput, low
latency services on the internet. CDN cache servers have limited storage and
bandwidth and implement state-of-the-art cache admission and eviction
algorithms to select the most popular and relevant content for the customers
served. The aim of this study was to utilize state-of-the-art recommender
system techniques for predicting ratings for cache content in CDN. Matrix
factorization was used in predicting content popularity which is valuable
information in content eviction and content admission algorithms run on CDN
edge servers. A custom implemented matrix factorization class and MyMediaLite
were utilized. The input CDN logs were received from a European
telecommunication service provider. We built a matrix factorization model with
that data and utilized grid search to tune its hyper-parameters. Experimental
results indicate that there is promise about the proposed approaches and we
showed that a low root mean square error value can be achieved on the real-life
CDN log data.
|
[
{
"created": "Wed, 5 Oct 2022 11:06:32 GMT",
"version": "v1"
}
] |
2022-11-16
|
[
[
"Kamuzora",
"Adolf",
""
],
[
"Skaf",
"Wadie",
""
],
[
"Birihanu",
"Ermiyas",
""
],
[
"Mahmud",
"Jiyan",
""
],
[
"Kiss",
"Péter",
""
],
[
"Jursonovics",
"Tamás",
""
],
[
"Pogrzeba",
"Peter",
""
],
[
"Lendák",
"Imre",
""
],
[
"Horváth",
"Tomáš",
""
]
] |
Content delivery networks (CDNs) are key components of high throughput, low latency services on the internet. CDN cache servers have limited storage and bandwidth and implement state-of-the-art cache admission and eviction algorithms to select the most popular and relevant content for the customers served. The aim of this study was to utilize state-of-the-art recommender system techniques for predicting ratings for cache content in CDN. Matrix factorization was used in predicting content popularity which is valuable information in content eviction and content admission algorithms run on CDN edge servers. A custom implemented matrix factorization class and MyMediaLite were utilized. The input CDN logs were received from a European telecommunication service provider. We built a matrix factorization model with that data and utilized grid search to tune its hyper-parameters. Experimental results indicate that there is promise about the proposed approaches and we showed that a low root mean square error value can be achieved on the real-life CDN log data.
|
2311.11901
|
Lei Fan
|
Lei Fan, Yiwen Ding, Dongdong Fan, Yong Wu, Maurice Pagnucco and Yang
Song
|
Identifying the Defective: Detecting Damaged Grains for Cereal
Appearance Inspection
|
Accepted by ECAI2023. https://github.com/hellodfan/AI4GrainInsp
| null |
10.3233/FAIA230329
| null |
cs.CV
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
Cereal grain plays a crucial role in the human diet as a major source of
essential nutrients. Grain Appearance Inspection (GAI) serves as an essential
process to determine grain quality and facilitate grain circulation and
processing. However, GAI is routinely performed manually by inspectors with
cumbersome procedures, which poses a significant bottleneck in smart
agriculture.
In this paper, we endeavor to develop an automated GAI system:AI4GrainInsp.
By analyzing the distinctive characteristics of grain kernels, we formulate GAI
as a ubiquitous problem: Anomaly Detection (AD), in which healthy and edible
kernels are considered normal samples while damaged grains or unknown objects
are regarded as anomalies. We further propose an AD model, called AD-GAI, which
is trained using only normal samples yet can identify anomalies during
inference. Moreover, we customize a prototype device for data acquisition and
create a large-scale dataset including 220K high-quality images of wheat and
maize kernels. Through extensive experiments, AD-GAI achieves considerable
performance in comparison with advanced AD methods, and AI4GrainInsp has highly
consistent performance compared to human experts and excels at inspection
efficiency over 20x speedup. The dataset, code and models will be released at
https://github.com/hellodfan/AI4GrainInsp.
|
[
{
"created": "Mon, 20 Nov 2023 16:35:16 GMT",
"version": "v1"
}
] |
2023-11-21
|
[
[
"Fan",
"Lei",
""
],
[
"Ding",
"Yiwen",
""
],
[
"Fan",
"Dongdong",
""
],
[
"Wu",
"Yong",
""
],
[
"Pagnucco",
"Maurice",
""
],
[
"Song",
"Yang",
""
]
] |
Cereal grain plays a crucial role in the human diet as a major source of essential nutrients. Grain Appearance Inspection (GAI) serves as an essential process to determine grain quality and facilitate grain circulation and processing. However, GAI is routinely performed manually by inspectors with cumbersome procedures, which poses a significant bottleneck in smart agriculture. In this paper, we endeavor to develop an automated GAI system:AI4GrainInsp. By analyzing the distinctive characteristics of grain kernels, we formulate GAI as a ubiquitous problem: Anomaly Detection (AD), in which healthy and edible kernels are considered normal samples while damaged grains or unknown objects are regarded as anomalies. We further propose an AD model, called AD-GAI, which is trained using only normal samples yet can identify anomalies during inference. Moreover, we customize a prototype device for data acquisition and create a large-scale dataset including 220K high-quality images of wheat and maize kernels. Through extensive experiments, AD-GAI achieves considerable performance in comparison with advanced AD methods, and AI4GrainInsp has highly consistent performance compared to human experts and excels at inspection efficiency over 20x speedup. The dataset, code and models will be released at https://github.com/hellodfan/AI4GrainInsp.
|
2305.03977
|
Da Ren
|
Da Ren, Yi Cai, Qing Li
|
An Adversarial Non-Autoregressive Model for Text Generation with
Incomplete Information
| null | null | null | null |
cs.CL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Non-autoregressive models have been widely studied in the Complete
Information Scenario (CIS), in which the input has complete information of
corresponding output. However, their explorations in the Incomplete Information
Scenario (IIS) are extremely limited. Our analyses reveal that the IIS's
incomplete input information will augment the inherent limitations of existing
non-autoregressive models trained under Maximum Likelihood Estimation. In this
paper, we propose for the IIS an Adversarial Non-autoregressive Transformer
(ANT) which has two features: 1) Position-Aware Self-Modulation to provide more
reasonable hidden representations, and 2) Dependency Feed Forward Network to
strengthen its capacity in dependency modeling. We compare ANT with other
mainstream models in the IIS and demonstrate that ANT can achieve comparable
performance with much fewer decoding iterations. Furthermore, we show its great
potential in various applications like latent interpolation and semi-supervised
learning.
|
[
{
"created": "Sat, 6 May 2023 08:43:33 GMT",
"version": "v1"
},
{
"created": "Fri, 1 Dec 2023 15:16:19 GMT",
"version": "v2"
}
] |
2023-12-04
|
[
[
"Ren",
"Da",
""
],
[
"Cai",
"Yi",
""
],
[
"Li",
"Qing",
""
]
] |
Non-autoregressive models have been widely studied in the Complete Information Scenario (CIS), in which the input has complete information of corresponding output. However, their explorations in the Incomplete Information Scenario (IIS) are extremely limited. Our analyses reveal that the IIS's incomplete input information will augment the inherent limitations of existing non-autoregressive models trained under Maximum Likelihood Estimation. In this paper, we propose for the IIS an Adversarial Non-autoregressive Transformer (ANT) which has two features: 1) Position-Aware Self-Modulation to provide more reasonable hidden representations, and 2) Dependency Feed Forward Network to strengthen its capacity in dependency modeling. We compare ANT with other mainstream models in the IIS and demonstrate that ANT can achieve comparable performance with much fewer decoding iterations. Furthermore, we show its great potential in various applications like latent interpolation and semi-supervised learning.
|
1812.10280
|
Torgeir Dings{\o}yr
|
Torgeir Dings{\o}yr, Nils Brede Moe, Helena Holmstrom Ohlsson
|
Towards an Understanding of Scaling Frameworks and Business Agility: A
Summary of the 6th International Workshop at XP2018
|
Summary of workshop at XP2018
| null | null | null |
cs.SE
|
http://creativecommons.org/licenses/by/4.0/
|
Large development projects and programs are conducted using agile development
methods, with an increasing body of advice from practitioners and from
research. This sixth workshop showed in increasing interest in scaling
frameworks and in topics related to achieving business agility. This article
summarizes four contributed papers, discussions in "open space" format and also
presents a revised research agenda for large-scale agile development.
|
[
{
"created": "Wed, 26 Dec 2018 10:45:08 GMT",
"version": "v1"
}
] |
2018-12-27
|
[
[
"Dingsøyr",
"Torgeir",
""
],
[
"Moe",
"Nils Brede",
""
],
[
"Ohlsson",
"Helena Holmstrom",
""
]
] |
Large development projects and programs are conducted using agile development methods, with an increasing body of advice from practitioners and from research. This sixth workshop showed in increasing interest in scaling frameworks and in topics related to achieving business agility. This article summarizes four contributed papers, discussions in "open space" format and also presents a revised research agenda for large-scale agile development.
|
2304.01201
|
Ruihan Yang
|
Ruihan Yang, Ge Yang, Xiaolong Wang
|
Neural Volumetric Memory for Visual Locomotion Control
|
CVPR 2023 Highlight. Our project page with videos is
https://rchalyang.github.io/NVM
| null | null | null |
cs.RO cs.AI cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
Legged robots have the potential to expand the reach of autonomy beyond paved
roads. In this work, we consider the difficult problem of locomotion on
challenging terrains using a single forward-facing depth camera. Due to the
partial observability of the problem, the robot has to rely on past
observations to infer the terrain currently beneath it. To solve this problem,
we follow the paradigm in computer vision that explicitly models the 3D
geometry of the scene and propose Neural Volumetric Memory (NVM), a geometric
memory architecture that explicitly accounts for the SE(3) equivariance of the
3D world. NVM aggregates feature volumes from multiple camera views by first
bringing them back to the ego-centric frame of the robot. We test the learned
visual-locomotion policy on a physical robot and show that our approach, which
explicitly introduces geometric priors during training, offers superior
performance than more na\"ive methods. We also include ablation studies and
show that the representations stored in the neural volumetric memory capture
sufficient geometric information to reconstruct the scene. Our project page
with videos is https://rchalyang.github.io/NVM .
|
[
{
"created": "Mon, 3 Apr 2023 17:59:56 GMT",
"version": "v1"
}
] |
2023-04-04
|
[
[
"Yang",
"Ruihan",
""
],
[
"Yang",
"Ge",
""
],
[
"Wang",
"Xiaolong",
""
]
] |
Legged robots have the potential to expand the reach of autonomy beyond paved roads. In this work, we consider the difficult problem of locomotion on challenging terrains using a single forward-facing depth camera. Due to the partial observability of the problem, the robot has to rely on past observations to infer the terrain currently beneath it. To solve this problem, we follow the paradigm in computer vision that explicitly models the 3D geometry of the scene and propose Neural Volumetric Memory (NVM), a geometric memory architecture that explicitly accounts for the SE(3) equivariance of the 3D world. NVM aggregates feature volumes from multiple camera views by first bringing them back to the ego-centric frame of the robot. We test the learned visual-locomotion policy on a physical robot and show that our approach, which explicitly introduces geometric priors during training, offers superior performance than more na\"ive methods. We also include ablation studies and show that the representations stored in the neural volumetric memory capture sufficient geometric information to reconstruct the scene. Our project page with videos is https://rchalyang.github.io/NVM .
|
2209.07660
|
Joshua Ott
|
Joshua Ott, Edward Balaban, Mykel J. Kochenderfer
|
Sequential Bayesian Optimization for Adaptive Informative Path Planning
with Multimodal Sensing
| null | null | null | null |
cs.AI cs.RO
|
http://creativecommons.org/licenses/by/4.0/
|
Adaptive Informative Path Planning with Multimodal Sensing (AIPPMS) considers
the problem of an agent equipped with multiple sensors, each with different
sensing accuracy and energy costs. The agent's goal is to explore the
environment and gather information subject to its resource constraints in
unknown, partially observable environments. Previous work has focused on the
less general Adaptive Informative Path Planning (AIPP) problem, which considers
only the effect of the agent's movement on received observations. The AIPPMS
problem adds additional complexity by requiring that the agent reasons jointly
about the effects of sensing and movement while balancing resource constraints
with information objectives. We formulate the AIPPMS problem as a belief Markov
decision process with Gaussian process beliefs and solve it using a sequential
Bayesian optimization approach with online planning. Our approach consistently
outperforms previous AIPPMS solutions by more than doubling the average reward
received in almost every experiment while also reducing the root-mean-square
error in the environment belief by 50%. We completely open-source our
implementation to aid in further development and comparison.
|
[
{
"created": "Fri, 16 Sep 2022 00:50:36 GMT",
"version": "v1"
}
] |
2022-09-19
|
[
[
"Ott",
"Joshua",
""
],
[
"Balaban",
"Edward",
""
],
[
"Kochenderfer",
"Mykel J.",
""
]
] |
Adaptive Informative Path Planning with Multimodal Sensing (AIPPMS) considers the problem of an agent equipped with multiple sensors, each with different sensing accuracy and energy costs. The agent's goal is to explore the environment and gather information subject to its resource constraints in unknown, partially observable environments. Previous work has focused on the less general Adaptive Informative Path Planning (AIPP) problem, which considers only the effect of the agent's movement on received observations. The AIPPMS problem adds additional complexity by requiring that the agent reasons jointly about the effects of sensing and movement while balancing resource constraints with information objectives. We formulate the AIPPMS problem as a belief Markov decision process with Gaussian process beliefs and solve it using a sequential Bayesian optimization approach with online planning. Our approach consistently outperforms previous AIPPMS solutions by more than doubling the average reward received in almost every experiment while also reducing the root-mean-square error in the environment belief by 50%. We completely open-source our implementation to aid in further development and comparison.
|
2009.06218
|
Jing Ji
|
Fanglan Zheng, Erihe, Kun Li, Jiang Tian, Xiaojia Xiang
|
A Vertical Federated Learning Method for Interpretable Scorecard and Its
Application in Credit Scoring
| null | null | null | null |
cs.LG stat.ML
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
With the success of big data and artificial intelligence in many fields, the
applications of big data driven models are expected in financial risk
management especially credit scoring and rating. Under the premise of data
privacy protection, we propose a projected gradient-based method in the
vertical federated learning framework for the traditional scorecard, which is
based on logistic regression with bounded constraints, namely FL-LRBC. The
latter enables multiple agencies to jointly train an optimized scorecard model
in a single training session. It leads to the formation of the model with
positive coefficients, while the time-consuming parameter-tuning process can be
avoided. Moreover, the performance in terms of both AUC and the
Kolmogorov-Smirnov (KS) statistics is significantly improved due to data
enrichment using FL-LRBC. At present, FL-LRBC has already been applied to
credit business in a China nation-wide financial holdings group.
|
[
{
"created": "Mon, 14 Sep 2020 06:26:09 GMT",
"version": "v1"
}
] |
2020-09-15
|
[
[
"Zheng",
"Fanglan",
""
],
[
"Erihe",
"",
""
],
[
"Li",
"Kun",
""
],
[
"Tian",
"Jiang",
""
],
[
"Xiang",
"Xiaojia",
""
]
] |
With the success of big data and artificial intelligence in many fields, the applications of big data driven models are expected in financial risk management especially credit scoring and rating. Under the premise of data privacy protection, we propose a projected gradient-based method in the vertical federated learning framework for the traditional scorecard, which is based on logistic regression with bounded constraints, namely FL-LRBC. The latter enables multiple agencies to jointly train an optimized scorecard model in a single training session. It leads to the formation of the model with positive coefficients, while the time-consuming parameter-tuning process can be avoided. Moreover, the performance in terms of both AUC and the Kolmogorov-Smirnov (KS) statistics is significantly improved due to data enrichment using FL-LRBC. At present, FL-LRBC has already been applied to credit business in a China nation-wide financial holdings group.
|
2012.00465
|
Daniel Barath
|
Yaqing Ding, Daniel Barath, Zuzana Kukelova
|
Minimal Solutions for Panoramic Stitching Given Gravity Prior
| null | null | null | null |
cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
When capturing panoramas, people tend to align their cameras with the
vertical axis, i.e., the direction of gravity. Moreover, modern devices, such
as smartphones and tablets, are equipped with an IMU (Inertial Measurement
Unit) that can measure the gravity vector accurately. Using this prior, the
y-axes of the cameras can be aligned or assumed to be already aligned, reducing
their relative orientation to 1-DOF (degree of freedom). Exploiting this
assumption, we propose new minimal solutions to panoramic image stitching of
images taken by cameras with coinciding optical centers, i.e., undergoing pure
rotation. We consider four practical camera configurations, assuming unknown
fixed or varying focal length with or without radial distortion. The solvers
are tested both on synthetic scenes and on more than 500k real image pairs from
the Sun360 dataset and from scenes captured by us using two smartphones
equipped with IMUs. It is shown, that they outperform the state-of-the-art both
in terms of accuracy and processing time.
|
[
{
"created": "Tue, 1 Dec 2020 13:17:36 GMT",
"version": "v1"
}
] |
2020-12-02
|
[
[
"Ding",
"Yaqing",
""
],
[
"Barath",
"Daniel",
""
],
[
"Kukelova",
"Zuzana",
""
]
] |
When capturing panoramas, people tend to align their cameras with the vertical axis, i.e., the direction of gravity. Moreover, modern devices, such as smartphones and tablets, are equipped with an IMU (Inertial Measurement Unit) that can measure the gravity vector accurately. Using this prior, the y-axes of the cameras can be aligned or assumed to be already aligned, reducing their relative orientation to 1-DOF (degree of freedom). Exploiting this assumption, we propose new minimal solutions to panoramic image stitching of images taken by cameras with coinciding optical centers, i.e., undergoing pure rotation. We consider four practical camera configurations, assuming unknown fixed or varying focal length with or without radial distortion. The solvers are tested both on synthetic scenes and on more than 500k real image pairs from the Sun360 dataset and from scenes captured by us using two smartphones equipped with IMUs. It is shown, that they outperform the state-of-the-art both in terms of accuracy and processing time.
|
2207.12757
|
Chun-Mao Lai
|
Chun-Mao Lai, Ming-Hao Hsu, Chao-Wei Huang, Yun-Nung Chen
|
Controllable User Dialogue Act Augmentation for Dialogue State Tracking
|
9 pages, 4 figures, accepted to sigdial 2022
| null | null | null |
cs.CL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Prior work has demonstrated that data augmentation is useful for improving
dialogue state tracking. However, there are many types of user utterances,
while the prior method only considered the simplest one for augmentation,
raising the concern about poor generalization capability. In order to better
cover diverse dialogue acts and control the generation quality, this paper
proposes controllable user dialogue act augmentation (CUDA-DST) to augment user
utterances with diverse behaviors. With the augmented data, different state
trackers gain improvement and show better robustness, achieving the
state-of-the-art performance on MultiWOZ 2.1
|
[
{
"created": "Tue, 26 Jul 2022 09:04:48 GMT",
"version": "v1"
}
] |
2022-07-27
|
[
[
"Lai",
"Chun-Mao",
""
],
[
"Hsu",
"Ming-Hao",
""
],
[
"Huang",
"Chao-Wei",
""
],
[
"Chen",
"Yun-Nung",
""
]
] |
Prior work has demonstrated that data augmentation is useful for improving dialogue state tracking. However, there are many types of user utterances, while the prior method only considered the simplest one for augmentation, raising the concern about poor generalization capability. In order to better cover diverse dialogue acts and control the generation quality, this paper proposes controllable user dialogue act augmentation (CUDA-DST) to augment user utterances with diverse behaviors. With the augmented data, different state trackers gain improvement and show better robustness, achieving the state-of-the-art performance on MultiWOZ 2.1
|
1911.09576
|
Gordon MacDonald
|
Gordon MacDonald and Andrew Godbout and Bryn Gillcash and Stephanie
Cairns
|
Volume-preserving Neural Networks
|
20 pages, 8 figures
| null | null | null |
cs.LG stat.ML
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We propose a novel approach to addressing the vanishing (or exploding)
gradient problem in deep neural networks. We construct a new architecture for
deep neural networks where all layers (except the output layer) of the network
are a combination of rotation, permutation, diagonal, and activation sublayers
which are all volume preserving. Our approach replaces the standard weight
matrix of a neural network with a combination of diagonal, rotational and
permutation matrices, all of which are volume-preserving. We introduce a
coupled activation function allowing us to preserve volume even in the
activation function portion of a neural network layer. This control on the
volume forces the gradient (on average) to maintain equilibrium and not explode
or vanish. To demonstrate our architecture we apply our volume-preserving
neural network model to two standard datasets.
|
[
{
"created": "Thu, 21 Nov 2019 16:10:41 GMT",
"version": "v1"
},
{
"created": "Fri, 22 Nov 2019 17:29:50 GMT",
"version": "v2"
},
{
"created": "Mon, 26 Apr 2021 16:05:32 GMT",
"version": "v3"
}
] |
2021-04-27
|
[
[
"MacDonald",
"Gordon",
""
],
[
"Godbout",
"Andrew",
""
],
[
"Gillcash",
"Bryn",
""
],
[
"Cairns",
"Stephanie",
""
]
] |
We propose a novel approach to addressing the vanishing (or exploding) gradient problem in deep neural networks. We construct a new architecture for deep neural networks where all layers (except the output layer) of the network are a combination of rotation, permutation, diagonal, and activation sublayers which are all volume preserving. Our approach replaces the standard weight matrix of a neural network with a combination of diagonal, rotational and permutation matrices, all of which are volume-preserving. We introduce a coupled activation function allowing us to preserve volume even in the activation function portion of a neural network layer. This control on the volume forces the gradient (on average) to maintain equilibrium and not explode or vanish. To demonstrate our architecture we apply our volume-preserving neural network model to two standard datasets.
|
2403.06465
|
Jianxun Lian
|
Jianxun Lian, Yuxuan Lei, Xu Huang, Jing Yao, Wei Xu, Xing Xie
|
RecAI: Leveraging Large Language Models for Next-Generation Recommender
Systems
|
4 pages. Webconf 2024 demo track
| null |
10.1145/3589335.3651242
| null |
cs.IR cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
This paper introduces RecAI, a practical toolkit designed to augment or even
revolutionize recommender systems with the advanced capabilities of Large
Language Models (LLMs). RecAI provides a suite of tools, including Recommender
AI Agent, Recommendation-oriented Language Models, Knowledge Plugin,
RecExplainer, and Evaluator, to facilitate the integration of LLMs into
recommender systems from multifaceted perspectives. The new generation of
recommender systems, empowered by LLMs, are expected to be more versatile,
explainable, conversational, and controllable, paving the way for more
intelligent and user-centric recommendation experiences. We hope the
open-source of RecAI can help accelerate evolution of new advanced recommender
systems. The source code of RecAI is available at
\url{https://github.com/microsoft/RecAI}.
|
[
{
"created": "Mon, 11 Mar 2024 07:07:02 GMT",
"version": "v1"
}
] |
2024-03-12
|
[
[
"Lian",
"Jianxun",
""
],
[
"Lei",
"Yuxuan",
""
],
[
"Huang",
"Xu",
""
],
[
"Yao",
"Jing",
""
],
[
"Xu",
"Wei",
""
],
[
"Xie",
"Xing",
""
]
] |
This paper introduces RecAI, a practical toolkit designed to augment or even revolutionize recommender systems with the advanced capabilities of Large Language Models (LLMs). RecAI provides a suite of tools, including Recommender AI Agent, Recommendation-oriented Language Models, Knowledge Plugin, RecExplainer, and Evaluator, to facilitate the integration of LLMs into recommender systems from multifaceted perspectives. The new generation of recommender systems, empowered by LLMs, are expected to be more versatile, explainable, conversational, and controllable, paving the way for more intelligent and user-centric recommendation experiences. We hope the open-source of RecAI can help accelerate evolution of new advanced recommender systems. The source code of RecAI is available at \url{https://github.com/microsoft/RecAI}.
|
1610.04551
|
Rafael Hurtado
|
Jorge Useche and Rafael Hurtado
|
Tonal consonance parameters link microscopic and macroscopic properties
of music exposing a hidden order in melody
|
11 pages, 7 figures. Supplemental material contains 3 figures and 3
tables. An spreadsheet .xlsx contains data, fitting parameters, determination
coefficients, expected values, and Lagrange multipliers
| null | null | null |
cs.SD cs.IT math.IT physics.data-an physics.soc-ph
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Consonance is related to the perception of pleasantness arising from a
combination of sounds and has been approached quantitatively using mathematical
relations, physics, information theory, and psychoacoustics. Tonal consonance
is present in timbre, musical tuning, harmony, and melody, and it is used for
conveying sensations, perceptions, and emotions in music. It involves the
physical properties of sound waves and is used to study melody and harmony
through musical intervals and chords. From the perspective of complexity, the
macroscopic properties of a system with many parts frequently rely on the
statistical properties of its constituent elements. Here we show how the tonal
consonance parameters for complex tones can be used to study complexity in
music. We apply this formalism to melody, showing that melodic lines in musical
pieces can be described in terms of the physical properties of melodic
intervals and the existence of an entropy extremalization principle subject to
psychoacoustic macroscopic constraints with musical meaning. This result
connects the human perception of consonance with the complexity of human
creativity in music through the physical properties of the musical stimulus.
|
[
{
"created": "Fri, 14 Oct 2016 17:42:41 GMT",
"version": "v1"
},
{
"created": "Fri, 27 Jan 2017 19:11:05 GMT",
"version": "v2"
},
{
"created": "Thu, 9 Feb 2017 19:07:40 GMT",
"version": "v3"
},
{
"created": "Sun, 23 Apr 2017 16:31:08 GMT",
"version": "v4"
}
] |
2017-04-25
|
[
[
"Useche",
"Jorge",
""
],
[
"Hurtado",
"Rafael",
""
]
] |
Consonance is related to the perception of pleasantness arising from a combination of sounds and has been approached quantitatively using mathematical relations, physics, information theory, and psychoacoustics. Tonal consonance is present in timbre, musical tuning, harmony, and melody, and it is used for conveying sensations, perceptions, and emotions in music. It involves the physical properties of sound waves and is used to study melody and harmony through musical intervals and chords. From the perspective of complexity, the macroscopic properties of a system with many parts frequently rely on the statistical properties of its constituent elements. Here we show how the tonal consonance parameters for complex tones can be used to study complexity in music. We apply this formalism to melody, showing that melodic lines in musical pieces can be described in terms of the physical properties of melodic intervals and the existence of an entropy extremalization principle subject to psychoacoustic macroscopic constraints with musical meaning. This result connects the human perception of consonance with the complexity of human creativity in music through the physical properties of the musical stimulus.
|
2306.04431
|
Tom Lamb
|
Tom A. Lamb, Rudy Brunel, Krishnamurthy DJ Dvijotham, M. Pawan Kumar,
Philip H. S. Torr, Francisco Eiras
|
Faithful Knowledge Distillation
|
7pgs (main content), 4 figures
| null | null | null |
cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
Knowledge distillation (KD) has received much attention due to its success in
compressing networks to allow for their deployment in resource-constrained
systems. While the problem of adversarial robustness has been studied before in
the KD setting, previous works overlook what we term the relative calibration
of the student network with respect to its teacher in terms of soft
confidences. In particular, we focus on two crucial questions with regard to a
teacher-student pair: (i) do the teacher and student disagree at points close
to correctly classified dataset examples, and (ii) is the distilled student as
confident as the teacher around dataset examples? These are critical questions
when considering the deployment of a smaller student network trained from a
robust teacher within a safety-critical setting. To address these questions, we
introduce a faithful imitation framework to discuss the relative calibration of
confidences and provide empirical and certified methods to evaluate the
relative calibration of a student w.r.t. its teacher. Further, to verifiably
align the relative calibration incentives of the student to those of its
teacher, we introduce faithful distillation. Our experiments on the MNIST,
Fashion-MNIST and CIFAR-10 datasets demonstrate the need for such an analysis
and the advantages of the increased verifiability of faithful distillation over
alternative adversarial distillation methods.
|
[
{
"created": "Wed, 7 Jun 2023 13:41:55 GMT",
"version": "v1"
},
{
"created": "Thu, 8 Jun 2023 09:50:27 GMT",
"version": "v2"
},
{
"created": "Fri, 11 Aug 2023 13:39:06 GMT",
"version": "v3"
}
] |
2023-08-14
|
[
[
"Lamb",
"Tom A.",
""
],
[
"Brunel",
"Rudy",
""
],
[
"Dvijotham",
"Krishnamurthy DJ",
""
],
[
"Kumar",
"M. Pawan",
""
],
[
"Torr",
"Philip H. S.",
""
],
[
"Eiras",
"Francisco",
""
]
] |
Knowledge distillation (KD) has received much attention due to its success in compressing networks to allow for their deployment in resource-constrained systems. While the problem of adversarial robustness has been studied before in the KD setting, previous works overlook what we term the relative calibration of the student network with respect to its teacher in terms of soft confidences. In particular, we focus on two crucial questions with regard to a teacher-student pair: (i) do the teacher and student disagree at points close to correctly classified dataset examples, and (ii) is the distilled student as confident as the teacher around dataset examples? These are critical questions when considering the deployment of a smaller student network trained from a robust teacher within a safety-critical setting. To address these questions, we introduce a faithful imitation framework to discuss the relative calibration of confidences and provide empirical and certified methods to evaluate the relative calibration of a student w.r.t. its teacher. Further, to verifiably align the relative calibration incentives of the student to those of its teacher, we introduce faithful distillation. Our experiments on the MNIST, Fashion-MNIST and CIFAR-10 datasets demonstrate the need for such an analysis and the advantages of the increased verifiability of faithful distillation over alternative adversarial distillation methods.
|
1910.05280
|
Masato Tamura
|
Masato Tamura, Tomokazu Murakami
|
Augmented Hard Example Mining for Generalizable Person Re-Identification
|
Submit to WACV2020
| null | null | null |
cs.CV cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Although the performance of person re-identification (Re-ID) has been much
improved by using sophisticated training methods and large-scale labelled
datasets, many existing methods make the impractical assumption that
information of a target domain can be utilized during training. In practice, a
Re-ID system often starts running as soon as it is deployed, hence training
with data from a target domain is unrealistic. To make Re-ID systems more
practical, methods have been proposed that achieve high performance without
information of a target domain. However, they need cumbersome tuning for
training and unusual operations for testing. In this paper, we propose
augmented hard example mining, which can be easily integrated to a common Re-ID
training process and can utilize sophisticated models without any network
modification. The method discovers hard examples on the basis of classification
probabilities, and to make the examples harder, various types of augmentation
are applied to the examples. Among those examples, excessively augmented ones
are eliminated by a classification based selection process. Extensive analysis
shows that our method successfully selects effective examples and achieves
state-of-the-art performance on publicly available benchmark datasets.
|
[
{
"created": "Fri, 11 Oct 2019 16:19:53 GMT",
"version": "v1"
}
] |
2019-10-14
|
[
[
"Tamura",
"Masato",
""
],
[
"Murakami",
"Tomokazu",
""
]
] |
Although the performance of person re-identification (Re-ID) has been much improved by using sophisticated training methods and large-scale labelled datasets, many existing methods make the impractical assumption that information of a target domain can be utilized during training. In practice, a Re-ID system often starts running as soon as it is deployed, hence training with data from a target domain is unrealistic. To make Re-ID systems more practical, methods have been proposed that achieve high performance without information of a target domain. However, they need cumbersome tuning for training and unusual operations for testing. In this paper, we propose augmented hard example mining, which can be easily integrated to a common Re-ID training process and can utilize sophisticated models without any network modification. The method discovers hard examples on the basis of classification probabilities, and to make the examples harder, various types of augmentation are applied to the examples. Among those examples, excessively augmented ones are eliminated by a classification based selection process. Extensive analysis shows that our method successfully selects effective examples and achieves state-of-the-art performance on publicly available benchmark datasets.
|
1408.6520
|
Shirin Sohrabi
|
Shirin Sohrabi and Octavian Udrea and Anton V. Riabov
|
Knowledge Engineering for Planning-Based Hypothesis Generation
|
This paper appears in the Proceedings of the Automated Planning and
Scheduling (ICAPS) Workshop on Knowledge Engineering for Planning and
Scheduling (KEPS)
| null | null | null |
cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this paper, we address the knowledge engineering problems for hypothesis
generation motivated by applications that require timely exploration of
hypotheses under unreliable observations. We looked at two applications:
malware detection and intensive care delivery. In intensive care, the goal is
to generate plausible hypotheses about the condition of the patient from
clinical observations and further refine these hypotheses to create a recovery
plan for the patient. Similarly, preventing malware spread within a corporate
network involves generating hypotheses from network traffic data and selecting
preventive actions. To this end, building on the already established
characterization and use of AI planning for similar problems, we propose use of
planning for the hypothesis generation problem. However, to deal with
uncertainty, incomplete model description and unreliable observations, we need
to use a planner capable of generating multiple high-quality plans. To capture
the model description we propose a language called LTS++ and a web-based tool
that enables the specification of the LTS++ model and a set of observations. We
also proposed a 9-step process that helps provide guidance to the domain expert
in specifying the LTS++ model. The hypotheses are then generated by running a
planner on the translated LTS++ model and the provided trace. The hypotheses
can be visualized and shown to the analyst or can be further investigated
automatically.
|
[
{
"created": "Wed, 27 Aug 2014 15:14:11 GMT",
"version": "v1"
}
] |
2014-08-29
|
[
[
"Sohrabi",
"Shirin",
""
],
[
"Udrea",
"Octavian",
""
],
[
"Riabov",
"Anton V.",
""
]
] |
In this paper, we address the knowledge engineering problems for hypothesis generation motivated by applications that require timely exploration of hypotheses under unreliable observations. We looked at two applications: malware detection and intensive care delivery. In intensive care, the goal is to generate plausible hypotheses about the condition of the patient from clinical observations and further refine these hypotheses to create a recovery plan for the patient. Similarly, preventing malware spread within a corporate network involves generating hypotheses from network traffic data and selecting preventive actions. To this end, building on the already established characterization and use of AI planning for similar problems, we propose use of planning for the hypothesis generation problem. However, to deal with uncertainty, incomplete model description and unreliable observations, we need to use a planner capable of generating multiple high-quality plans. To capture the model description we propose a language called LTS++ and a web-based tool that enables the specification of the LTS++ model and a set of observations. We also proposed a 9-step process that helps provide guidance to the domain expert in specifying the LTS++ model. The hypotheses are then generated by running a planner on the translated LTS++ model and the provided trace. The hypotheses can be visualized and shown to the analyst or can be further investigated automatically.
|
2101.09588
|
Xiaobin Xiong
|
Xiaobin Xiong and Aaron Ames
|
3D Underactuated Bipedal Walking via H-LIP based Gait Synthesis and
Stepping Stabilization
|
20 pages, 24 figures. Paper under review, comments are sincerely
welcome
| null | null | null |
cs.RO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this paper, we holistically present a Hybrid-Linear Inverted Pendulum
(H-LIP) based approach for synthesizing and stabilizing 3D foot-underactuated
bipedal walking, with an emphasis on thorough hardware realization. The H-LIP
is proposed to capture the essential components of the underactuated and
actuated part of the robotic walking. The robot walking gait is then directly
synthesized based on the H-LIP. We comprehensively characterize the periodic
orbits of the H-LIP and provably derive the stepping stabilization via its
step-to-step (S2S) dynamics, which is then utilized to approximate the S2S
dynamics of the horizontal state of the center of mass (COM) of the robotic
walking. The approximation facilities a H-LIP based stepping controller to
provide desired step sizes to stabilize the robotic walking. By realizing the
desired step sizes, the robot achieves dynamic and stable walking. The approach
is fully evaluated in both simulation and experiment on the 3D underactuated
bipedal robot Cassie, which demonstrates dynamic walking behaviors with both
high versatility and robustness.
|
[
{
"created": "Sat, 23 Jan 2021 21:28:04 GMT",
"version": "v1"
},
{
"created": "Fri, 5 Feb 2021 22:35:25 GMT",
"version": "v2"
},
{
"created": "Wed, 3 Nov 2021 00:24:10 GMT",
"version": "v3"
}
] |
2021-11-04
|
[
[
"Xiong",
"Xiaobin",
""
],
[
"Ames",
"Aaron",
""
]
] |
In this paper, we holistically present a Hybrid-Linear Inverted Pendulum (H-LIP) based approach for synthesizing and stabilizing 3D foot-underactuated bipedal walking, with an emphasis on thorough hardware realization. The H-LIP is proposed to capture the essential components of the underactuated and actuated part of the robotic walking. The robot walking gait is then directly synthesized based on the H-LIP. We comprehensively characterize the periodic orbits of the H-LIP and provably derive the stepping stabilization via its step-to-step (S2S) dynamics, which is then utilized to approximate the S2S dynamics of the horizontal state of the center of mass (COM) of the robotic walking. The approximation facilities a H-LIP based stepping controller to provide desired step sizes to stabilize the robotic walking. By realizing the desired step sizes, the robot achieves dynamic and stable walking. The approach is fully evaluated in both simulation and experiment on the 3D underactuated bipedal robot Cassie, which demonstrates dynamic walking behaviors with both high versatility and robustness.
|
0805.4374
|
Jin Xu
|
Jin Xu, Yi Cao, and Biao Chen
|
Capacity Bounds for Broadcast Channels with Confidential Messages
|
27 pages, 1 figure, submitted to IEEE Transaction on Information
Theory
| null |
10.1109/TIT.2009.2027500
| null |
cs.IT math.IT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this paper, we study capacity bounds for discrete memoryless broadcast
channels with confidential messages. Two private messages as well as a common
message are transmitted; the common message is to be decoded by both receivers,
while each private message is only for its intended receiver. In addition, each
private message is to be kept secret from the unintended receiver where secrecy
is measured by equivocation. We propose both inner and outer bounds to the rate
equivocation region for broadcast channels with confidential messages. The
proposed inner bound generalizes Csisz\'{a}r and K\"{o}rner's rate equivocation
region for broadcast channels with a single confidential message, Liu {\em et
al}'s achievable rate region for broadcast channels with perfect secrecy,
Marton's and Gel'fand and Pinsker's achievable rate region for general
broadcast channels. Our proposed outer bounds, together with the inner bound,
helps establish the rate equivocation region of several classes of discrete
memoryless broadcast channels with confidential messages, including less noisy,
deterministic, and semi-deterministic channels. Furthermore, specializing to
the general broadcast channel by removing the confidentiality constraint, our
proposed outer bounds reduce to new capacity outer bounds for the discrete
memory broadcast channel.
|
[
{
"created": "Wed, 28 May 2008 15:36:46 GMT",
"version": "v1"
}
] |
2016-11-17
|
[
[
"Xu",
"Jin",
""
],
[
"Cao",
"Yi",
""
],
[
"Chen",
"Biao",
""
]
] |
In this paper, we study capacity bounds for discrete memoryless broadcast channels with confidential messages. Two private messages as well as a common message are transmitted; the common message is to be decoded by both receivers, while each private message is only for its intended receiver. In addition, each private message is to be kept secret from the unintended receiver where secrecy is measured by equivocation. We propose both inner and outer bounds to the rate equivocation region for broadcast channels with confidential messages. The proposed inner bound generalizes Csisz\'{a}r and K\"{o}rner's rate equivocation region for broadcast channels with a single confidential message, Liu {\em et al}'s achievable rate region for broadcast channels with perfect secrecy, Marton's and Gel'fand and Pinsker's achievable rate region for general broadcast channels. Our proposed outer bounds, together with the inner bound, helps establish the rate equivocation region of several classes of discrete memoryless broadcast channels with confidential messages, including less noisy, deterministic, and semi-deterministic channels. Furthermore, specializing to the general broadcast channel by removing the confidentiality constraint, our proposed outer bounds reduce to new capacity outer bounds for the discrete memory broadcast channel.
|
1812.05816
|
Songyang Zhang
|
Songyang Zhang, Weimin Lei, Wei Zhang, Yunchong Guan
|
Shared Bottleneck Detecction Based on Trend Line Regression for
Multipath Transmission
| null | null | null | null |
cs.NI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The current deployed multipath congestion control algorithms couple all the
subflows together to avoid bandwidth occupation aggressiveness if the subflows
of multipath transmission protocol share common bottleneck with single path
TCP. The coupled congestion control algorithms can guarantee well fairness
property in common bottleneck but result in rate increase conservativeness in
none-sharing bottleneck situation. Thus, the throughput of multipath session
can be further improved when combing with effective shared bottleneck detection
mechanism. This paper proposes a delay trend line regression method to detect
if flows share common bottleneck. Deduced from TCP fluid model, the packet
round trip delay signal shows linear increase property during the queue
building up process of the narrowest link and the delay trend line slopes of
two flows are in close proximity if they traverse the same bottleneck link. The
proposed method is implemented on multipath QUIC golang codebase and extensive
simulations are performed to validate its effectiveness in detecting out flows
traversing common bottleneck. If the subflows are detected out via a common
bottleneck, the sender would perform coupled congestion control algorithm and
perform congestion control seperately on flow level in none sharing bottleneck
case. Results show a multipath session with two subflows can obtain 74\% gain
on average in throughput compared with single path connection when Linked
Increases Algorithm (LIA) is in combination with trend line regession shared
bottlenck detection algorithm in none shared bottleneck, and show well fairness
property in common bottleneck scenarios.
|
[
{
"created": "Fri, 14 Dec 2018 08:19:39 GMT",
"version": "v1"
}
] |
2018-12-17
|
[
[
"Zhang",
"Songyang",
""
],
[
"Lei",
"Weimin",
""
],
[
"Zhang",
"Wei",
""
],
[
"Guan",
"Yunchong",
""
]
] |
The current deployed multipath congestion control algorithms couple all the subflows together to avoid bandwidth occupation aggressiveness if the subflows of multipath transmission protocol share common bottleneck with single path TCP. The coupled congestion control algorithms can guarantee well fairness property in common bottleneck but result in rate increase conservativeness in none-sharing bottleneck situation. Thus, the throughput of multipath session can be further improved when combing with effective shared bottleneck detection mechanism. This paper proposes a delay trend line regression method to detect if flows share common bottleneck. Deduced from TCP fluid model, the packet round trip delay signal shows linear increase property during the queue building up process of the narrowest link and the delay trend line slopes of two flows are in close proximity if they traverse the same bottleneck link. The proposed method is implemented on multipath QUIC golang codebase and extensive simulations are performed to validate its effectiveness in detecting out flows traversing common bottleneck. If the subflows are detected out via a common bottleneck, the sender would perform coupled congestion control algorithm and perform congestion control seperately on flow level in none sharing bottleneck case. Results show a multipath session with two subflows can obtain 74\% gain on average in throughput compared with single path connection when Linked Increases Algorithm (LIA) is in combination with trend line regession shared bottlenck detection algorithm in none shared bottleneck, and show well fairness property in common bottleneck scenarios.
|
2306.06238
|
Harvey Dam
|
Harvey Dam, Vinu Joseph, Aditya Bhaskara, Ganesh Gopalakrishnan,
Saurav Muralidharan, Michael Garland
|
Understanding the Effect of the Long Tail on Neural Network Compression
| null | null | null | null |
cs.LG cs.AI cs.CV
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Network compression is now a mature sub-field of neural network research:
over the last decade, significant progress has been made towards reducing the
size of models and speeding up inference, while maintaining the classification
accuracy. However, many works have observed that focusing on just the overall
accuracy can be misguided. E.g., it has been shown that mismatches between the
full and compressed models can be biased towards under-represented classes.
This raises the important research question, can we achieve network compression
while maintaining "semantic equivalence" with the original network? In this
work, we study this question in the context of the "long tail" phenomenon in
computer vision datasets observed by Feldman, et al. They argue that
memorization of certain inputs (appropriately defined) is essential to
achieving good generalization. As compression limits the capacity of a network
(and hence also its ability to memorize), we study the question: are mismatches
between the full and compressed models correlated with the memorized training
data? We present positive evidence in this direction for image classification
tasks, by considering different base architectures and compression schemes.
|
[
{
"created": "Fri, 9 Jun 2023 20:18:05 GMT",
"version": "v1"
},
{
"created": "Mon, 19 Jun 2023 09:46:09 GMT",
"version": "v2"
},
{
"created": "Tue, 27 Jun 2023 23:14:16 GMT",
"version": "v3"
}
] |
2023-06-29
|
[
[
"Dam",
"Harvey",
""
],
[
"Joseph",
"Vinu",
""
],
[
"Bhaskara",
"Aditya",
""
],
[
"Gopalakrishnan",
"Ganesh",
""
],
[
"Muralidharan",
"Saurav",
""
],
[
"Garland",
"Michael",
""
]
] |
Network compression is now a mature sub-field of neural network research: over the last decade, significant progress has been made towards reducing the size of models and speeding up inference, while maintaining the classification accuracy. However, many works have observed that focusing on just the overall accuracy can be misguided. E.g., it has been shown that mismatches between the full and compressed models can be biased towards under-represented classes. This raises the important research question, can we achieve network compression while maintaining "semantic equivalence" with the original network? In this work, we study this question in the context of the "long tail" phenomenon in computer vision datasets observed by Feldman, et al. They argue that memorization of certain inputs (appropriately defined) is essential to achieving good generalization. As compression limits the capacity of a network (and hence also its ability to memorize), we study the question: are mismatches between the full and compressed models correlated with the memorized training data? We present positive evidence in this direction for image classification tasks, by considering different base architectures and compression schemes.
|
1806.03483
|
Chengyuan Zhang
|
Chengyuan Zhang, Ruipeng Chen, Lei Zhu, Anfeng Liu, Yunwu Lin and Fang
Huang
|
Hierarchical Information Quadtree: Efficient Spatial Temporal Image
Search for Multimedia Stream
|
Published at Multimedia Tools and Applications. arXiv admin note:
text overlap with arXiv:1805.02009
| null |
10.1007/s11042-018-6284-y
| null |
cs.MM cs.IR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Massive amount of multimedia data that contain times- tamps and geographical
information are being generated at an unprecedented scale in many emerging
applications such as photo sharing web site and social networks applications.
Due to their importance, a large body of work has focused on efficiently
computing various spatial image queries. In this paper,we study the spatial
temporal image query which considers three important constraints during the
search including time recency, spatial proximity and visual relevance. A novel
index structure, namely Hierarchical Information Quadtree(\hiq), to efficiently
insert/delete spatial temporal images with high arrive rates. Base on \hiq an
efficient algorithm is developed to support spatial temporal image query. We
show via extensive experimentation with real spatial databases clearly
demonstrate the efficiency of our methods.
|
[
{
"created": "Sat, 9 Jun 2018 14:53:07 GMT",
"version": "v1"
},
{
"created": "Wed, 15 Aug 2018 07:53:15 GMT",
"version": "v2"
}
] |
2018-08-16
|
[
[
"Zhang",
"Chengyuan",
""
],
[
"Chen",
"Ruipeng",
""
],
[
"Zhu",
"Lei",
""
],
[
"Liu",
"Anfeng",
""
],
[
"Lin",
"Yunwu",
""
],
[
"Huang",
"Fang",
""
]
] |
Massive amount of multimedia data that contain times- tamps and geographical information are being generated at an unprecedented scale in many emerging applications such as photo sharing web site and social networks applications. Due to their importance, a large body of work has focused on efficiently computing various spatial image queries. In this paper,we study the spatial temporal image query which considers three important constraints during the search including time recency, spatial proximity and visual relevance. A novel index structure, namely Hierarchical Information Quadtree(\hiq), to efficiently insert/delete spatial temporal images with high arrive rates. Base on \hiq an efficient algorithm is developed to support spatial temporal image query. We show via extensive experimentation with real spatial databases clearly demonstrate the efficiency of our methods.
|
1604.07095
|
Xiaoxiao Guo
|
Xiaoxiao Guo, Satinder Singh, Richard Lewis and Honglak Lee
|
Deep Learning for Reward Design to Improve Monte Carlo Tree Search in
ATARI Games
|
In 25th International Joint Conference on Artificial Intelligence
(IJCAI), 2016
| null | null | null |
cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Monte Carlo Tree Search (MCTS) methods have proven powerful in planning for
sequential decision-making problems such as Go and video games, but their
performance can be poor when the planning depth and sampling trajectories are
limited or when the rewards are sparse. We present an adaptation of PGRD
(policy-gradient for reward-design) for learning a reward-bonus function to
improve UCT (a MCTS algorithm). Unlike previous applications of PGRD in which
the space of reward-bonus functions was limited to linear functions of
hand-coded state-action-features, we use PGRD with a multi-layer convolutional
neural network to automatically learn features from raw perception as well as
to adapt the non-linear reward-bonus function parameters. We also adopt a
variance-reducing gradient method to improve PGRD's performance. The new method
improves UCT's performance on multiple ATARI games compared to UCT without the
reward bonus. Combining PGRD and Deep Learning in this way should make adapting
rewards for MCTS algorithms far more widely and practically applicable than
before.
|
[
{
"created": "Sun, 24 Apr 2016 23:51:18 GMT",
"version": "v1"
}
] |
2016-04-26
|
[
[
"Guo",
"Xiaoxiao",
""
],
[
"Singh",
"Satinder",
""
],
[
"Lewis",
"Richard",
""
],
[
"Lee",
"Honglak",
""
]
] |
Monte Carlo Tree Search (MCTS) methods have proven powerful in planning for sequential decision-making problems such as Go and video games, but their performance can be poor when the planning depth and sampling trajectories are limited or when the rewards are sparse. We present an adaptation of PGRD (policy-gradient for reward-design) for learning a reward-bonus function to improve UCT (a MCTS algorithm). Unlike previous applications of PGRD in which the space of reward-bonus functions was limited to linear functions of hand-coded state-action-features, we use PGRD with a multi-layer convolutional neural network to automatically learn features from raw perception as well as to adapt the non-linear reward-bonus function parameters. We also adopt a variance-reducing gradient method to improve PGRD's performance. The new method improves UCT's performance on multiple ATARI games compared to UCT without the reward bonus. Combining PGRD and Deep Learning in this way should make adapting rewards for MCTS algorithms far more widely and practically applicable than before.
|
2011.07964
|
Martin Hole\v{c}ek
|
Martin Hole\v{c}ek
|
Learning from similarity and information extraction from structured
documents
|
17 pages, 9 figures, manuscript for the IJDAR journal special issue
for ICDAR conference
|
Hole\v{c}ek, M. 2021 Learning from similarity and information
extraction from structured documents; International Journal on Document
Analysis and Recognition (IJDAR) 2021/06/11
|
10.1007/s10032-021-00375-3
| null |
cs.CL cs.IR cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The automation of document processing is gaining recent attention due to the
great potential to reduce manual work through improved methods and hardware.
Neural networks have been successfully applied before - even though they have
been trained only on relatively small datasets with hundreds of documents so
far. To successfully explore deep learning techniques and improve the
information extraction results, a dataset with more than twenty-five thousand
documents has been compiled, anonymized and is published as a part of this
work. We will expand our previous work where we proved that convolutions, graph
convolutions and self-attention can work together and exploit all the
information present in a structured document. Taking the fully trainable method
one step further, we will now design and examine various approaches to using
siamese networks, concepts of similarity, one-shot learning and context/memory
awareness. The aim is to improve micro F1 of per-word classification on the
huge real-world document dataset. The results verify the hypothesis that
trainable access to a similar (yet still different) page together with its
already known target information improves the information extraction.
Furthermore, the experiments confirm that all proposed architecture parts are
all required to beat the previous results. The best model improves the previous
state-of-the-art results by an 8.25 gain in F1 score. Qualitative analysis is
provided to verify that the new model performs better for all target classes.
Additionally, multiple structural observations about the causes of the
underperformance of some architectures are revealed. All the source codes,
parameters and implementation details are published together with the dataset
in the hope to push the research boundaries since all the techniques used in
this work are not problem-specific and can be generalized for other tasks and
contexts.
|
[
{
"created": "Sat, 17 Oct 2020 21:34:52 GMT",
"version": "v1"
},
{
"created": "Sat, 13 Mar 2021 21:36:56 GMT",
"version": "v2"
}
] |
2021-06-15
|
[
[
"Holeček",
"Martin",
""
]
] |
The automation of document processing is gaining recent attention due to the great potential to reduce manual work through improved methods and hardware. Neural networks have been successfully applied before - even though they have been trained only on relatively small datasets with hundreds of documents so far. To successfully explore deep learning techniques and improve the information extraction results, a dataset with more than twenty-five thousand documents has been compiled, anonymized and is published as a part of this work. We will expand our previous work where we proved that convolutions, graph convolutions and self-attention can work together and exploit all the information present in a structured document. Taking the fully trainable method one step further, we will now design and examine various approaches to using siamese networks, concepts of similarity, one-shot learning and context/memory awareness. The aim is to improve micro F1 of per-word classification on the huge real-world document dataset. The results verify the hypothesis that trainable access to a similar (yet still different) page together with its already known target information improves the information extraction. Furthermore, the experiments confirm that all proposed architecture parts are all required to beat the previous results. The best model improves the previous state-of-the-art results by an 8.25 gain in F1 score. Qualitative analysis is provided to verify that the new model performs better for all target classes. Additionally, multiple structural observations about the causes of the underperformance of some architectures are revealed. All the source codes, parameters and implementation details are published together with the dataset in the hope to push the research boundaries since all the techniques used in this work are not problem-specific and can be generalized for other tasks and contexts.
|
2307.09036
|
Yingchaojie Feng
|
Yingchaojie Feng, Xingbo Wang, Kam Kwai Wong, Sijia Wang, Yuhong Lu,
Minfeng Zhu, Baicheng Wang, Wei Chen
|
PromptMagician: Interactive Prompt Engineering for Text-to-Image
Creation
|
Accepted full paper for IEEE VIS 2023
| null |
10.1109/TVCG.2023.3327168
| null |
cs.AI cs.HC
|
http://creativecommons.org/licenses/by/4.0/
|
Generative text-to-image models have gained great popularity among the public
for their powerful capability to generate high-quality images based on natural
language prompts. However, developing effective prompts for desired images can
be challenging due to the complexity and ambiguity of natural language. This
research proposes PromptMagician, a visual analysis system that helps users
explore the image results and refine the input prompts. The backbone of our
system is a prompt recommendation model that takes user prompts as input,
retrieves similar prompt-image pairs from DiffusionDB, and identifies special
(important and relevant) prompt keywords. To facilitate interactive prompt
refinement, PromptMagician introduces a multi-level visualization for the
cross-modal embedding of the retrieved images and recommended keywords, and
supports users in specifying multiple criteria for personalized exploration.
Two usage scenarios, a user study, and expert interviews demonstrate the
effectiveness and usability of our system, suggesting it facilitates prompt
engineering and improves the creativity support of the generative text-to-image
model.
|
[
{
"created": "Tue, 18 Jul 2023 07:46:25 GMT",
"version": "v1"
},
{
"created": "Tue, 15 Aug 2023 09:44:57 GMT",
"version": "v2"
}
] |
2023-11-02
|
[
[
"Feng",
"Yingchaojie",
""
],
[
"Wang",
"Xingbo",
""
],
[
"Wong",
"Kam Kwai",
""
],
[
"Wang",
"Sijia",
""
],
[
"Lu",
"Yuhong",
""
],
[
"Zhu",
"Minfeng",
""
],
[
"Wang",
"Baicheng",
""
],
[
"Chen",
"Wei",
""
]
] |
Generative text-to-image models have gained great popularity among the public for their powerful capability to generate high-quality images based on natural language prompts. However, developing effective prompts for desired images can be challenging due to the complexity and ambiguity of natural language. This research proposes PromptMagician, a visual analysis system that helps users explore the image results and refine the input prompts. The backbone of our system is a prompt recommendation model that takes user prompts as input, retrieves similar prompt-image pairs from DiffusionDB, and identifies special (important and relevant) prompt keywords. To facilitate interactive prompt refinement, PromptMagician introduces a multi-level visualization for the cross-modal embedding of the retrieved images and recommended keywords, and supports users in specifying multiple criteria for personalized exploration. Two usage scenarios, a user study, and expert interviews demonstrate the effectiveness and usability of our system, suggesting it facilitates prompt engineering and improves the creativity support of the generative text-to-image model.
|
2209.06692
|
Lingyu Du
|
Lingyu Du, Guohao Lan
|
FreeGaze: Resource-efficient Gaze Estimation via Frequency Domain
Contrastive Learning
| null | null | null | null |
cs.CV cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Gaze estimation is of great importance to many scientific fields and daily
applications, ranging from fundamental research in cognitive psychology to
attention-aware mobile systems. While recent advancements in deep learning have
yielded remarkable successes in building highly accurate gaze estimation
systems, the associated high computational cost and the reliance on large-scale
labeled gaze data for supervised learning place challenges on the practical use
of existing solutions. To move beyond these limitations, we present FreeGaze, a
resource-efficient framework for unsupervised gaze representation learning.
FreeGaze incorporates the frequency domain gaze estimation and the contrastive
gaze representation learning in its design. The former significantly alleviates
the computational burden in both system calibration and gaze estimation, and
dramatically reduces the system latency; while the latter overcomes the data
labeling hurdle of existing supervised learning-based counterparts, and ensures
efficient gaze representation learning in the absence of gaze label. Our
evaluation on two gaze estimation datasets shows that FreeGaze can achieve
comparable gaze estimation accuracy with existing supervised learning-based
approach, while enabling up to 6.81 and 1.67 times speedup in system
calibration and gaze estimation, respectively.
|
[
{
"created": "Wed, 14 Sep 2022 14:51:52 GMT",
"version": "v1"
}
] |
2022-09-15
|
[
[
"Du",
"Lingyu",
""
],
[
"Lan",
"Guohao",
""
]
] |
Gaze estimation is of great importance to many scientific fields and daily applications, ranging from fundamental research in cognitive psychology to attention-aware mobile systems. While recent advancements in deep learning have yielded remarkable successes in building highly accurate gaze estimation systems, the associated high computational cost and the reliance on large-scale labeled gaze data for supervised learning place challenges on the practical use of existing solutions. To move beyond these limitations, we present FreeGaze, a resource-efficient framework for unsupervised gaze representation learning. FreeGaze incorporates the frequency domain gaze estimation and the contrastive gaze representation learning in its design. The former significantly alleviates the computational burden in both system calibration and gaze estimation, and dramatically reduces the system latency; while the latter overcomes the data labeling hurdle of existing supervised learning-based counterparts, and ensures efficient gaze representation learning in the absence of gaze label. Our evaluation on two gaze estimation datasets shows that FreeGaze can achieve comparable gaze estimation accuracy with existing supervised learning-based approach, while enabling up to 6.81 and 1.67 times speedup in system calibration and gaze estimation, respectively.
|
2011.10134
|
Quanquan Gu
|
Dongruo Zhou and Jiahao Chen and Quanquan Gu
|
Provable Multi-Objective Reinforcement Learning with Generative Models
|
10 pages, Workshop on Real-World Reinforcement Learning at the 34th
Conference on Neural Information ProcessingSystems (NeurIPS 2020), Vancouver,
Canada
| null | null | null |
cs.LG cs.AI math.OC
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Multi-objective reinforcement learning (MORL) is an extension of ordinary,
single-objective reinforcement learning (RL) that is applicable to many
real-world tasks where multiple objectives exist without known relative costs.
We study the problem of single policy MORL, which learns an optimal policy
given the preference of objectives. Existing methods require strong assumptions
such as exact knowledge of the multi-objective Markov decision process, and are
analyzed in the limit of infinite data and time. We propose a new algorithm
called model-based envelop value iteration (EVI), which generalizes the
enveloped multi-objective $Q$-learning algorithm in Yang et al., 2019. Our
method can learn a near-optimal value function with polynomial sample
complexity and linear convergence speed. To the best of our knowledge, this is
the first finite-sample analysis of MORL algorithms.
|
[
{
"created": "Thu, 19 Nov 2020 22:35:31 GMT",
"version": "v1"
},
{
"created": "Mon, 11 Jan 2021 07:28:13 GMT",
"version": "v2"
}
] |
2021-01-12
|
[
[
"Zhou",
"Dongruo",
""
],
[
"Chen",
"Jiahao",
""
],
[
"Gu",
"Quanquan",
""
]
] |
Multi-objective reinforcement learning (MORL) is an extension of ordinary, single-objective reinforcement learning (RL) that is applicable to many real-world tasks where multiple objectives exist without known relative costs. We study the problem of single policy MORL, which learns an optimal policy given the preference of objectives. Existing methods require strong assumptions such as exact knowledge of the multi-objective Markov decision process, and are analyzed in the limit of infinite data and time. We propose a new algorithm called model-based envelop value iteration (EVI), which generalizes the enveloped multi-objective $Q$-learning algorithm in Yang et al., 2019. Our method can learn a near-optimal value function with polynomial sample complexity and linear convergence speed. To the best of our knowledge, this is the first finite-sample analysis of MORL algorithms.
|
1301.3369
|
Yuichiro Fujiwara
|
Yuichiro Fujiwara
|
Self-synchronizing pulse position modulation with error tolerance
|
11 pages, 1 figure, 3 tables. Final accepted version for publication
in the IEEE Transactions on Information Theory. This version incorporates
minor corrections and some improvements including additional explicit
examples, performance comparisons, and a discussion on symbol error rates for
use in an FSO link
|
IEEE Transactions on Information Theory 59 (2013) 5352-5362
|
10.1109/TIT.2013.2262094
| null |
cs.IT math.CO math.IT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Pulse position modulation (PPM) is a popular signal modulation technique
which creates M-ary data by means of the position of a pulse within a time
interval. While PPM and its variations have great advantages in many contexts,
this type of modulation is vulnerable to loss of synchronization, potentially
causing a severe error floor or throughput penalty even when little or no noise
is assumed. Another disadvantage is that this type of modulation typically
offers no error correction mechanism on its own, making them sensitive to
intersymbol interference and environmental noise. In this paper we propose a
coding theoretic variation of PPM that allows for significantly more efficient
symbol and frame synchronization as well as strong error correction. The
proposed scheme can be divided into a synchronization layer and a modulation
layer. This makes our technique compatible with major existing techniques such
as standard PPM, multipluse PPM, and expurgated PPM as well in that the scheme
can be realized by adding a simple synchronization layer to one of these
standard techniques. We also develop a generalization of expurgated PPM suited
for the modulation layer of the proposed self-synchronizing modulation scheme.
This generalized PPM can also be used as stand-alone error-correcting PPM with
a larger number of available symbols.
|
[
{
"created": "Tue, 15 Jan 2013 14:51:12 GMT",
"version": "v1"
},
{
"created": "Sun, 5 May 2013 03:19:36 GMT",
"version": "v2"
}
] |
2013-08-20
|
[
[
"Fujiwara",
"Yuichiro",
""
]
] |
Pulse position modulation (PPM) is a popular signal modulation technique which creates M-ary data by means of the position of a pulse within a time interval. While PPM and its variations have great advantages in many contexts, this type of modulation is vulnerable to loss of synchronization, potentially causing a severe error floor or throughput penalty even when little or no noise is assumed. Another disadvantage is that this type of modulation typically offers no error correction mechanism on its own, making them sensitive to intersymbol interference and environmental noise. In this paper we propose a coding theoretic variation of PPM that allows for significantly more efficient symbol and frame synchronization as well as strong error correction. The proposed scheme can be divided into a synchronization layer and a modulation layer. This makes our technique compatible with major existing techniques such as standard PPM, multipluse PPM, and expurgated PPM as well in that the scheme can be realized by adding a simple synchronization layer to one of these standard techniques. We also develop a generalization of expurgated PPM suited for the modulation layer of the proposed self-synchronizing modulation scheme. This generalized PPM can also be used as stand-alone error-correcting PPM with a larger number of available symbols.
|
2002.05915
|
Jinglian He
|
Jinglian He, Kaiqiang Yu and Yuanming Shi
|
Coordinated Passive Beamforming for Distributed Intelligent Reflecting
Surfaces Network
|
Accepted by Pro. IEEE Veh. Technol. Conf. (VTC), Antwerp, Belgium,
May 2020
| null | null | null |
cs.IT eess.SP math.IT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Intelligent reflecting surface (IRS) is a proposing technology in 6G to
enhance the performance of wireless networks by smartly reconfiguring the
propagation environment with a large number of passive reflecting elements.
However, current works mainly focus on single IRS-empowered wireless networks,
where the channel rank deficiency problem has emerged. In this paper, we
propose a distributed IRS-empowered communication network architecture, where
multiple source-destination pairs communicate through multiple distributed
IRSs. We further contribute to maximize the achievable sum-rates in this
network via jointly optimizing the transmit power vector at the sources and the
phase shift matrix with passive beamforming at all distributed IRSs.
Unfortunately, this problem turns out to be non-convex and highly intractable,
for which an alternating approach is developed via solving the resulting
fractional programming problems alternatively. In particular, the closed-form
expressions are proposed for coordinated passive beamforming at IRSs. The
numerical results will demonstrate the algorithmic advantages and desirable
performances of the distributed IRS-empowered communication network.
|
[
{
"created": "Fri, 14 Feb 2020 08:31:27 GMT",
"version": "v1"
}
] |
2020-02-17
|
[
[
"He",
"Jinglian",
""
],
[
"Yu",
"Kaiqiang",
""
],
[
"Shi",
"Yuanming",
""
]
] |
Intelligent reflecting surface (IRS) is a proposing technology in 6G to enhance the performance of wireless networks by smartly reconfiguring the propagation environment with a large number of passive reflecting elements. However, current works mainly focus on single IRS-empowered wireless networks, where the channel rank deficiency problem has emerged. In this paper, we propose a distributed IRS-empowered communication network architecture, where multiple source-destination pairs communicate through multiple distributed IRSs. We further contribute to maximize the achievable sum-rates in this network via jointly optimizing the transmit power vector at the sources and the phase shift matrix with passive beamforming at all distributed IRSs. Unfortunately, this problem turns out to be non-convex and highly intractable, for which an alternating approach is developed via solving the resulting fractional programming problems alternatively. In particular, the closed-form expressions are proposed for coordinated passive beamforming at IRSs. The numerical results will demonstrate the algorithmic advantages and desirable performances of the distributed IRS-empowered communication network.
|
1508.02082
|
Binjie Benjamin Lim
|
Benjamin Lim
|
Vulnerability Analysis of GWireless
| null | null | null | null |
cs.CR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Wireless networking has become very popular in recent years due to the
increase in adoption of mobile devices. As more and more employees demand for
Wi-Fi access for their devices, more companies have been jumping onto the
"Bring Your Own Device" (BYOD) bandwagon[1] to appease their employees. One
such example of an enterprise wireless infrastructure is the George Washington
University's GWireless.
For this project, I will attempt to capture hashes of authentication
credentials from users who are connecting to the GWireless network using what
is commonly known as the "evil twin" attack. I will document the hardware,
software used and steps taken to configure the devices. I will then evaluate
the feasibility of such an attack, explore variations of the attack and
document measures that can be taken to prevent such an attack.
|
[
{
"created": "Sun, 9 Aug 2015 20:45:50 GMT",
"version": "v1"
}
] |
2015-08-11
|
[
[
"Lim",
"Benjamin",
""
]
] |
Wireless networking has become very popular in recent years due to the increase in adoption of mobile devices. As more and more employees demand for Wi-Fi access for their devices, more companies have been jumping onto the "Bring Your Own Device" (BYOD) bandwagon[1] to appease their employees. One such example of an enterprise wireless infrastructure is the George Washington University's GWireless. For this project, I will attempt to capture hashes of authentication credentials from users who are connecting to the GWireless network using what is commonly known as the "evil twin" attack. I will document the hardware, software used and steps taken to configure the devices. I will then evaluate the feasibility of such an attack, explore variations of the attack and document measures that can be taken to prevent such an attack.
|
2209.00084
|
Sudeep Pasricha
|
Febin Sunny, Mahdi Nikdast, Sudeep Pasricha
|
RecLight: A Recurrent Neural Network Accelerator with Integrated Silicon
Photonics
| null | null | null | null |
cs.LG cs.AR cs.NE
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Recurrent Neural Networks (RNNs) are used in applications that learn
dependencies in data sequences, such as speech recognition, human activity
recognition, and anomaly detection. In recent years, newer RNN variants, such
as GRUs and LSTMs, have been used for implementing these applications. As many
of these applications are employed in real-time scenarios, accelerating
RNN/LSTM/GRU inference is crucial. In this paper, we propose a novel photonic
hardware accelerator called RecLight for accelerating simple RNNs, GRUs, and
LSTMs. Simulation results indicate that RecLight achieves 37x lower
energy-per-bit and 10% better throughput compared to the state-of-the-art.
|
[
{
"created": "Wed, 31 Aug 2022 19:36:01 GMT",
"version": "v1"
}
] |
2022-09-02
|
[
[
"Sunny",
"Febin",
""
],
[
"Nikdast",
"Mahdi",
""
],
[
"Pasricha",
"Sudeep",
""
]
] |
Recurrent Neural Networks (RNNs) are used in applications that learn dependencies in data sequences, such as speech recognition, human activity recognition, and anomaly detection. In recent years, newer RNN variants, such as GRUs and LSTMs, have been used for implementing these applications. As many of these applications are employed in real-time scenarios, accelerating RNN/LSTM/GRU inference is crucial. In this paper, we propose a novel photonic hardware accelerator called RecLight for accelerating simple RNNs, GRUs, and LSTMs. Simulation results indicate that RecLight achieves 37x lower energy-per-bit and 10% better throughput compared to the state-of-the-art.
|
cs/0206028
|
Wolfgang Eiden
|
Wolfgang Eiden
|
Knowledge management for enterprises (Wissensmanagement fuer
Unternehmen)
|
published in January 2000, 22 pages, 13 figures, german
| null | null | null |
cs.IR cs.AI
| null |
Although knowledge is one of the most valuable resource of enterprises and an
important production and competition factor, this intellectual potential is
often used (or maintained) only inadequate by the enterprises. Therefore, in a
globalised and growing market the optimal usage of existing knowledge
represents a key factor for enterprises of the future. Here, knowledge
management systems should engage facilitating. Because geographically far
distributed establishments cause, however, a distributed system, this paper
should uncover the spectrum connected with it and present a possible basic
approach which is based on ontologies and modern, platform independent
technologies. Last but not least this attempt, as well as general questions of
the knowledge management, are discussed.
|
[
{
"created": "Wed, 19 Jun 2002 22:13:41 GMT",
"version": "v1"
},
{
"created": "Thu, 1 Aug 2002 00:09:30 GMT",
"version": "v2"
}
] |
2016-08-31
|
[
[
"Eiden",
"Wolfgang",
""
]
] |
Although knowledge is one of the most valuable resource of enterprises and an important production and competition factor, this intellectual potential is often used (or maintained) only inadequate by the enterprises. Therefore, in a globalised and growing market the optimal usage of existing knowledge represents a key factor for enterprises of the future. Here, knowledge management systems should engage facilitating. Because geographically far distributed establishments cause, however, a distributed system, this paper should uncover the spectrum connected with it and present a possible basic approach which is based on ontologies and modern, platform independent technologies. Last but not least this attempt, as well as general questions of the knowledge management, are discussed.
|
2304.05008
|
Dongqi Han
|
Dongqi Han, Kenji Doya, Dongsheng Li, Jun Tani
|
Habits and goals in synergy: a variational Bayesian framework for
behavior
| null |
Nat Commun 15, 4461 (2024)
|
10.1038/s41467-024-48577-7
| null |
cs.LG cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
How to behave efficiently and flexibly is a central problem for understanding
biological agents and creating intelligent embodied AI. It has been well known
that behavior can be classified as two types: reward-maximizing habitual
behavior, which is fast while inflexible; and goal-directed behavior, which is
flexible while slow. Conventionally, habitual and goal-directed behaviors are
considered handled by two distinct systems in the brain. Here, we propose to
bridge the gap between the two behaviors, drawing on the principles of
variational Bayesian theory. We incorporate both behaviors in one framework by
introducing a Bayesian latent variable called "intention". The habitual
behavior is generated by using prior distribution of intention, which is
goal-less; and the goal-directed behavior is generated by the posterior
distribution of intention, which is conditioned on the goal. Building on this
idea, we present a novel Bayesian framework for modeling behaviors. Our
proposed framework enables skill sharing between the two kinds of behaviors,
and by leveraging the idea of predictive coding, it enables an agent to
seamlessly generalize from habitual to goal-directed behavior without requiring
additional training. The proposed framework suggests a fresh perspective for
cognitive science and embodied AI, highlighting the potential for greater
integration between habitual and goal-directed behaviors.
|
[
{
"created": "Tue, 11 Apr 2023 06:28:14 GMT",
"version": "v1"
}
] |
2024-07-09
|
[
[
"Han",
"Dongqi",
""
],
[
"Doya",
"Kenji",
""
],
[
"Li",
"Dongsheng",
""
],
[
"Tani",
"Jun",
""
]
] |
How to behave efficiently and flexibly is a central problem for understanding biological agents and creating intelligent embodied AI. It has been well known that behavior can be classified as two types: reward-maximizing habitual behavior, which is fast while inflexible; and goal-directed behavior, which is flexible while slow. Conventionally, habitual and goal-directed behaviors are considered handled by two distinct systems in the brain. Here, we propose to bridge the gap between the two behaviors, drawing on the principles of variational Bayesian theory. We incorporate both behaviors in one framework by introducing a Bayesian latent variable called "intention". The habitual behavior is generated by using prior distribution of intention, which is goal-less; and the goal-directed behavior is generated by the posterior distribution of intention, which is conditioned on the goal. Building on this idea, we present a novel Bayesian framework for modeling behaviors. Our proposed framework enables skill sharing between the two kinds of behaviors, and by leveraging the idea of predictive coding, it enables an agent to seamlessly generalize from habitual to goal-directed behavior without requiring additional training. The proposed framework suggests a fresh perspective for cognitive science and embodied AI, highlighting the potential for greater integration between habitual and goal-directed behaviors.
|
2008.11459
|
Xin Kong
|
Xin Kong, Xuemeng Yang, Guangyao Zhai, Xiangrui Zhao, Xianfang Zeng,
Mengmeng Wang, Yong Liu, Wanlong Li, Feng Wen
|
Semantic Graph Based Place Recognition for 3D Point Clouds
|
8 pages. Accpeted by IROS-2020
| null | null | null |
cs.CV cs.RO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Due to the difficulty in generating the effective descriptors which are
robust to occlusion and viewpoint changes, place recognition for 3D point cloud
remains an open issue. Unlike most of the existing methods that focus on
extracting local, global, and statistical features of raw point clouds, our
method aims at the semantic level that can be superior in terms of robustness
to environmental changes. Inspired by the perspective of humans, who recognize
scenes through identifying semantic objects and capturing their relations, this
paper presents a novel semantic graph based approach for place recognition.
First, we propose a novel semantic graph representation for the point cloud
scenes by reserving the semantic and topological information of the raw point
cloud. Thus, place recognition is modeled as a graph matching problem. Then we
design a fast and effective graph similarity network to compute the similarity.
Exhaustive evaluations on the KITTI dataset show that our approach is robust to
the occlusion as well as viewpoint changes and outperforms the state-of-the-art
methods with a large margin. Our code is available at:
\url{https://github.com/kxhit/SG_PR}.
|
[
{
"created": "Wed, 26 Aug 2020 09:27:26 GMT",
"version": "v1"
}
] |
2020-08-27
|
[
[
"Kong",
"Xin",
""
],
[
"Yang",
"Xuemeng",
""
],
[
"Zhai",
"Guangyao",
""
],
[
"Zhao",
"Xiangrui",
""
],
[
"Zeng",
"Xianfang",
""
],
[
"Wang",
"Mengmeng",
""
],
[
"Liu",
"Yong",
""
],
[
"Li",
"Wanlong",
""
],
[
"Wen",
"Feng",
""
]
] |
Due to the difficulty in generating the effective descriptors which are robust to occlusion and viewpoint changes, place recognition for 3D point cloud remains an open issue. Unlike most of the existing methods that focus on extracting local, global, and statistical features of raw point clouds, our method aims at the semantic level that can be superior in terms of robustness to environmental changes. Inspired by the perspective of humans, who recognize scenes through identifying semantic objects and capturing their relations, this paper presents a novel semantic graph based approach for place recognition. First, we propose a novel semantic graph representation for the point cloud scenes by reserving the semantic and topological information of the raw point cloud. Thus, place recognition is modeled as a graph matching problem. Then we design a fast and effective graph similarity network to compute the similarity. Exhaustive evaluations on the KITTI dataset show that our approach is robust to the occlusion as well as viewpoint changes and outperforms the state-of-the-art methods with a large margin. Our code is available at: \url{https://github.com/kxhit/SG_PR}.
|
2206.14452
|
Yiqi Deng
|
Yiqi Deng and Siu Ming Yiu
|
Deep Multiple Instance Learning For Forecasting Stock Trends Using
Financial News
|
17 pages, 4 figures
| null | null | null |
cs.LG q-fin.CP
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
A major source of information can be taken from financial news articles,
which have some correlations about the fluctuation of stock trends. In this
paper, we investigate the influences of financial news on the stock trends,
from a multi-instance view. The intuition behind this is based on the news
uncertainty of varying intervals of news occurrences and the lack of annotation
in every single financial news. Under the scenario of Multiple Instance
Learning (MIL) where training instances are arranged in bags, and a label is
assigned for the entire bag instead of instances, we develop a flexible and
adaptive multi-instance learning model and evaluate its ability in directional
movement forecast of Standard & Poors 500 index on financial news dataset.
Specifically, we treat each trading day as one bag, with certain amounts of
news happening on each trading day as instances in each bag. Experiment results
demonstrate that our proposed multi-instance-based framework gains outstanding
results in terms of the accuracy of trend prediction, compared with other
state-of-art approaches and baselines.
|
[
{
"created": "Wed, 29 Jun 2022 08:00:13 GMT",
"version": "v1"
}
] |
2022-06-30
|
[
[
"Deng",
"Yiqi",
""
],
[
"Yiu",
"Siu Ming",
""
]
] |
A major source of information can be taken from financial news articles, which have some correlations about the fluctuation of stock trends. In this paper, we investigate the influences of financial news on the stock trends, from a multi-instance view. The intuition behind this is based on the news uncertainty of varying intervals of news occurrences and the lack of annotation in every single financial news. Under the scenario of Multiple Instance Learning (MIL) where training instances are arranged in bags, and a label is assigned for the entire bag instead of instances, we develop a flexible and adaptive multi-instance learning model and evaluate its ability in directional movement forecast of Standard & Poors 500 index on financial news dataset. Specifically, we treat each trading day as one bag, with certain amounts of news happening on each trading day as instances in each bag. Experiment results demonstrate that our proposed multi-instance-based framework gains outstanding results in terms of the accuracy of trend prediction, compared with other state-of-art approaches and baselines.
|
2208.01721
|
Aresh Dadlani
|
Sarina Jami, Iman Sahebi, Mohammad M. Sabermahani, Seyed P.
Shariatpanahi, Aresh Dadlani, Behrouz Maham
|
Rumor Stance Classification in Online Social Networks: The
State-of-the-Art, Prospects, and Future Challenges
|
16 pages, 3 figures, journal
| null | null | null |
cs.SI cs.NI
|
http://creativecommons.org/licenses/by/4.0/
|
The emergence of the Internet as a ubiquitous technology has facilitated the
rapid evolution of social media as the leading virtual platform for
communication, content sharing, and information dissemination. In spite of
revolutionizing the way news is delivered to people, this technology has also
brought along with itself inevitable demerits. One such drawback is the spread
of rumors expedited by social media platforms, which may provoke doubt and
fear. Therefore, it is essential to debunk rumors before their widespread use.
Over the years, many studies have been conducted to develop effective rumor
verification systems. One aspect of such studies focuses on rumor stance
classification, which involves the task of utilizing user viewpoints regarding
a rumorous post to better predict the veracity of a rumor. Relying on user
stances in rumor verification has gained significant importance, for it has
resulted in significant improvements in the model performance. In this paper,
we conduct a comprehensive literature review of rumor stance classification in
complex online social networks (OSNs). In particular, we present a thorough
description of these approaches and compare their performances. Moreover, we
introduce multiple datasets available for this purpose and highlight their
limitations. Finally, challenges and future directions are discussed to
stimulate further relevant research efforts.
|
[
{
"created": "Tue, 2 Aug 2022 20:07:49 GMT",
"version": "v1"
},
{
"created": "Mon, 31 Oct 2022 19:37:10 GMT",
"version": "v2"
}
] |
2022-11-02
|
[
[
"Jami",
"Sarina",
""
],
[
"Sahebi",
"Iman",
""
],
[
"Sabermahani",
"Mohammad M.",
""
],
[
"Shariatpanahi",
"Seyed P.",
""
],
[
"Dadlani",
"Aresh",
""
],
[
"Maham",
"Behrouz",
""
]
] |
The emergence of the Internet as a ubiquitous technology has facilitated the rapid evolution of social media as the leading virtual platform for communication, content sharing, and information dissemination. In spite of revolutionizing the way news is delivered to people, this technology has also brought along with itself inevitable demerits. One such drawback is the spread of rumors expedited by social media platforms, which may provoke doubt and fear. Therefore, it is essential to debunk rumors before their widespread use. Over the years, many studies have been conducted to develop effective rumor verification systems. One aspect of such studies focuses on rumor stance classification, which involves the task of utilizing user viewpoints regarding a rumorous post to better predict the veracity of a rumor. Relying on user stances in rumor verification has gained significant importance, for it has resulted in significant improvements in the model performance. In this paper, we conduct a comprehensive literature review of rumor stance classification in complex online social networks (OSNs). In particular, we present a thorough description of these approaches and compare their performances. Moreover, we introduce multiple datasets available for this purpose and highlight their limitations. Finally, challenges and future directions are discussed to stimulate further relevant research efforts.
|
1808.09772
|
Antoine Tixier
|
Antoine J.-P. Tixier
|
Notes on Deep Learning for NLP
|
work in progress
| null | null | null |
cs.CL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
My notes on Deep Learning for NLP.
|
[
{
"created": "Wed, 29 Aug 2018 12:58:45 GMT",
"version": "v1"
},
{
"created": "Thu, 30 Aug 2018 17:44:54 GMT",
"version": "v2"
}
] |
2018-08-31
|
[
[
"Tixier",
"Antoine J. -P.",
""
]
] |
My notes on Deep Learning for NLP.
|
1804.03230
|
Tien-Ju Yang
|
Tien-Ju Yang, Andrew Howard, Bo Chen, Xiao Zhang, Alec Go, Mark
Sandler, Vivienne Sze, Hartwig Adam
|
NetAdapt: Platform-Aware Neural Network Adaptation for Mobile
Applications
|
Accepted by ECCV 2018
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
This work proposes an algorithm, called NetAdapt, that automatically adapts a
pre-trained deep neural network to a mobile platform given a resource budget.
While many existing algorithms simplify networks based on the number of MACs or
weights, optimizing those indirect metrics may not necessarily reduce the
direct metrics, such as latency and energy consumption. To solve this problem,
NetAdapt incorporates direct metrics into its adaptation algorithm. These
direct metrics are evaluated using empirical measurements, so that detailed
knowledge of the platform and toolchain is not required. NetAdapt automatically
and progressively simplifies a pre-trained network until the resource budget is
met while maximizing the accuracy. Experiment results show that NetAdapt
achieves better accuracy versus latency trade-offs on both mobile CPU and
mobile GPU, compared with the state-of-the-art automated network simplification
algorithms. For image classification on the ImageNet dataset, NetAdapt achieves
up to a 1.7$\times$ speedup in measured inference latency with equal or higher
accuracy on MobileNets (V1&V2).
|
[
{
"created": "Mon, 9 Apr 2018 20:45:26 GMT",
"version": "v1"
},
{
"created": "Fri, 28 Sep 2018 19:20:16 GMT",
"version": "v2"
}
] |
2018-10-02
|
[
[
"Yang",
"Tien-Ju",
""
],
[
"Howard",
"Andrew",
""
],
[
"Chen",
"Bo",
""
],
[
"Zhang",
"Xiao",
""
],
[
"Go",
"Alec",
""
],
[
"Sandler",
"Mark",
""
],
[
"Sze",
"Vivienne",
""
],
[
"Adam",
"Hartwig",
""
]
] |
This work proposes an algorithm, called NetAdapt, that automatically adapts a pre-trained deep neural network to a mobile platform given a resource budget. While many existing algorithms simplify networks based on the number of MACs or weights, optimizing those indirect metrics may not necessarily reduce the direct metrics, such as latency and energy consumption. To solve this problem, NetAdapt incorporates direct metrics into its adaptation algorithm. These direct metrics are evaluated using empirical measurements, so that detailed knowledge of the platform and toolchain is not required. NetAdapt automatically and progressively simplifies a pre-trained network until the resource budget is met while maximizing the accuracy. Experiment results show that NetAdapt achieves better accuracy versus latency trade-offs on both mobile CPU and mobile GPU, compared with the state-of-the-art automated network simplification algorithms. For image classification on the ImageNet dataset, NetAdapt achieves up to a 1.7$\times$ speedup in measured inference latency with equal or higher accuracy on MobileNets (V1&V2).
|
2304.00057
|
Khandaker Foysal Haque
|
Khandaker Foysal Haque, Milin Zhang, Francesco Restuccia
|
SiMWiSense: Simultaneous Multi-Subject Activity Classification Through
Wi-Fi Signals
|
This work has been accepted for publication in IEEE WoWMoM 2023
| null |
10.1109/WoWMoM57956.2023.00019
| null |
cs.NI
|
http://creativecommons.org/licenses/by/4.0/
|
Recent advances in Wi-Fi sensing have ushered in a plethora of pervasive
applications in home surveillance, remote healthcare, road safety, and home
entertainment, among others. Most of the existing works are limited to the
activity classification of a single human subject at a given time. Conversely,
a more realistic scenario is to achieve simultaneous, multi-subject activity
classification. The first key challenge in that context is that the number of
classes grows exponentially with the number of subjects and activities.
Moreover, it is known that Wi-Fi sensing systems struggle to adapt to new
environments and subjects. To address both issues, we propose SiMWiSense, the
first framework for simultaneous multi-subject activity classification based on
Wi-Fi that generalizes to multiple environments and subjects. We address the
scalability issue by using the Channel State Information (CSI) computed from
the device positioned closest to the subject. We experimentally prove this
intuition by confirming that the best accuracy is experienced when the CSI
computed by the transceiver positioned closest to the subject is used for
classification. To address the generalization issue, we develop a brand-new
few-shot learning algorithm named Feature Reusable Embedding Learning (FREL).
Through an extensive data collection campaign in 3 different environments and 3
subjects performing 20 different activities simultaneously, we demonstrate that
SiMWiSense achieves classification accuracy of up to 97%, while FREL improves
the accuracy by 85% in comparison to a traditional Convolutional Neural Network
(CNN) and up to 20% when compared to the state-of-the-art few-shot embedding
learning (FSEL), by using only 15 seconds of additional data for each class.
For reproducibility purposes, we share our 1TB dataset and code repository.
|
[
{
"created": "Fri, 31 Mar 2023 18:19:23 GMT",
"version": "v1"
}
] |
2023-09-11
|
[
[
"Haque",
"Khandaker Foysal",
""
],
[
"Zhang",
"Milin",
""
],
[
"Restuccia",
"Francesco",
""
]
] |
Recent advances in Wi-Fi sensing have ushered in a plethora of pervasive applications in home surveillance, remote healthcare, road safety, and home entertainment, among others. Most of the existing works are limited to the activity classification of a single human subject at a given time. Conversely, a more realistic scenario is to achieve simultaneous, multi-subject activity classification. The first key challenge in that context is that the number of classes grows exponentially with the number of subjects and activities. Moreover, it is known that Wi-Fi sensing systems struggle to adapt to new environments and subjects. To address both issues, we propose SiMWiSense, the first framework for simultaneous multi-subject activity classification based on Wi-Fi that generalizes to multiple environments and subjects. We address the scalability issue by using the Channel State Information (CSI) computed from the device positioned closest to the subject. We experimentally prove this intuition by confirming that the best accuracy is experienced when the CSI computed by the transceiver positioned closest to the subject is used for classification. To address the generalization issue, we develop a brand-new few-shot learning algorithm named Feature Reusable Embedding Learning (FREL). Through an extensive data collection campaign in 3 different environments and 3 subjects performing 20 different activities simultaneously, we demonstrate that SiMWiSense achieves classification accuracy of up to 97%, while FREL improves the accuracy by 85% in comparison to a traditional Convolutional Neural Network (CNN) and up to 20% when compared to the state-of-the-art few-shot embedding learning (FSEL), by using only 15 seconds of additional data for each class. For reproducibility purposes, we share our 1TB dataset and code repository.
|
1710.02192
|
Juan Castorena
|
Juan Castorena and Siddharth Agarwal
|
Ground Edge based LIDAR Localization without a Reflectivity Calibration
for Autonomous Driving
| null | null |
10.1109/LRA.2017.2748180
| null |
cs.RO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this work we propose an alternative formulation to the problem of ground
reflectivity grid based localization involving laser scanned data from multiple
LIDARs mounted on autonomous vehicles. The driving idea of our localization
formulation is an alternative edge reflectivity grid representation which is
invariant to laser source, angle of incidence, range and robot surveying
motion. Such property eliminates the need of the post-factory reflectivity
calibration whose time requirements are infeasible in mass produced
robots/vehicles. Our experiments demonstrate that we can achieve better
performance than state of the art on ground reflectivity inference-map based
localization at no additional computational burden.
|
[
{
"created": "Thu, 5 Oct 2017 19:56:32 GMT",
"version": "v1"
}
] |
2017-10-09
|
[
[
"Castorena",
"Juan",
""
],
[
"Agarwal",
"Siddharth",
""
]
] |
In this work we propose an alternative formulation to the problem of ground reflectivity grid based localization involving laser scanned data from multiple LIDARs mounted on autonomous vehicles. The driving idea of our localization formulation is an alternative edge reflectivity grid representation which is invariant to laser source, angle of incidence, range and robot surveying motion. Such property eliminates the need of the post-factory reflectivity calibration whose time requirements are infeasible in mass produced robots/vehicles. Our experiments demonstrate that we can achieve better performance than state of the art on ground reflectivity inference-map based localization at no additional computational burden.
|
2304.05687
|
Rock Yuren Pang
|
Rock Yuren Pang, Katharina Reinecke
|
Anticipating Unintended Consequences of Technology Using Insights from
Creativity Support Tools
|
In CHI '23 Workshop on Designing Technology and Policy
Simultaneously: Towards A Research Agenda and New Practice, April 23, 2023
| null | null | null |
cs.HC
|
http://creativecommons.org/licenses/by/4.0/
|
Our society has been increasingly witnessing a number of negative, unintended
consequences of digital technologies. While post-hoc policy regulation is
crucial in addressing these issues, reasonably anticipating the consequences
before deploying technology can help mitigate potential harm to society in the
first place. Yet, the quest to anticipate potential harms can be difficult
without seeing digital technologies deployed in the real world. In this
position paper, we argue that anticipating unintended consequences of
technology can be facilitated through creativity-enhancing interventions, such
as by building on existing knowledge and insights from diverse stakeholders.
Using lessons learned from prior work on creativity-support tools, the HCI
community is uniquely equipped to design novel systems that aid in anticipating
negative unintended consequences of technology on society.
|
[
{
"created": "Wed, 12 Apr 2023 08:25:22 GMT",
"version": "v1"
}
] |
2023-04-13
|
[
[
"Pang",
"Rock Yuren",
""
],
[
"Reinecke",
"Katharina",
""
]
] |
Our society has been increasingly witnessing a number of negative, unintended consequences of digital technologies. While post-hoc policy regulation is crucial in addressing these issues, reasonably anticipating the consequences before deploying technology can help mitigate potential harm to society in the first place. Yet, the quest to anticipate potential harms can be difficult without seeing digital technologies deployed in the real world. In this position paper, we argue that anticipating unintended consequences of technology can be facilitated through creativity-enhancing interventions, such as by building on existing knowledge and insights from diverse stakeholders. Using lessons learned from prior work on creativity-support tools, the HCI community is uniquely equipped to design novel systems that aid in anticipating negative unintended consequences of technology on society.
|
1602.05181
|
Arindam Biswas
|
Arindam Biswas
|
A Simple Condition for the Existence of Transversals
| null | null | null | null |
cs.DM math.CO
|
http://creativecommons.org/licenses/by-sa/4.0/
|
Hall's Theorem is a basic result in Combinatorics which states that the
obvious necesssary condition for a finite family of sets to have a transversal
is also sufficient. We present a sufficient (but not necessary) condition on
the sizes of the sets in the family and the sizes of their intersections so
that a transversal exists. Using this, we prove that in a bipartite graph $G$
(bipartition $\{A, B\}$), without 4-cycles, if $\deg(v) \geq \sqrt{2e|A|}$ for
all $v \in A$, then $G$ has a matching of size $|A|$.
|
[
{
"created": "Tue, 16 Feb 2016 20:54:24 GMT",
"version": "v1"
}
] |
2016-02-17
|
[
[
"Biswas",
"Arindam",
""
]
] |
Hall's Theorem is a basic result in Combinatorics which states that the obvious necesssary condition for a finite family of sets to have a transversal is also sufficient. We present a sufficient (but not necessary) condition on the sizes of the sets in the family and the sizes of their intersections so that a transversal exists. Using this, we prove that in a bipartite graph $G$ (bipartition $\{A, B\}$), without 4-cycles, if $\deg(v) \geq \sqrt{2e|A|}$ for all $v \in A$, then $G$ has a matching of size $|A|$.
|
1910.00294
|
Yunsu Kim
|
Yunsu Kim, Duc Thanh Tran, Hermann Ney
|
When and Why is Document-level Context Useful in Neural Machine
Translation?
|
DiscoMT 2019 camera-ready
| null | null | null |
cs.CL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Document-level context has received lots of attention for compensating neural
machine translation (NMT) of isolated sentences. However, recent advances in
document-level NMT focus on sophisticated integration of the context,
explaining its improvement with only a few selected examples or targeted test
sets. We extensively quantify the causes of improvements by a document-level
model in general test sets, clarifying the limit of the usefulness of
document-level context in NMT. We show that most of the improvements are not
interpretable as utilizing the context. We also show that a minimal encoding is
sufficient for the context modeling and very long context is not helpful for
NMT.
|
[
{
"created": "Tue, 1 Oct 2019 10:40:26 GMT",
"version": "v1"
}
] |
2019-10-02
|
[
[
"Kim",
"Yunsu",
""
],
[
"Tran",
"Duc Thanh",
""
],
[
"Ney",
"Hermann",
""
]
] |
Document-level context has received lots of attention for compensating neural machine translation (NMT) of isolated sentences. However, recent advances in document-level NMT focus on sophisticated integration of the context, explaining its improvement with only a few selected examples or targeted test sets. We extensively quantify the causes of improvements by a document-level model in general test sets, clarifying the limit of the usefulness of document-level context in NMT. We show that most of the improvements are not interpretable as utilizing the context. We also show that a minimal encoding is sufficient for the context modeling and very long context is not helpful for NMT.
|
2303.10631
|
Peter Kostol\'anyi
|
Peter Kostol\'anyi
|
Bideterministic Weighted Automata
|
This is an extended version of an article published in the
proceedings of the conference CAI 2022
| null | null | null |
cs.FL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
A finite automaton is called bideterministic if it is both deterministic and
codeterministic -- that is, if it is deterministic and its transpose is
deterministic as well. The study of such automata in a weighted setting is
initiated. All trim bideterministic weighted automata over integral domains and
over positive semirings are proved to be minimal. On the contrary, it is
observed that this property does not hold over commutative rings in general:
non-minimal trim bideterministic weighted automata do exist over all semirings
that are not zero-divisor free, and over many such semirings, these automata
might not even admit equivalents that are both minimal and bideterministic. The
problem of determining whether a given rational series is realised by a
bideterministic automaton is shown to be decidable over fields and over
tropical semirings. An example of a positive semiring over which this problem
becomes undecidable is given as well.
|
[
{
"created": "Sun, 19 Mar 2023 11:22:34 GMT",
"version": "v1"
},
{
"created": "Wed, 17 May 2023 15:24:39 GMT",
"version": "v2"
},
{
"created": "Fri, 29 Sep 2023 14:08:04 GMT",
"version": "v3"
}
] |
2023-10-02
|
[
[
"Kostolányi",
"Peter",
""
]
] |
A finite automaton is called bideterministic if it is both deterministic and codeterministic -- that is, if it is deterministic and its transpose is deterministic as well. The study of such automata in a weighted setting is initiated. All trim bideterministic weighted automata over integral domains and over positive semirings are proved to be minimal. On the contrary, it is observed that this property does not hold over commutative rings in general: non-minimal trim bideterministic weighted automata do exist over all semirings that are not zero-divisor free, and over many such semirings, these automata might not even admit equivalents that are both minimal and bideterministic. The problem of determining whether a given rational series is realised by a bideterministic automaton is shown to be decidable over fields and over tropical semirings. An example of a positive semiring over which this problem becomes undecidable is given as well.
|
2404.03899
|
Jinwook Kim
|
Hyunyoung Jang, Jinwook Kim, Jeongmi Lee
|
Effects of Multisensory Feedback on the Perception and Performance of
Virtual Reality Hand-Retargeted Interaction
|
17 pages, 8 figures, 2 tables
| null | null | null |
cs.HC
|
http://creativecommons.org/licenses/by/4.0/
|
Retargeting methods that modify the visual representation of real movements
have been widely used to expand the interaction space and create engaging
virtual reality experiences. For optimal user experience and performance, it is
essential to specify the perception of retargeting and utilize the appropriate
range of modification parameters. However, previous studies mostly concentrated
on whether users perceived the target sense or not and rarely examined the
perceptual accuracy and sensitivity to retargeting. Moreover, it is unknown how
the perception and performance in hand-retargeted interactions are influenced
by multisensory feedback. In this study, we used rigorous psychophysical
methods to specify users' perceptual accuracy and sensitivity to
hand-retargeting and provide acceptable ranges of retargeting parameters. We
also presented different multisensory feedback simultaneously with the
retargeting to probe its effect on users' perception and task performance. The
experimental results showed that providing continuous multisensory feedback,
proportionate to the distance between the virtual hand and the targeted
destination, heightened the accuracy of users' perception of hand retargeting
without altering their perceptual sensitivity. Furthermore, the utilization of
multisensory feedback considerably improved the precision of task performance,
particularly at lower gain factors. Based on these findings, we propose design
guidelines and potential applications of VR hand-retargeted interactions and
multisensory feedback for optimal user experience and performance.
|
[
{
"created": "Fri, 5 Apr 2024 05:44:11 GMT",
"version": "v1"
}
] |
2024-04-08
|
[
[
"Jang",
"Hyunyoung",
""
],
[
"Kim",
"Jinwook",
""
],
[
"Lee",
"Jeongmi",
""
]
] |
Retargeting methods that modify the visual representation of real movements have been widely used to expand the interaction space and create engaging virtual reality experiences. For optimal user experience and performance, it is essential to specify the perception of retargeting and utilize the appropriate range of modification parameters. However, previous studies mostly concentrated on whether users perceived the target sense or not and rarely examined the perceptual accuracy and sensitivity to retargeting. Moreover, it is unknown how the perception and performance in hand-retargeted interactions are influenced by multisensory feedback. In this study, we used rigorous psychophysical methods to specify users' perceptual accuracy and sensitivity to hand-retargeting and provide acceptable ranges of retargeting parameters. We also presented different multisensory feedback simultaneously with the retargeting to probe its effect on users' perception and task performance. The experimental results showed that providing continuous multisensory feedback, proportionate to the distance between the virtual hand and the targeted destination, heightened the accuracy of users' perception of hand retargeting without altering their perceptual sensitivity. Furthermore, the utilization of multisensory feedback considerably improved the precision of task performance, particularly at lower gain factors. Based on these findings, we propose design guidelines and potential applications of VR hand-retargeted interactions and multisensory feedback for optimal user experience and performance.
|
1703.07611
|
Georgios Zogopoulos-Papaliakos
|
Georgios Zogopoulos-Papaliakos, Kostas J. Kyriakopoulos
|
On the Selection of Calculable Residual Generators for UAV Fault
Diagnosis
| null | null |
10.1109/MED.2016.7536003
| null |
cs.SY
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Structural Analysis is an established method for Fault Detection and
Identification (FDI) in large-scale systems, enabling the discovery of
Analytical Redundancy Relations (ARRs) which serve as residual generators.
However, most techniques used to enumerate ARRs do not specify the matching
used to calculate each of those ARRs. This can result in non-implementable or
unusable residual generators, in the presence of non-invertibilities in the
equations involved or in lack of computational tools. In this paper, we propose
a methodology which combines a priori and a posteriori information in order to
reduce the time required to find implementable, usable residual generators of
minimum cost. The method is applied to a fixed-wing Unmanned Aerial Vehicle
(UAV) model.
|
[
{
"created": "Wed, 22 Mar 2017 12:01:09 GMT",
"version": "v1"
}
] |
2017-03-23
|
[
[
"Zogopoulos-Papaliakos",
"Georgios",
""
],
[
"Kyriakopoulos",
"Kostas J.",
""
]
] |
Structural Analysis is an established method for Fault Detection and Identification (FDI) in large-scale systems, enabling the discovery of Analytical Redundancy Relations (ARRs) which serve as residual generators. However, most techniques used to enumerate ARRs do not specify the matching used to calculate each of those ARRs. This can result in non-implementable or unusable residual generators, in the presence of non-invertibilities in the equations involved or in lack of computational tools. In this paper, we propose a methodology which combines a priori and a posteriori information in order to reduce the time required to find implementable, usable residual generators of minimum cost. The method is applied to a fixed-wing Unmanned Aerial Vehicle (UAV) model.
|
1805.04625
|
Shun Watanabe
|
Himanshu Tyagi, Shun Watanabe
|
Strong Converse using Change of Measure Arguments
|
35 pages, no figure; v2 updated references
| null | null | null |
cs.IT math.IT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The strong converse for a coding theorem shows that the optimal asymptotic
rate possible with vanishing error cannot be improved by allowing a fixed
error. Building on a method introduced by Gu and Effros for centralized coding
problems, we develop a general and simple recipe for proving strong converse
that is applicable for distributed problems as well. Heuristically, our proof
of strong converse mimics the standard steps for proving a weak converse,
except that we apply those steps to a modified distribution obtained by
conditioning the original distribution on the event that no error occurs. A key
component of our recipe is the replacement of the hard Markov constraints
implied by the distributed nature of the problem with a soft information cost
using a variational formula introduced by Oohama. We illustrate our method by
providing a short proof of the strong converse for the Wyner-Ziv problem and
strong converse theorems for interactive function computation, common
randomness and secret key agreement, and the wiretap channel; the latter three
strong converse problems were open prior to this work.
|
[
{
"created": "Sat, 12 May 2018 00:34:37 GMT",
"version": "v1"
},
{
"created": "Wed, 21 Aug 2019 14:13:36 GMT",
"version": "v2"
}
] |
2019-08-22
|
[
[
"Tyagi",
"Himanshu",
""
],
[
"Watanabe",
"Shun",
""
]
] |
The strong converse for a coding theorem shows that the optimal asymptotic rate possible with vanishing error cannot be improved by allowing a fixed error. Building on a method introduced by Gu and Effros for centralized coding problems, we develop a general and simple recipe for proving strong converse that is applicable for distributed problems as well. Heuristically, our proof of strong converse mimics the standard steps for proving a weak converse, except that we apply those steps to a modified distribution obtained by conditioning the original distribution on the event that no error occurs. A key component of our recipe is the replacement of the hard Markov constraints implied by the distributed nature of the problem with a soft information cost using a variational formula introduced by Oohama. We illustrate our method by providing a short proof of the strong converse for the Wyner-Ziv problem and strong converse theorems for interactive function computation, common randomness and secret key agreement, and the wiretap channel; the latter three strong converse problems were open prior to this work.
|
1101.3220
|
Andreas Schenk
|
Andreas Schenk and Robert F.H. Fischer
|
Decision-Feedback Differential Detection in Impulse-Radio Ultra-Wideband
Systems
|
Preprint of manuscript accepted for presentation in "IEEE
Transactions on Communications"
| null | null | null |
cs.IT math.IT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this paper we present decision-feedback differential detection (DF-DD)
schemes for autocorrelation-based detection in impulse-radio ultra-wideband
(IR-UWB) systems, a signaling scheme regarded as a promising candidate in
particular for low-complexity wireless sensor networks. To this end, we first
discuss ideal noncoherent sequence estimation and approximations thereof based
on block-wise multiple-symbol differential detection (MSDD) and the Viterbi
algorithm (VA) from the perspective of tree-search/trellis decoding. Exploiting
relations well-known from tree-search decoding, we are able to derive the novel
decision-feedback differential detection (DF-DD) schemes. A comprehensive
comparison with respect to performance and complexity of the presented schemes
in a typical IR-UWB scenario reveals---along with novel insights in techniques
for complexity reduction of the sphere decoder applied for MSDD---that sorted
DF-DD achieves close-to-optimum performance at very low, and in particular
constant receiver complexity.
|
[
{
"created": "Mon, 17 Jan 2011 14:08:26 GMT",
"version": "v1"
}
] |
2011-01-18
|
[
[
"Schenk",
"Andreas",
""
],
[
"Fischer",
"Robert F. H.",
""
]
] |
In this paper we present decision-feedback differential detection (DF-DD) schemes for autocorrelation-based detection in impulse-radio ultra-wideband (IR-UWB) systems, a signaling scheme regarded as a promising candidate in particular for low-complexity wireless sensor networks. To this end, we first discuss ideal noncoherent sequence estimation and approximations thereof based on block-wise multiple-symbol differential detection (MSDD) and the Viterbi algorithm (VA) from the perspective of tree-search/trellis decoding. Exploiting relations well-known from tree-search decoding, we are able to derive the novel decision-feedback differential detection (DF-DD) schemes. A comprehensive comparison with respect to performance and complexity of the presented schemes in a typical IR-UWB scenario reveals---along with novel insights in techniques for complexity reduction of the sphere decoder applied for MSDD---that sorted DF-DD achieves close-to-optimum performance at very low, and in particular constant receiver complexity.
|
1901.01255
|
Tolga Birdal
|
Tolga Birdal and Benjamin Busam and Nassir Navab and Slobodan Ilic and
Peter Sturm
|
Generic Primitive Detection in Point Clouds Using Novel Minimal Quadric
Fits
|
Submitted to IEEE Transactions on Pattern Analysis and Machine
Intelligence (T-PAMI). arXiv admin note: substantial text overlap with
arXiv:1803.07191
| null | null | null |
cs.CV cs.CG cs.GR cs.RO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We present a novel and effective method for detecting 3D primitives in
cluttered, unorganized point clouds, without axillary segmentation or type
specification. We consider the quadric surfaces for encapsulating the basic
building blocks of our environments - planes, spheres, ellipsoids, cones or
cylinders, in a unified fashion. Moreover, quadrics allow us to model higher
degree of freedom shapes, such as hyperboloids or paraboloids that could be
used in non-rigid settings.
We begin by contributing two novel quadric fits targeting 3D point sets that
are endowed with tangent space information. Based upon the idea of aligning the
quadric gradients with the surface normals, our first formulation is exact and
requires as low as four oriented points. The second fit approximates the first,
and reduces the computational effort. We theoretically analyze these fits with
rigor, and give algebraic and geometric arguments. Next, by re-parameterizing
the solution, we devise a new local Hough voting scheme on the null-space
coefficients that is combined with RANSAC, reducing the complexity from
$O(N^4)$ to $O(N^3)$ (three points). To the best of our knowledge, this is the
first method capable of performing a generic cross-type multi-object primitive
detection in difficult scenes without segmentation. Our extensive qualitative
and quantitative results show that our method is efficient and flexible, as
well as being accurate.
|
[
{
"created": "Fri, 4 Jan 2019 12:09:50 GMT",
"version": "v1"
}
] |
2019-01-08
|
[
[
"Birdal",
"Tolga",
""
],
[
"Busam",
"Benjamin",
""
],
[
"Navab",
"Nassir",
""
],
[
"Ilic",
"Slobodan",
""
],
[
"Sturm",
"Peter",
""
]
] |
We present a novel and effective method for detecting 3D primitives in cluttered, unorganized point clouds, without axillary segmentation or type specification. We consider the quadric surfaces for encapsulating the basic building blocks of our environments - planes, spheres, ellipsoids, cones or cylinders, in a unified fashion. Moreover, quadrics allow us to model higher degree of freedom shapes, such as hyperboloids or paraboloids that could be used in non-rigid settings. We begin by contributing two novel quadric fits targeting 3D point sets that are endowed with tangent space information. Based upon the idea of aligning the quadric gradients with the surface normals, our first formulation is exact and requires as low as four oriented points. The second fit approximates the first, and reduces the computational effort. We theoretically analyze these fits with rigor, and give algebraic and geometric arguments. Next, by re-parameterizing the solution, we devise a new local Hough voting scheme on the null-space coefficients that is combined with RANSAC, reducing the complexity from $O(N^4)$ to $O(N^3)$ (three points). To the best of our knowledge, this is the first method capable of performing a generic cross-type multi-object primitive detection in difficult scenes without segmentation. Our extensive qualitative and quantitative results show that our method is efficient and flexible, as well as being accurate.
|
2407.11902
|
Qi Li
|
Qi Li, Runpeng Yu, Xinchao Wang
|
Encapsulating Knowledge in One Prompt
| null | null | null | null |
cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
This paradigm encapsulates knowledge from various models into a solitary
prompt without altering the original models or requiring access to the training
data, which enables us to achieve efficient and convenient knowledge transfer
in more realistic scenarios. From a practicality standpoint, this paradigm not
only for the first time proves the effectiveness of Visual Prompt in data
inaccessible contexts, but also solves the problems of low model reusability
and high storage resource consumption faced by traditional Data-Free Knowledge
Transfer, which means that we can realize the parallel knowledge transfer of
multiple models without modifying any source model. Extensive experiments
across various datasets and models demonstrate the efficacy of the proposed
KiOP knowledge transfer paradigm. Without access to real training data and with
rigorous storage capacity constraints, it is also capable of yielding
considerable outcomes when dealing with cross-model backbone setups and
handling parallel knowledge transfer processing requests with multiple (more
than 2) models.
|
[
{
"created": "Tue, 16 Jul 2024 16:35:23 GMT",
"version": "v1"
}
] |
2024-07-17
|
[
[
"Li",
"Qi",
""
],
[
"Yu",
"Runpeng",
""
],
[
"Wang",
"Xinchao",
""
]
] |
This paradigm encapsulates knowledge from various models into a solitary prompt without altering the original models or requiring access to the training data, which enables us to achieve efficient and convenient knowledge transfer in more realistic scenarios. From a practicality standpoint, this paradigm not only for the first time proves the effectiveness of Visual Prompt in data inaccessible contexts, but also solves the problems of low model reusability and high storage resource consumption faced by traditional Data-Free Knowledge Transfer, which means that we can realize the parallel knowledge transfer of multiple models without modifying any source model. Extensive experiments across various datasets and models demonstrate the efficacy of the proposed KiOP knowledge transfer paradigm. Without access to real training data and with rigorous storage capacity constraints, it is also capable of yielding considerable outcomes when dealing with cross-model backbone setups and handling parallel knowledge transfer processing requests with multiple (more than 2) models.
|
2212.11726
|
David Kuric
|
David Kuric, Herke van Hoof
|
Reusable Options through Gradient-based Meta Learning
|
Published in Transactions on Machine Learning Research (TMLR)
| null | null | null |
cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Hierarchical methods in reinforcement learning have the potential to reduce
the amount of decisions that the agent needs to perform when learning new
tasks. However, finding reusable useful temporal abstractions that facilitate
fast learning remains a challenging problem. Recently, several deep learning
approaches were proposed to learn such temporal abstractions in the form of
options in an end-to-end manner. In this work, we point out several
shortcomings of these methods and discuss their potential negative
consequences. Subsequently, we formulate the desiderata for reusable options
and use these to frame the problem of learning options as a gradient-based
meta-learning problem. This allows us to formulate an objective that explicitly
incentivizes options which allow a higher-level decision maker to adjust in few
steps to different tasks. Experimentally, we show that our method is able to
learn transferable components which accelerate learning and performs better
than existing prior methods developed for this setting. Additionally, we
perform ablations to quantify the impact of using gradient-based meta-learning
as well as other proposed changes.
|
[
{
"created": "Thu, 22 Dec 2022 14:19:35 GMT",
"version": "v1"
},
{
"created": "Tue, 4 Apr 2023 10:46:54 GMT",
"version": "v2"
}
] |
2023-04-05
|
[
[
"Kuric",
"David",
""
],
[
"van Hoof",
"Herke",
""
]
] |
Hierarchical methods in reinforcement learning have the potential to reduce the amount of decisions that the agent needs to perform when learning new tasks. However, finding reusable useful temporal abstractions that facilitate fast learning remains a challenging problem. Recently, several deep learning approaches were proposed to learn such temporal abstractions in the form of options in an end-to-end manner. In this work, we point out several shortcomings of these methods and discuss their potential negative consequences. Subsequently, we formulate the desiderata for reusable options and use these to frame the problem of learning options as a gradient-based meta-learning problem. This allows us to formulate an objective that explicitly incentivizes options which allow a higher-level decision maker to adjust in few steps to different tasks. Experimentally, we show that our method is able to learn transferable components which accelerate learning and performs better than existing prior methods developed for this setting. Additionally, we perform ablations to quantify the impact of using gradient-based meta-learning as well as other proposed changes.
|
1812.07129
|
Ashkan Ebadi
|
Ashkan Ebadi, Patrick J. Tighe, Lei Zhang, Parisa Rashidi
|
Does the Position of Surgical Service Providers in Intra-Operative
Networks Matter? Analyzing the Impact of Influencing Factors on Patients'
Outcome
|
17 pages, 3 Figures, 5 Tables PrePrint
| null | null | null |
cs.SI cs.CY
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We analyzed the relation between surgical service providers' network
structure and surgical team size with patient outcome during the operation. We
did correlation analysis to evaluate the associations among the network
structure measures in the intra-operative networks of surgical service
providers. We focused on intra-operative networks of surgical service
providers, in a quaternary-care academic medical center, using retrospective
Electronic Medical Record (EMR) data. We used de-identified intra-operative
data for adult patients who received nonambulatory/nonobstetric surgery in a
main operating room at Shands at the University of Florida between June 1, 2011
and November 1, 2014. The intra-operative dataset contained 30,211 unique
surgical cases. To perform the analysis, we created the networks of surgical
service providers and calculated several network structure measures at both
team and individual levels. We considered number of patients' complications as
the target variable and assessed its interrelations with the calculated network
measures along with other influencing factors (e.g. surgical team size, type of
surgery). Our results confirm the significant role of interactions among
surgical providers on patient outcome. In addition, we observed that highly
central providers at the global network level are more likely to be associated
with a lower number of surgical complications, while locally important
providers might be associated with higher number of complications. We also
found a positive relation between age of patients and number of complications.
|
[
{
"created": "Tue, 18 Dec 2018 01:36:47 GMT",
"version": "v1"
}
] |
2018-12-19
|
[
[
"Ebadi",
"Ashkan",
""
],
[
"Tighe",
"Patrick J.",
""
],
[
"Zhang",
"Lei",
""
],
[
"Rashidi",
"Parisa",
""
]
] |
We analyzed the relation between surgical service providers' network structure and surgical team size with patient outcome during the operation. We did correlation analysis to evaluate the associations among the network structure measures in the intra-operative networks of surgical service providers. We focused on intra-operative networks of surgical service providers, in a quaternary-care academic medical center, using retrospective Electronic Medical Record (EMR) data. We used de-identified intra-operative data for adult patients who received nonambulatory/nonobstetric surgery in a main operating room at Shands at the University of Florida between June 1, 2011 and November 1, 2014. The intra-operative dataset contained 30,211 unique surgical cases. To perform the analysis, we created the networks of surgical service providers and calculated several network structure measures at both team and individual levels. We considered number of patients' complications as the target variable and assessed its interrelations with the calculated network measures along with other influencing factors (e.g. surgical team size, type of surgery). Our results confirm the significant role of interactions among surgical providers on patient outcome. In addition, we observed that highly central providers at the global network level are more likely to be associated with a lower number of surgical complications, while locally important providers might be associated with higher number of complications. We also found a positive relation between age of patients and number of complications.
|
2107.03506
|
Szymon Talaga
|
Agnieszka Rychwalska, Szymon Talaga, Karolina Ziembowicz, Dariusz
Jemielniak
|
Communication networks and group effectiveness: the case of English
Wikipedia
|
Preprint (before peer-review)
| null | null | null |
cs.SI
|
http://creativecommons.org/licenses/by/4.0/
|
A selection of intellectual goods produced by online communities - e.g. open
source software or knowledge bases like Wikipedia - are in daily use by a broad
audience, and thus their quality impacts the public at large. Yet, it is still
unclear what contributes to the effectiveness of such online peer production
systems: what conditions or social processes help them deliver quality
products. Specifically, while co-contribution (i.e. bipartite networks) are
often investigated in online collaboration, the role of interpersonal
communication in coordination of online peer-production is much less
investigated. To address this gap we have reconstructed networks of personal
communication (direct messaging) between Wikipedia editors gathered in so
called Wikiprojects - teams of contributors who focus on articles within
specific topical areas. We found that effective projects exchange larger volume
of direct messages and that their communication structure allows for complex
coordination: for sharing of information locally through selective ties, and at
the same time globally across the whole group. To verify how these network
measures relate to the subjective perception of importance of group
communication we conducted semi-structured interviews with members of selected
projects. Our interviewees used direct communication for providing feedback,
for maintaining close relations and for tapping on the social capital of the
Wikipedia community. Our results underscore the importance of communication
structure in online collaboration: online peer production communities rely on
interpersonal communication to coordinate their work and to maintain high
levels of engagement. Design of platforms for such communities should allow for
ample group level communication as well as for one-on-one interactions.
|
[
{
"created": "Wed, 7 Jul 2021 22:29:08 GMT",
"version": "v1"
}
] |
2021-07-09
|
[
[
"Rychwalska",
"Agnieszka",
""
],
[
"Talaga",
"Szymon",
""
],
[
"Ziembowicz",
"Karolina",
""
],
[
"Jemielniak",
"Dariusz",
""
]
] |
A selection of intellectual goods produced by online communities - e.g. open source software or knowledge bases like Wikipedia - are in daily use by a broad audience, and thus their quality impacts the public at large. Yet, it is still unclear what contributes to the effectiveness of such online peer production systems: what conditions or social processes help them deliver quality products. Specifically, while co-contribution (i.e. bipartite networks) are often investigated in online collaboration, the role of interpersonal communication in coordination of online peer-production is much less investigated. To address this gap we have reconstructed networks of personal communication (direct messaging) between Wikipedia editors gathered in so called Wikiprojects - teams of contributors who focus on articles within specific topical areas. We found that effective projects exchange larger volume of direct messages and that their communication structure allows for complex coordination: for sharing of information locally through selective ties, and at the same time globally across the whole group. To verify how these network measures relate to the subjective perception of importance of group communication we conducted semi-structured interviews with members of selected projects. Our interviewees used direct communication for providing feedback, for maintaining close relations and for tapping on the social capital of the Wikipedia community. Our results underscore the importance of communication structure in online collaboration: online peer production communities rely on interpersonal communication to coordinate their work and to maintain high levels of engagement. Design of platforms for such communities should allow for ample group level communication as well as for one-on-one interactions.
|
2310.06904
|
Deepti Ghadiyaram
|
Piero Esposito, Parmida Atighehchian, Anastasis Germanidis and Deepti
Ghadiyaram
|
Mitigating stereotypical biases in text to image generative systems
|
4 figures, 8 pages
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
State-of-the-art generative text-to-image models are known to exhibit social
biases and over-represent certain groups like people of perceived lighter skin
tones and men in their outcomes. In this work, we propose a method to mitigate
such biases and ensure that the outcomes are fair across different groups of
people. We do this by finetuning text-to-image models on synthetic data that
varies in perceived skin tones and genders constructed from diverse text
prompts. These text prompts are constructed from multiplicative combinations of
ethnicities, genders, professions, age groups, and so on, resulting in diverse
synthetic data. Our diversity finetuned (DFT) model improves the group fairness
metric by 150% for perceived skin tone and 97.7% for perceived gender. Compared
to baselines, DFT models generate more people with perceived darker skin tone
and more women. To foster open research, we will release all text prompts and
code to generate training images.
|
[
{
"created": "Tue, 10 Oct 2023 18:01:52 GMT",
"version": "v1"
}
] |
2023-10-12
|
[
[
"Esposito",
"Piero",
""
],
[
"Atighehchian",
"Parmida",
""
],
[
"Germanidis",
"Anastasis",
""
],
[
"Ghadiyaram",
"Deepti",
""
]
] |
State-of-the-art generative text-to-image models are known to exhibit social biases and over-represent certain groups like people of perceived lighter skin tones and men in their outcomes. In this work, we propose a method to mitigate such biases and ensure that the outcomes are fair across different groups of people. We do this by finetuning text-to-image models on synthetic data that varies in perceived skin tones and genders constructed from diverse text prompts. These text prompts are constructed from multiplicative combinations of ethnicities, genders, professions, age groups, and so on, resulting in diverse synthetic data. Our diversity finetuned (DFT) model improves the group fairness metric by 150% for perceived skin tone and 97.7% for perceived gender. Compared to baselines, DFT models generate more people with perceived darker skin tone and more women. To foster open research, we will release all text prompts and code to generate training images.
|
2407.11678
|
Luwei Sun
|
Luwei Sun, Dongrui Shen and Han Feng
|
Theoretical Insights into CycleGAN: Analyzing Approximation and
Estimation Errors in Unpaired Data Generation
| null | null | null | null |
cs.LG math.ST stat.ML stat.TH
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this paper, we focus on analyzing the excess risk of the unpaired data
generation model, called CycleGAN. Unlike classical GANs, CycleGAN not only
transforms data between two unpaired distributions but also ensures the
mappings are consistent, which is encouraged by the cycle-consistency term
unique to CycleGAN. The increasing complexity of model structure and the
addition of the cycle-consistency term in CycleGAN present new challenges for
error analysis. By considering the impact of both the model architecture and
training procedure, the risk is decomposed into two terms: approximation error
and estimation error. These two error terms are analyzed separately and
ultimately combined by considering the trade-off between them. Each component
is rigorously analyzed; the approximation error through constructing
approximations of the optimal transport maps, and the estimation error through
establishing an upper bound using Rademacher complexity. Our analysis not only
isolates these errors but also explores the trade-offs between them, which
provides a theoretical insights of how CycleGAN's architecture and training
procedures influence its performance.
|
[
{
"created": "Tue, 16 Jul 2024 12:53:53 GMT",
"version": "v1"
}
] |
2024-07-17
|
[
[
"Sun",
"Luwei",
""
],
[
"Shen",
"Dongrui",
""
],
[
"Feng",
"Han",
""
]
] |
In this paper, we focus on analyzing the excess risk of the unpaired data generation model, called CycleGAN. Unlike classical GANs, CycleGAN not only transforms data between two unpaired distributions but also ensures the mappings are consistent, which is encouraged by the cycle-consistency term unique to CycleGAN. The increasing complexity of model structure and the addition of the cycle-consistency term in CycleGAN present new challenges for error analysis. By considering the impact of both the model architecture and training procedure, the risk is decomposed into two terms: approximation error and estimation error. These two error terms are analyzed separately and ultimately combined by considering the trade-off between them. Each component is rigorously analyzed; the approximation error through constructing approximations of the optimal transport maps, and the estimation error through establishing an upper bound using Rademacher complexity. Our analysis not only isolates these errors but also explores the trade-offs between them, which provides a theoretical insights of how CycleGAN's architecture and training procedures influence its performance.
|
2204.05989
|
Sam Zhang
|
Sam Zhang, K. Hunter Wapman, Daniel B. Larremore, Aaron Clauset
|
Labor advantages drive the greater productivity of faculty at elite
universities
|
22 pages, 11 figures
| null | null | null |
cs.DL
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Faculty at prestigious institutions dominate scientific discourse, with the
small proportion of researchers at elite universities producing a
disproportionate share of all research publications. Environmental prestige is
known to drive such epistemic disparity, but the mechanisms by which it causes
increased faculty productivity remain unknown. Here we combine employment,
publication, and federal survey data for 78,802 tenure-track faculty at 262
PhD-granting institutions in the American university system between 2008--2017
to show through multiple lines of evidence that the greater availability of
funded graduate and postdoctoral labor at more prestigious institutions drives
the environmental effect of prestige on productivity. In particular, we show
that greater environmental prestige leads to larger faculty-led research
groups, which drive higher faculty productivity, primarily in disciplines with
research group collaboration norms. In contrast, we show that productivity does
not increase substantially with prestige for either faculty papers published
without group members, nor group members themselves. The disproportionate
scientific productivity of elite researchers is thus largely explained by their
substantial labor advantage, indicating a more limited role for prestige itself
in predicting scientific contributions.
|
[
{
"created": "Tue, 12 Apr 2022 17:55:09 GMT",
"version": "v1"
}
] |
2022-04-13
|
[
[
"Zhang",
"Sam",
""
],
[
"Wapman",
"K. Hunter",
""
],
[
"Larremore",
"Daniel B.",
""
],
[
"Clauset",
"Aaron",
""
]
] |
Faculty at prestigious institutions dominate scientific discourse, with the small proportion of researchers at elite universities producing a disproportionate share of all research publications. Environmental prestige is known to drive such epistemic disparity, but the mechanisms by which it causes increased faculty productivity remain unknown. Here we combine employment, publication, and federal survey data for 78,802 tenure-track faculty at 262 PhD-granting institutions in the American university system between 2008--2017 to show through multiple lines of evidence that the greater availability of funded graduate and postdoctoral labor at more prestigious institutions drives the environmental effect of prestige on productivity. In particular, we show that greater environmental prestige leads to larger faculty-led research groups, which drive higher faculty productivity, primarily in disciplines with research group collaboration norms. In contrast, we show that productivity does not increase substantially with prestige for either faculty papers published without group members, nor group members themselves. The disproportionate scientific productivity of elite researchers is thus largely explained by their substantial labor advantage, indicating a more limited role for prestige itself in predicting scientific contributions.
|
2010.04466
|
Robert Tjarko Lange
|
Robert Tjarko Lange and Henning Sprekeler
|
Learning Not to Learn: Nature versus Nurture in Silico
| null | null | null | null |
cs.LG cs.AI cs.NE q-bio.NC
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Animals are equipped with a rich innate repertoire of sensory, behavioral and
motor skills, which allows them to interact with the world immediately after
birth. At the same time, many behaviors are highly adaptive and can be tailored
to specific environments by means of learning. In this work, we use
mathematical analysis and the framework of meta-learning (or 'learning to
learn') to answer when it is beneficial to learn such an adaptive strategy and
when to hard-code a heuristic behavior. We find that the interplay of
ecological uncertainty, task complexity and the agents' lifetime has crucial
effects on the meta-learned amortized Bayesian inference performed by an agent.
There exist two regimes: One in which meta-learning yields a learning algorithm
that implements task-dependent information-integration and a second regime in
which meta-learning imprints a heuristic or 'hard-coded' behavior. Further
analysis reveals that non-adaptive behaviors are not only optimal for aspects
of the environment that are stable across individuals, but also in situations
where an adaptation to the environment would in fact be highly beneficial, but
could not be done quickly enough to be exploited within the remaining lifetime.
Hard-coded behaviors should hence not only be those that always work, but also
those that are too complex to be learned within a reasonable time frame.
|
[
{
"created": "Fri, 9 Oct 2020 09:47:40 GMT",
"version": "v1"
},
{
"created": "Thu, 4 Mar 2021 11:27:16 GMT",
"version": "v2"
},
{
"created": "Sun, 1 May 2022 08:38:27 GMT",
"version": "v3"
}
] |
2022-05-03
|
[
[
"Lange",
"Robert Tjarko",
""
],
[
"Sprekeler",
"Henning",
""
]
] |
Animals are equipped with a rich innate repertoire of sensory, behavioral and motor skills, which allows them to interact with the world immediately after birth. At the same time, many behaviors are highly adaptive and can be tailored to specific environments by means of learning. In this work, we use mathematical analysis and the framework of meta-learning (or 'learning to learn') to answer when it is beneficial to learn such an adaptive strategy and when to hard-code a heuristic behavior. We find that the interplay of ecological uncertainty, task complexity and the agents' lifetime has crucial effects on the meta-learned amortized Bayesian inference performed by an agent. There exist two regimes: One in which meta-learning yields a learning algorithm that implements task-dependent information-integration and a second regime in which meta-learning imprints a heuristic or 'hard-coded' behavior. Further analysis reveals that non-adaptive behaviors are not only optimal for aspects of the environment that are stable across individuals, but also in situations where an adaptation to the environment would in fact be highly beneficial, but could not be done quickly enough to be exploited within the remaining lifetime. Hard-coded behaviors should hence not only be those that always work, but also those that are too complex to be learned within a reasonable time frame.
|
2106.03804
|
Daniel Rebain
|
Daniel Rebain, Ke Li, Vincent Sitzmann, Soroosh Yazdani, Kwang Moo Yi,
Andrea Tagliasacchi
|
Deep Medial Fields
| null | null | null | null |
cs.GR cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Implicit representations of geometry, such as occupancy fields or signed
distance fields (SDF), have recently re-gained popularity in encoding 3D solid
shape in a functional form. In this work, we introduce medial fields: a field
function derived from the medial axis transform (MAT) that makes available
information about the underlying 3D geometry that is immediately useful for a
number of downstream tasks. In particular, the medial field encodes the local
thickness of a 3D shape, and enables O(1) projection of a query point onto the
medial axis. To construct the medial field we require nothing but the SDF of
the shape itself, thus allowing its straightforward incorporation in any
application that relies on signed distance fields. Working in unison with the
O(1) surface projection supported by the SDF, the medial field opens the door
for an entirely new set of efficient, shape-aware operations on implicit
representations. We present three such applications, including a modification
to sphere tracing that renders implicit representations with better convergence
properties, a fast construction method for memory-efficient rigid-body
collision proxies, and an efficient approximation of ambient occlusion that
remains stable with respect to viewpoint variations.
|
[
{
"created": "Mon, 7 Jun 2021 17:15:38 GMT",
"version": "v1"
}
] |
2021-06-08
|
[
[
"Rebain",
"Daniel",
""
],
[
"Li",
"Ke",
""
],
[
"Sitzmann",
"Vincent",
""
],
[
"Yazdani",
"Soroosh",
""
],
[
"Yi",
"Kwang Moo",
""
],
[
"Tagliasacchi",
"Andrea",
""
]
] |
Implicit representations of geometry, such as occupancy fields or signed distance fields (SDF), have recently re-gained popularity in encoding 3D solid shape in a functional form. In this work, we introduce medial fields: a field function derived from the medial axis transform (MAT) that makes available information about the underlying 3D geometry that is immediately useful for a number of downstream tasks. In particular, the medial field encodes the local thickness of a 3D shape, and enables O(1) projection of a query point onto the medial axis. To construct the medial field we require nothing but the SDF of the shape itself, thus allowing its straightforward incorporation in any application that relies on signed distance fields. Working in unison with the O(1) surface projection supported by the SDF, the medial field opens the door for an entirely new set of efficient, shape-aware operations on implicit representations. We present three such applications, including a modification to sphere tracing that renders implicit representations with better convergence properties, a fast construction method for memory-efficient rigid-body collision proxies, and an efficient approximation of ambient occlusion that remains stable with respect to viewpoint variations.
|
2206.08257
|
Amirhossein Reisizadeh
|
Romain Cosson, Ali Jadbabaie, Anuran Makur, Amirhossein Reisizadeh,
Devavrat Shah
|
Gradient Descent for Low-Rank Functions
|
26 pages, 2 figures
| null | null | null |
cs.LG math.OC
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Several recent empirical studies demonstrate that important machine learning
tasks, e.g., training deep neural networks, exhibit low-rank structure, where
the loss function varies significantly in only a few directions of the input
space. In this paper, we leverage such low-rank structure to reduce the high
computational cost of canonical gradient-based methods such as gradient descent
(GD). Our proposed \emph{Low-Rank Gradient Descent} (LRGD) algorithm finds an
$\epsilon$-approximate stationary point of a $p$-dimensional function by first
identifying $r \leq p$ significant directions, and then estimating the true
$p$-dimensional gradient at every iteration by computing directional
derivatives only along those $r$ directions. We establish that the "directional
oracle complexities" of LRGD for strongly convex and non-convex objective
functions are $\mathcal{O}(r \log(1/\epsilon) + rp)$ and
$\mathcal{O}(r/\epsilon^2 + rp)$, respectively. When $r \ll p$, these
complexities are smaller than the known complexities of $\mathcal{O}(p
\log(1/\epsilon))$ and $\mathcal{O}(p/\epsilon^2)$ of {\gd} in the strongly
convex and non-convex settings, respectively. Thus, LRGD significantly reduces
the computational cost of gradient-based methods for sufficiently low-rank
functions. In the course of our analysis, we also formally define and
characterize the classes of exact and approximately low-rank functions.
|
[
{
"created": "Thu, 16 Jun 2022 15:58:05 GMT",
"version": "v1"
}
] |
2022-06-17
|
[
[
"Cosson",
"Romain",
""
],
[
"Jadbabaie",
"Ali",
""
],
[
"Makur",
"Anuran",
""
],
[
"Reisizadeh",
"Amirhossein",
""
],
[
"Shah",
"Devavrat",
""
]
] |
Several recent empirical studies demonstrate that important machine learning tasks, e.g., training deep neural networks, exhibit low-rank structure, where the loss function varies significantly in only a few directions of the input space. In this paper, we leverage such low-rank structure to reduce the high computational cost of canonical gradient-based methods such as gradient descent (GD). Our proposed \emph{Low-Rank Gradient Descent} (LRGD) algorithm finds an $\epsilon$-approximate stationary point of a $p$-dimensional function by first identifying $r \leq p$ significant directions, and then estimating the true $p$-dimensional gradient at every iteration by computing directional derivatives only along those $r$ directions. We establish that the "directional oracle complexities" of LRGD for strongly convex and non-convex objective functions are $\mathcal{O}(r \log(1/\epsilon) + rp)$ and $\mathcal{O}(r/\epsilon^2 + rp)$, respectively. When $r \ll p$, these complexities are smaller than the known complexities of $\mathcal{O}(p \log(1/\epsilon))$ and $\mathcal{O}(p/\epsilon^2)$ of {\gd} in the strongly convex and non-convex settings, respectively. Thus, LRGD significantly reduces the computational cost of gradient-based methods for sufficiently low-rank functions. In the course of our analysis, we also formally define and characterize the classes of exact and approximately low-rank functions.
|
1705.10664
|
Jiaji Zhou
|
Jiaji Zhou, J. Andrew Bagnell and Matthew T. Mason
|
A Fast Stochastic Contact Model for Planar Pushing and Grasping: Theory
and Experimental Validation
|
Robotics: Science and Systems 2017
| null | null | null |
cs.RO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Based on the convex force-motion polynomial model for quasi-static sliding,
we derive the kinematic contact model to determine the contact modes and
instantaneous object motion on a supporting surface given a position controlled
manipulator. The inherently stochastic object-to-surface friction distribution
is modelled by sampling physically consistent parameters from appropriate
distributions, with only one parameter to control the amount of noise. Thanks
to the high fidelity and smoothness of convex polynomial models, the mechanics
of patch contact is captured while being computationally efficient without mode
selection at support points. The motion equations for both single and multiple
frictional contacts are given. Simulation based on the model is validated with
robotic pushing and grasping experiments.
|
[
{
"created": "Tue, 30 May 2017 14:21:28 GMT",
"version": "v1"
}
] |
2017-05-31
|
[
[
"Zhou",
"Jiaji",
""
],
[
"Bagnell",
"J. Andrew",
""
],
[
"Mason",
"Matthew T.",
""
]
] |
Based on the convex force-motion polynomial model for quasi-static sliding, we derive the kinematic contact model to determine the contact modes and instantaneous object motion on a supporting surface given a position controlled manipulator. The inherently stochastic object-to-surface friction distribution is modelled by sampling physically consistent parameters from appropriate distributions, with only one parameter to control the amount of noise. Thanks to the high fidelity and smoothness of convex polynomial models, the mechanics of patch contact is captured while being computationally efficient without mode selection at support points. The motion equations for both single and multiple frictional contacts are given. Simulation based on the model is validated with robotic pushing and grasping experiments.
|
2008.04717
|
Lukas Holzbaur
|
Lukas Holzbaur, Rina Polyanskaya, Nikita Polyanskii, Ilya Vorobyev and
Eitan Yaakobi
|
Lifted Multiplicity Codes
| null | null | null | null |
cs.IT cs.DM math.IT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Lifted Reed-Solomon codes and multiplicity codes are two classes of
evaluation codes that allow for the design of high-rate codes that can recover
every codeword or information symbol from many disjoint sets. Recently, the
underlying approaches have been combined to construct lifted bi-variate
multiplicity codes, that can further improve on the rate. We continue the study
of these codes by providing lower bounds on the rate and distance for lifted
multiplicity codes obtained from polynomials in an arbitrary number of
variables. Specifically, we investigate a subcode of a lifted multiplicity code
formed by the linear span of $m$-variate monomials whose restriction to an
arbitrary line in $\mathbb{F}_q^m$ is equivalent to a low-degree uni-variate
polynomial. We find the tight asymptotic behavior of the fraction of such
monomials when the number of variables $m$ is fixed and the alphabet size
$q=2^\ell$ is large. For some parameter regimes, lifted multiplicity codes are
then shown to have a better trade-off between redundancy and the number of
disjoint recovering sets for every codeword or information symbol than
previously known constructions. Additionally, we present a local
self-correction algorithm for lifted multiplicity codes.
|
[
{
"created": "Tue, 11 Aug 2020 14:24:52 GMT",
"version": "v1"
},
{
"created": "Thu, 29 Oct 2020 14:41:53 GMT",
"version": "v2"
}
] |
2020-10-30
|
[
[
"Holzbaur",
"Lukas",
""
],
[
"Polyanskaya",
"Rina",
""
],
[
"Polyanskii",
"Nikita",
""
],
[
"Vorobyev",
"Ilya",
""
],
[
"Yaakobi",
"Eitan",
""
]
] |
Lifted Reed-Solomon codes and multiplicity codes are two classes of evaluation codes that allow for the design of high-rate codes that can recover every codeword or information symbol from many disjoint sets. Recently, the underlying approaches have been combined to construct lifted bi-variate multiplicity codes, that can further improve on the rate. We continue the study of these codes by providing lower bounds on the rate and distance for lifted multiplicity codes obtained from polynomials in an arbitrary number of variables. Specifically, we investigate a subcode of a lifted multiplicity code formed by the linear span of $m$-variate monomials whose restriction to an arbitrary line in $\mathbb{F}_q^m$ is equivalent to a low-degree uni-variate polynomial. We find the tight asymptotic behavior of the fraction of such monomials when the number of variables $m$ is fixed and the alphabet size $q=2^\ell$ is large. For some parameter regimes, lifted multiplicity codes are then shown to have a better trade-off between redundancy and the number of disjoint recovering sets for every codeword or information symbol than previously known constructions. Additionally, we present a local self-correction algorithm for lifted multiplicity codes.
|
1909.05663
|
Riccardo La Grassa
|
Ignazio Gallo, Shah Nawaz, Alessandro Calefati, Riccardo La Grassa,
Nicola Landro
|
Picture What you Read
|
7 pages, Dicta2019 conference
| null | null | null |
cs.CV cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
Visualization refers to our ability to create an image in our head based on
the text we read or the words we hear. It is one of the many skills that makes
reading comprehension possible. Convolutional Neural Networks (CNN) are an
excellent tool for recognizing and classifying text documents. In addition, it
can generate images conditioned on natural language. In this work, we utilize
CNNs capabilities to generate realistic images representative of the text
illustrating the semantic concept. We conducted various experiments to
highlight the capacity of the proposed model to generate representative images
of the text descriptions used as input to the proposed model.
|
[
{
"created": "Mon, 9 Sep 2019 11:26:35 GMT",
"version": "v1"
}
] |
2019-09-13
|
[
[
"Gallo",
"Ignazio",
""
],
[
"Nawaz",
"Shah",
""
],
[
"Calefati",
"Alessandro",
""
],
[
"La Grassa",
"Riccardo",
""
],
[
"Landro",
"Nicola",
""
]
] |
Visualization refers to our ability to create an image in our head based on the text we read or the words we hear. It is one of the many skills that makes reading comprehension possible. Convolutional Neural Networks (CNN) are an excellent tool for recognizing and classifying text documents. In addition, it can generate images conditioned on natural language. In this work, we utilize CNNs capabilities to generate realistic images representative of the text illustrating the semantic concept. We conducted various experiments to highlight the capacity of the proposed model to generate representative images of the text descriptions used as input to the proposed model.
|
1606.05830
|
Cesar Cadena
|
Cesar Cadena, Luca Carlone, Henry Carrillo, Yasir Latif, Davide
Scaramuzza, Jose Neira, Ian Reid, John J. Leonard
|
Past, Present, and Future of Simultaneous Localization And Mapping:
Towards the Robust-Perception Age
| null |
IEEE Transactions on Robotics 32 (6) pp 1309-1332, 2016
|
10.1109/TRO.2016.2624754
| null |
cs.RO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Simultaneous Localization and Mapping (SLAM)consists in the concurrent
construction of a model of the environment (the map), and the estimation of the
state of the robot moving within it. The SLAM community has made astonishing
progress over the last 30 years, enabling large-scale real-world applications,
and witnessing a steady transition of this technology to industry. We survey
the current state of SLAM. We start by presenting what is now the de-facto
standard formulation for SLAM. We then review related work, covering a broad
set of topics including robustness and scalability in long-term mapping, metric
and semantic representations for mapping, theoretical performance guarantees,
active SLAM and exploration, and other new frontiers. This paper simultaneously
serves as a position paper and tutorial to those who are users of SLAM. By
looking at the published research with a critical eye, we delineate open
challenges and new research issues, that still deserve careful scientific
investigation. The paper also contains the authors' take on two questions that
often animate discussions during robotics conferences: Do robots need SLAM? and
Is SLAM solved?
|
[
{
"created": "Sun, 19 Jun 2016 03:23:53 GMT",
"version": "v1"
},
{
"created": "Wed, 20 Jul 2016 06:32:40 GMT",
"version": "v2"
},
{
"created": "Wed, 23 Nov 2016 15:07:09 GMT",
"version": "v3"
},
{
"created": "Mon, 30 Jan 2017 12:05:48 GMT",
"version": "v4"
}
] |
2017-01-31
|
[
[
"Cadena",
"Cesar",
""
],
[
"Carlone",
"Luca",
""
],
[
"Carrillo",
"Henry",
""
],
[
"Latif",
"Yasir",
""
],
[
"Scaramuzza",
"Davide",
""
],
[
"Neira",
"Jose",
""
],
[
"Reid",
"Ian",
""
],
[
"Leonard",
"John J.",
""
]
] |
Simultaneous Localization and Mapping (SLAM)consists in the concurrent construction of a model of the environment (the map), and the estimation of the state of the robot moving within it. The SLAM community has made astonishing progress over the last 30 years, enabling large-scale real-world applications, and witnessing a steady transition of this technology to industry. We survey the current state of SLAM. We start by presenting what is now the de-facto standard formulation for SLAM. We then review related work, covering a broad set of topics including robustness and scalability in long-term mapping, metric and semantic representations for mapping, theoretical performance guarantees, active SLAM and exploration, and other new frontiers. This paper simultaneously serves as a position paper and tutorial to those who are users of SLAM. By looking at the published research with a critical eye, we delineate open challenges and new research issues, that still deserve careful scientific investigation. The paper also contains the authors' take on two questions that often animate discussions during robotics conferences: Do robots need SLAM? and Is SLAM solved?
|
1802.07440
|
Telikepalli Kavitha
|
Telikepalli Kavitha
|
Max-size popular matchings and extensions
|
26 pages, 10 figures
| null | null | null |
cs.DS
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We consider the max-size popular matching problem in a roommates instance G =
(V,E) with strict preference lists. A matching M is popular if there is no
matching M' in G such that the vertices that prefer M' to M outnumber those
that prefer M to M'. We show it is NP-hard to compute a max-size popular
matching in G. This is in contrast to the tractability of this problem in
bipartite graphs where a max-size popular matching can be computed in linear
time. We define a subclass of max-size popular matchings called strongly
dominant matchings and show a linear time algorithm to solve the strongly
dominant matching problem in a roommates instance.
We consider a generalization of the max-size popular matching problem in
bipartite graphs: this is the max-weight popular matching problem where there
is also an edge weight function w and we seek a popular matching of largest
weight. We show this is an NP-hard problem and this is so even when w(e) is
either 1 or 2 for every edge e. We also show an algorithm with running time
O*(2^{n/4}) to find a max-weight popular matching matching in G = (A U B,E)$ on
n vertices.
|
[
{
"created": "Wed, 21 Feb 2018 06:43:21 GMT",
"version": "v1"
}
] |
2018-02-22
|
[
[
"Kavitha",
"Telikepalli",
""
]
] |
We consider the max-size popular matching problem in a roommates instance G = (V,E) with strict preference lists. A matching M is popular if there is no matching M' in G such that the vertices that prefer M' to M outnumber those that prefer M to M'. We show it is NP-hard to compute a max-size popular matching in G. This is in contrast to the tractability of this problem in bipartite graphs where a max-size popular matching can be computed in linear time. We define a subclass of max-size popular matchings called strongly dominant matchings and show a linear time algorithm to solve the strongly dominant matching problem in a roommates instance. We consider a generalization of the max-size popular matching problem in bipartite graphs: this is the max-weight popular matching problem where there is also an edge weight function w and we seek a popular matching of largest weight. We show this is an NP-hard problem and this is so even when w(e) is either 1 or 2 for every edge e. We also show an algorithm with running time O*(2^{n/4}) to find a max-weight popular matching matching in G = (A U B,E)$ on n vertices.
|
cs/0608035
|
Kohei Suenaga
|
Naoki Kobayashi, Kohei Suenaga, and Lucian Wischik
|
Resource Usage Analysis for the Pi-Calculus
| null |
Logical Methods in Computer Science, Volume 2, Issue 3 (September
13, 2006) lmcs:2246
|
10.2168/LMCS-2(3:4)2006
| null |
cs.PL cs.LO
| null |
We propose a type-based resource usage analysis for the π-calculus
extended with resource creation/access primitives. The goal of the resource
usage analysis is to statically check that a program accesses resources such as
files and memory in a valid manner. Our type system is an extension of previous
behavioral type systems for the π-calculus, and can guarantee the safety
property that no invalid access is performed, as well as the property that
necessary accesses (such as the close operation for a file) are eventually
performed unless the program diverges. A sound type inference algorithm for the
type system is also developed to free the programmer from the burden of writing
complex type annotations. Based on the algorithm, we have implemented a
prototype resource usage analyzer for the π-calculus. To the authors'
knowledge, ours is the first type-based resource usage analysis that deals with
an expressive concurrent language like the pi-calculus.
|
[
{
"created": "Mon, 7 Aug 2006 05:22:42 GMT",
"version": "v1"
},
{
"created": "Wed, 13 Sep 2006 11:05:09 GMT",
"version": "v2"
}
] |
2017-01-11
|
[
[
"Kobayashi",
"Naoki",
""
],
[
"Suenaga",
"Kohei",
""
],
[
"Wischik",
"Lucian",
""
]
] |
We propose a type-based resource usage analysis for the π-calculus extended with resource creation/access primitives. The goal of the resource usage analysis is to statically check that a program accesses resources such as files and memory in a valid manner. Our type system is an extension of previous behavioral type systems for the π-calculus, and can guarantee the safety property that no invalid access is performed, as well as the property that necessary accesses (such as the close operation for a file) are eventually performed unless the program diverges. A sound type inference algorithm for the type system is also developed to free the programmer from the burden of writing complex type annotations. Based on the algorithm, we have implemented a prototype resource usage analyzer for the π-calculus. To the authors' knowledge, ours is the first type-based resource usage analysis that deals with an expressive concurrent language like the pi-calculus.
|
2107.09283
|
Hyunji Chung
|
Jungheum Park, Hyunji Chung
|
Toward Trustworthy Urban IT Systems: The Bright and Dark Sides of Smart
City Development
|
1 figure
| null | null | null |
cs.CY
|
http://creativecommons.org/publicdomain/zero/1.0/
|
In smart cities built on information and communication technology, citizens
and various IT systems interoperate in harmony. Cloud computing and
Internet-of-Things technologies that have been developed for a long time are
making modern cities smarter. Smart cities can have a positive impact on
citizens, but they can also make cities dangerous. Today, with the emerging
reality of smart cities, this paper looks at both the bright and dark sides and
provides a foundation for supporting work-related tasks of IT professionals as
well as non-IT experts involved in urban design and development.
|
[
{
"created": "Tue, 20 Jul 2021 06:54:08 GMT",
"version": "v1"
}
] |
2021-07-21
|
[
[
"Park",
"Jungheum",
""
],
[
"Chung",
"Hyunji",
""
]
] |
In smart cities built on information and communication technology, citizens and various IT systems interoperate in harmony. Cloud computing and Internet-of-Things technologies that have been developed for a long time are making modern cities smarter. Smart cities can have a positive impact on citizens, but they can also make cities dangerous. Today, with the emerging reality of smart cities, this paper looks at both the bright and dark sides and provides a foundation for supporting work-related tasks of IT professionals as well as non-IT experts involved in urban design and development.
|
1612.09134
|
Antonio Manuel Lopez Pe\~na
|
Antonio M. Lopez, Jiaolong Xu, Jose L. Gomez, David Vazquez, German
Ros
|
From Virtual to Real World Visual Perception using Domain Adaptation --
The DPM as Example
|
Invited book chapter to appear in "Domain Adaptation in Computer
Vision Applications", Springer Series: Advances in Computer Vision and
Pattern Recognition, Edited by Gabriela Csurka
| null | null | null |
cs.CV cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Supervised learning tends to produce more accurate classifiers than
unsupervised learning in general. This implies that training data is preferred
with annotations. When addressing visual perception challenges, such as
localizing certain object classes within an image, the learning of the involved
classifiers turns out to be a practical bottleneck. The reason is that, at
least, we have to frame object examples with bounding boxes in thousands of
images. A priori, the more complex the model is regarding its number of
parameters, the more annotated examples are required. This annotation task is
performed by human oracles, which ends up in inaccuracies and errors in the
annotations (aka ground truth) since the task is inherently very cumbersome and
sometimes ambiguous. As an alternative we have pioneered the use of virtual
worlds for collecting such annotations automatically and with high precision.
However, since the models learned with virtual data must operate in the real
world, we still need to perform domain adaptation (DA). In this chapter we
revisit the DA of a deformable part-based model (DPM) as an exemplifying case
of virtual- to-real-world DA. As a use case, we address the challenge of
vehicle detection for driver assistance, using different publicly available
virtual-world data. While doing so, we investigate questions such as: how does
the domain gap behave due to virtual-vs-real data with respect to dominant
object appearance per domain, as well as the role of photo-realism in the
virtual world.
|
[
{
"created": "Thu, 29 Dec 2016 13:16:22 GMT",
"version": "v1"
}
] |
2016-12-30
|
[
[
"Lopez",
"Antonio M.",
""
],
[
"Xu",
"Jiaolong",
""
],
[
"Gomez",
"Jose L.",
""
],
[
"Vazquez",
"David",
""
],
[
"Ros",
"German",
""
]
] |
Supervised learning tends to produce more accurate classifiers than unsupervised learning in general. This implies that training data is preferred with annotations. When addressing visual perception challenges, such as localizing certain object classes within an image, the learning of the involved classifiers turns out to be a practical bottleneck. The reason is that, at least, we have to frame object examples with bounding boxes in thousands of images. A priori, the more complex the model is regarding its number of parameters, the more annotated examples are required. This annotation task is performed by human oracles, which ends up in inaccuracies and errors in the annotations (aka ground truth) since the task is inherently very cumbersome and sometimes ambiguous. As an alternative we have pioneered the use of virtual worlds for collecting such annotations automatically and with high precision. However, since the models learned with virtual data must operate in the real world, we still need to perform domain adaptation (DA). In this chapter we revisit the DA of a deformable part-based model (DPM) as an exemplifying case of virtual- to-real-world DA. As a use case, we address the challenge of vehicle detection for driver assistance, using different publicly available virtual-world data. While doing so, we investigate questions such as: how does the domain gap behave due to virtual-vs-real data with respect to dominant object appearance per domain, as well as the role of photo-realism in the virtual world.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.