id
stringlengths 9
16
| submitter
stringlengths 3
64
⌀ | authors
stringlengths 5
6.63k
| title
stringlengths 7
245
| comments
stringlengths 1
482
⌀ | journal-ref
stringlengths 4
382
⌀ | doi
stringlengths 9
151
⌀ | report-no
stringclasses 984
values | categories
stringlengths 5
108
| license
stringclasses 9
values | abstract
stringlengths 83
3.41k
| versions
listlengths 1
20
| update_date
timestamp[s]date 2007-05-23 00:00:00
2025-04-11 00:00:00
| authors_parsed
sequencelengths 1
427
| prompt
stringlengths 166
3.49k
| label
stringclasses 2
values | prob
float64 0.5
0.98
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1612.00534 | Bo Li | Bo Li, Tianfu Wu, Shuai Shao, Lun Zhang and Rufeng Chu | Object Detection via Aspect Ratio and Context Aware Region-based
Convolutional Networks | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Jointly integrating aspect ratio and context has been extensively studied and
shown performance improvement in traditional object detection systems such as
the DPMs. It, however, has been largely ignored in deep neural network based
detection systems. This paper presents a method of integrating a mixture of
object models and region-based convolutional networks for accurate object
detection. Each mixture component accounts for both object aspect ratio and
multi-scale contextual information explicitly: (i) it exploits a mixture of
tiling configurations in the RoI pooling to remedy the warping artifacts caused
by a single type RoI pooling (e.g., with equally-sized 7 x 7 cells), and to
respect the underlying object shapes more; (ii) it "looks from both the inside
and the outside of a RoI" by incorporating contextual information at two
scales: global context pooled from the whole image and local context pooled
from the surrounding of a RoI. To facilitate accurate detection, this paper
proposes a multi-stage detection scheme for integrating the mixture of object
models, which utilizes the detection results of the model at the previous stage
as the proposals for the current in both training and testing. The proposed
method is called the aspect ratio and context aware region-based convolutional
network (ARC-R-CNN). In experiments, ARC-R-CNN shows very competitive results
with Faster R-CNN [41] and R-FCN [10] on two datasets: the PASCAL VOC and the
Microsoft COCO. It obtains significantly better mAP performance using high IoU
thresholds on both datasets.
| [
{
"version": "v1",
"created": "Fri, 2 Dec 2016 01:20:02 GMT"
},
{
"version": "v2",
"created": "Wed, 22 Mar 2017 16:28:24 GMT"
}
] | 2017-03-23T00:00:00 | [
[
"Li",
"Bo",
""
],
[
"Wu",
"Tianfu",
""
],
[
"Shao",
"Shuai",
""
],
[
"Zhang",
"Lun",
""
],
[
"Chu",
"Rufeng",
""
]
] | TITLE: Object Detection via Aspect Ratio and Context Aware Region-based
Convolutional Networks
ABSTRACT: Jointly integrating aspect ratio and context has been extensively studied and
shown performance improvement in traditional object detection systems such as
the DPMs. It, however, has been largely ignored in deep neural network based
detection systems. This paper presents a method of integrating a mixture of
object models and region-based convolutional networks for accurate object
detection. Each mixture component accounts for both object aspect ratio and
multi-scale contextual information explicitly: (i) it exploits a mixture of
tiling configurations in the RoI pooling to remedy the warping artifacts caused
by a single type RoI pooling (e.g., with equally-sized 7 x 7 cells), and to
respect the underlying object shapes more; (ii) it "looks from both the inside
and the outside of a RoI" by incorporating contextual information at two
scales: global context pooled from the whole image and local context pooled
from the surrounding of a RoI. To facilitate accurate detection, this paper
proposes a multi-stage detection scheme for integrating the mixture of object
models, which utilizes the detection results of the model at the previous stage
as the proposals for the current in both training and testing. The proposed
method is called the aspect ratio and context aware region-based convolutional
network (ARC-R-CNN). In experiments, ARC-R-CNN shows very competitive results
with Faster R-CNN [41] and R-FCN [10] on two datasets: the PASCAL VOC and the
Microsoft COCO. It obtains significantly better mAP performance using high IoU
thresholds on both datasets.
| no_new_dataset | 0.954223 |
1702.05060 | Xuanzhe Liu | Xuanzhe Liu, Huoran Li, Xuan Lu, Tao Xie, Qiaozhu Mei, Hong Mei, Feng
Feng | Mining Behavioral Patterns from Millions of Android Users | 29pages | IEEE Transactions on Software Engineering, 2017 | 10.1109/TSE.2017.2685387 | null | cs.CY cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The prevalence of smart mobile devices has promoted the popularity of mobile
applications (a.k.a. apps). Supporting mobility has become a promising trend in
software engineering research. This article presents an empirical study of
behavioral service profiles collected from millions of users whose devices are
deployed with Wandoujia, a leading Android app store service in China. The
dataset of Wandoujia service profiles consists of two kinds of user behavioral
data from using 0.28 million free Android apps, including (1) app management
activities (i.e., downloading, updating, and uninstalling apps) from over 17
million unique users and (2) app network usage from over 6 million unique
users. We explore multiple aspects of such behavioral data and present patterns
of app usage. Based on the findings as well as derived knowledge, we also
suggest some new open opportunities and challenges that can be explored by the
research community, including app development, deployment, delivery, revenue,
etc.
| [
{
"version": "v1",
"created": "Tue, 14 Feb 2017 11:31:13 GMT"
},
{
"version": "v2",
"created": "Wed, 22 Mar 2017 12:07:29 GMT"
}
] | 2017-03-23T00:00:00 | [
[
"Liu",
"Xuanzhe",
""
],
[
"Li",
"Huoran",
""
],
[
"Lu",
"Xuan",
""
],
[
"Xie",
"Tao",
""
],
[
"Mei",
"Qiaozhu",
""
],
[
"Mei",
"Hong",
""
],
[
"Feng",
"Feng",
""
]
] | TITLE: Mining Behavioral Patterns from Millions of Android Users
ABSTRACT: The prevalence of smart mobile devices has promoted the popularity of mobile
applications (a.k.a. apps). Supporting mobility has become a promising trend in
software engineering research. This article presents an empirical study of
behavioral service profiles collected from millions of users whose devices are
deployed with Wandoujia, a leading Android app store service in China. The
dataset of Wandoujia service profiles consists of two kinds of user behavioral
data from using 0.28 million free Android apps, including (1) app management
activities (i.e., downloading, updating, and uninstalling apps) from over 17
million unique users and (2) app network usage from over 6 million unique
users. We explore multiple aspects of such behavioral data and present patterns
of app usage. Based on the findings as well as derived knowledge, we also
suggest some new open opportunities and challenges that can be explored by the
research community, including app development, deployment, delivery, revenue,
etc.
| new_dataset | 0.896433 |
1703.02437 | Santiago Manen | Santiago Manen, Michael Gygli, Dengxin Dai, Luc Van Gool | PathTrack: Fast Trajectory Annotation with Path Supervision | 10 pages, ICCV submission | null | null | null | cs.CV cs.LG cs.MM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Progress in Multiple Object Tracking (MOT) has been historically limited by
the size of the available datasets. We present an efficient framework to
annotate trajectories and use it to produce a MOT dataset of unprecedented
size. In our novel path supervision the annotator loosely follows the object
with the cursor while watching the video, providing a path annotation for each
object in the sequence. Our approach is able to turn such weak annotations into
dense box trajectories. Our experiments on existing datasets prove that our
framework produces more accurate annotations than the state of the art, in a
fraction of the time. We further validate our approach by crowdsourcing the
PathTrack dataset, with more than 15,000 person trajectories in 720 sequences.
Tracking approaches can benefit training on such large-scale datasets, as did
object recognition. We prove this by re-training an off-the-shelf person
matching network, originally trained on the MOT15 dataset, almost halving the
misclassification rate. Additionally, training on our data consistently
improves tracking results, both on our dataset and on MOT15. On the latter, we
improve the top-performing tracker (NOMT) dropping the number of IDSwitches by
18% and fragments by 5%.
| [
{
"version": "v1",
"created": "Tue, 7 Mar 2017 15:36:39 GMT"
},
{
"version": "v2",
"created": "Wed, 22 Mar 2017 07:08:34 GMT"
}
] | 2017-03-23T00:00:00 | [
[
"Manen",
"Santiago",
""
],
[
"Gygli",
"Michael",
""
],
[
"Dai",
"Dengxin",
""
],
[
"Van Gool",
"Luc",
""
]
] | TITLE: PathTrack: Fast Trajectory Annotation with Path Supervision
ABSTRACT: Progress in Multiple Object Tracking (MOT) has been historically limited by
the size of the available datasets. We present an efficient framework to
annotate trajectories and use it to produce a MOT dataset of unprecedented
size. In our novel path supervision the annotator loosely follows the object
with the cursor while watching the video, providing a path annotation for each
object in the sequence. Our approach is able to turn such weak annotations into
dense box trajectories. Our experiments on existing datasets prove that our
framework produces more accurate annotations than the state of the art, in a
fraction of the time. We further validate our approach by crowdsourcing the
PathTrack dataset, with more than 15,000 person trajectories in 720 sequences.
Tracking approaches can benefit training on such large-scale datasets, as did
object recognition. We prove this by re-training an off-the-shelf person
matching network, originally trained on the MOT15 dataset, almost halving the
misclassification rate. Additionally, training on our data consistently
improves tracking results, both on our dataset and on MOT15. On the latter, we
improve the top-performing tracker (NOMT) dropping the number of IDSwitches by
18% and fragments by 5%.
| new_dataset | 0.665574 |
1703.05884 | Ashton Fagg | Hamed Kiani Galoogahi, Ashton Fagg, Chen Huang, Deva Ramanan, Simon
Lucey | Need for Speed: A Benchmark for Higher Frame Rate Object Tracking | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we propose the first higher frame rate video dataset (called
Need for Speed - NfS) and benchmark for visual object tracking. The dataset
consists of 100 videos (380K frames) captured with now commonly available
higher frame rate (240 FPS) cameras from real world scenarios. All frames are
annotated with axis aligned bounding boxes and all sequences are manually
labelled with nine visual attributes - such as occlusion, fast motion,
background clutter, etc. Our benchmark provides an extensive evaluation of many
recent and state-of-the-art trackers on higher frame rate sequences. We ranked
each of these trackers according to their tracking accuracy and real-time
performance. One of our surprising conclusions is that at higher frame rates,
simple trackers such as correlation filters outperform complex methods based on
deep networks. This suggests that for practical applications (such as in
robotics or embedded vision), one needs to carefully tradeoff bandwidth
constraints associated with higher frame rate acquisition, computational costs
of real-time analysis, and the required application accuracy. Our dataset and
benchmark allows for the first time (to our knowledge) systematic exploration
of such issues, and will be made available to allow for further research in
this space.
| [
{
"version": "v1",
"created": "Fri, 17 Mar 2017 04:18:25 GMT"
},
{
"version": "v2",
"created": "Tue, 21 Mar 2017 22:35:09 GMT"
}
] | 2017-03-23T00:00:00 | [
[
"Galoogahi",
"Hamed Kiani",
""
],
[
"Fagg",
"Ashton",
""
],
[
"Huang",
"Chen",
""
],
[
"Ramanan",
"Deva",
""
],
[
"Lucey",
"Simon",
""
]
] | TITLE: Need for Speed: A Benchmark for Higher Frame Rate Object Tracking
ABSTRACT: In this paper, we propose the first higher frame rate video dataset (called
Need for Speed - NfS) and benchmark for visual object tracking. The dataset
consists of 100 videos (380K frames) captured with now commonly available
higher frame rate (240 FPS) cameras from real world scenarios. All frames are
annotated with axis aligned bounding boxes and all sequences are manually
labelled with nine visual attributes - such as occlusion, fast motion,
background clutter, etc. Our benchmark provides an extensive evaluation of many
recent and state-of-the-art trackers on higher frame rate sequences. We ranked
each of these trackers according to their tracking accuracy and real-time
performance. One of our surprising conclusions is that at higher frame rates,
simple trackers such as correlation filters outperform complex methods based on
deep networks. This suggests that for practical applications (such as in
robotics or embedded vision), one needs to carefully tradeoff bandwidth
constraints associated with higher frame rate acquisition, computational costs
of real-time analysis, and the required application accuracy. Our dataset and
benchmark allows for the first time (to our knowledge) systematic exploration
of such issues, and will be made available to allow for further research in
this space.
| new_dataset | 0.967595 |
1703.07255 | Hao Wang | Hao Wang, Xiaodan Liang, Hao Zhang, Dit-Yan Yeung, Eric P. Xing | ZM-Net: Real-time Zero-shot Image Manipulation Network | null | null | null | null | cs.CV cs.AI cs.GR cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Many problems in image processing and computer vision (e.g. colorization,
style transfer) can be posed as 'manipulating' an input image into a
corresponding output image given a user-specified guiding signal. A holy-grail
solution towards generic image manipulation should be able to efficiently alter
an input image with any personalized signals (even signals unseen during
training), such as diverse paintings and arbitrary descriptive attributes.
However, existing methods are either inefficient to simultaneously process
multiple signals (let alone generalize to unseen signals), or unable to handle
signals from other modalities. In this paper, we make the first attempt to
address the zero-shot image manipulation task. We cast this problem as
manipulating an input image according to a parametric model whose key
parameters can be conditionally generated from any guiding signal (even unseen
ones). To this end, we propose the Zero-shot Manipulation Net (ZM-Net), a
fully-differentiable architecture that jointly optimizes an
image-transformation network (TNet) and a parameter network (PNet). The PNet
learns to generate key transformation parameters for the TNet given any guiding
signal while the TNet performs fast zero-shot image manipulation according to
both signal-dependent parameters from the PNet and signal-invariant parameters
from the TNet itself. Extensive experiments show that our ZM-Net can perform
high-quality image manipulation conditioned on different forms of guiding
signals (e.g. style images and attributes) in real-time (tens of milliseconds
per image) even for unseen signals. Moreover, a large-scale style dataset with
over 20,000 style images is also constructed to promote further research.
| [
{
"version": "v1",
"created": "Tue, 21 Mar 2017 15:01:59 GMT"
},
{
"version": "v2",
"created": "Wed, 22 Mar 2017 17:08:40 GMT"
}
] | 2017-03-23T00:00:00 | [
[
"Wang",
"Hao",
""
],
[
"Liang",
"Xiaodan",
""
],
[
"Zhang",
"Hao",
""
],
[
"Yeung",
"Dit-Yan",
""
],
[
"Xing",
"Eric P.",
""
]
] | TITLE: ZM-Net: Real-time Zero-shot Image Manipulation Network
ABSTRACT: Many problems in image processing and computer vision (e.g. colorization,
style transfer) can be posed as 'manipulating' an input image into a
corresponding output image given a user-specified guiding signal. A holy-grail
solution towards generic image manipulation should be able to efficiently alter
an input image with any personalized signals (even signals unseen during
training), such as diverse paintings and arbitrary descriptive attributes.
However, existing methods are either inefficient to simultaneously process
multiple signals (let alone generalize to unseen signals), or unable to handle
signals from other modalities. In this paper, we make the first attempt to
address the zero-shot image manipulation task. We cast this problem as
manipulating an input image according to a parametric model whose key
parameters can be conditionally generated from any guiding signal (even unseen
ones). To this end, we propose the Zero-shot Manipulation Net (ZM-Net), a
fully-differentiable architecture that jointly optimizes an
image-transformation network (TNet) and a parameter network (PNet). The PNet
learns to generate key transformation parameters for the TNet given any guiding
signal while the TNet performs fast zero-shot image manipulation according to
both signal-dependent parameters from the PNet and signal-invariant parameters
from the TNet itself. Extensive experiments show that our ZM-Net can perform
high-quality image manipulation conditioned on different forms of guiding
signals (e.g. style images and attributes) in real-time (tens of milliseconds
per image) even for unseen signals. Moreover, a large-scale style dataset with
over 20,000 style images is also constructed to promote further research.
| no_new_dataset | 0.948728 |
1703.07362 | James Bagrow | James P. Bagrow | Information spreading during emergencies and anomalous events | 19 pages, 11 figures | null | null | null | cs.SI cs.CY physics.data-an physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The most critical time for information to spread is in the aftermath of a
serious emergency, crisis, or disaster. Individuals affected by such situations
can now turn to an array of communication channels, from mobile phone calls and
text messages to social media posts, when alerting social ties. These channels
drastically improve the speed of information in a time-sensitive event, and
provide extant records of human dynamics during and afterward the event.
Retrospective analysis of such anomalous events provides researchers with a
class of "found experiments" that may be used to better understand social
spreading. In this chapter, we study information spreading due to a number of
emergency events, including the Boston Marathon Bombing and a plane crash at a
western European airport. We also contrast the different information which may
be gleaned by social media data compared with mobile phone data and we estimate
the rate of anomalous events in a mobile phone dataset using a proposed anomaly
detection method.
| [
{
"version": "v1",
"created": "Tue, 21 Mar 2017 18:00:07 GMT"
}
] | 2017-03-23T00:00:00 | [
[
"Bagrow",
"James P.",
""
]
] | TITLE: Information spreading during emergencies and anomalous events
ABSTRACT: The most critical time for information to spread is in the aftermath of a
serious emergency, crisis, or disaster. Individuals affected by such situations
can now turn to an array of communication channels, from mobile phone calls and
text messages to social media posts, when alerting social ties. These channels
drastically improve the speed of information in a time-sensitive event, and
provide extant records of human dynamics during and afterward the event.
Retrospective analysis of such anomalous events provides researchers with a
class of "found experiments" that may be used to better understand social
spreading. In this chapter, we study information spreading due to a number of
emergency events, including the Boston Marathon Bombing and a plane crash at a
western European airport. We also contrast the different information which may
be gleaned by social media data compared with mobile phone data and we estimate
the rate of anomalous events in a mobile phone dataset using a proposed anomaly
detection method.
| no_new_dataset | 0.939081 |
1703.07402 | Nicolai Wojke | Nicolai Wojke and Alex Bewley and Dietrich Paulus | Simple Online and Realtime Tracking with a Deep Association Metric | 5 pages, 1 figure | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Simple Online and Realtime Tracking (SORT) is a pragmatic approach to
multiple object tracking with a focus on simple, effective algorithms. In this
paper, we integrate appearance information to improve the performance of SORT.
Due to this extension we are able to track objects through longer periods of
occlusions, effectively reducing the number of identity switches. In spirit of
the original framework we place much of the computational complexity into an
offline pre-training stage where we learn a deep association metric on a
large-scale person re-identification dataset. During online application, we
establish measurement-to-track associations using nearest neighbor queries in
visual appearance space. Experimental evaluation shows that our extensions
reduce the number of identity switches by 45%, achieving overall competitive
performance at high frame rates.
| [
{
"version": "v1",
"created": "Tue, 21 Mar 2017 19:40:25 GMT"
}
] | 2017-03-23T00:00:00 | [
[
"Wojke",
"Nicolai",
""
],
[
"Bewley",
"Alex",
""
],
[
"Paulus",
"Dietrich",
""
]
] | TITLE: Simple Online and Realtime Tracking with a Deep Association Metric
ABSTRACT: Simple Online and Realtime Tracking (SORT) is a pragmatic approach to
multiple object tracking with a focus on simple, effective algorithms. In this
paper, we integrate appearance information to improve the performance of SORT.
Due to this extension we are able to track objects through longer periods of
occlusions, effectively reducing the number of identity switches. In spirit of
the original framework we place much of the computational complexity into an
offline pre-training stage where we learn a deep association metric on a
large-scale person re-identification dataset. During online application, we
establish measurement-to-track associations using nearest neighbor queries in
visual appearance space. Experimental evaluation shows that our extensions
reduce the number of identity switches by 45%, achieving overall competitive
performance at high frame rates.
| no_new_dataset | 0.951142 |
1703.07403 | Kevin Moerman | Andr\'e M.J. Sprengers, Matthan W.A. Caan, Kevin M. Moerman, Aart J.
Nederveen, Rolf M.J.N. Lamerichs, Jaap Stoker | A scale space based algorithm for automated segmentation of single shot
tagged MRI of shearing deformation | null | Magnetic Resonance Materials in Physics, Biology and Medicine,
April 2013, Volume 26, Issue 2, pp 229-238 | 10.1007/s10334-012-0332-9 | null | physics.med-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Object This study proposes a scale space based algorithm for automated
segmentation of single-shot tagged images of modest SNR. Furthermore the
algorithm was designed for analysis of discontinuous or shearing types of
motion, i.e. segmentation of broken tag patterns.
Materials and methods The proposed algorithm utilizes non-linear scale space
for automatic segmentation of single-shot tagged images. The algorithm's
ability to automatically segment tagged shearing motion was evaluated in a
numerical simulation and in vivo. A typical shearing deformation was simulated
in a Shepp-Logan phantom allowing for quantitative evaluation of the
algorithm's success rate as a function of both SNR and the amount of
deformation. For a qualitative in vivo evaluation tagged images showing
deformations in the calf muscles and eye movement in a healthy volunteer were
acquired.
Results Both the numerical simulation and the in vivo tagged data
demonstrated the algorithm's ability for automated segmentation of single-shot
tagged MR provided that SNR of the images is above 10 and the amount of
deformation does not exceed the tag spacing. The latter constraint can be met
by adjusting the tag delay or the tag spacing.
Conclusion The scale space based algorithm for automatic segmentation of
single-shot tagged MR enables the application of tagged MR to complex
(shearing) deformation and the processing of datasets with relatively low SNR.
| [
{
"version": "v1",
"created": "Tue, 21 Mar 2017 19:41:27 GMT"
}
] | 2017-03-23T00:00:00 | [
[
"Sprengers",
"André M. J.",
""
],
[
"Caan",
"Matthan W. A.",
""
],
[
"Moerman",
"Kevin M.",
""
],
[
"Nederveen",
"Aart J.",
""
],
[
"Lamerichs",
"Rolf M. J. N.",
""
],
[
"Stoker",
"Jaap",
""
]
] | TITLE: A scale space based algorithm for automated segmentation of single shot
tagged MRI of shearing deformation
ABSTRACT: Object This study proposes a scale space based algorithm for automated
segmentation of single-shot tagged images of modest SNR. Furthermore the
algorithm was designed for analysis of discontinuous or shearing types of
motion, i.e. segmentation of broken tag patterns.
Materials and methods The proposed algorithm utilizes non-linear scale space
for automatic segmentation of single-shot tagged images. The algorithm's
ability to automatically segment tagged shearing motion was evaluated in a
numerical simulation and in vivo. A typical shearing deformation was simulated
in a Shepp-Logan phantom allowing for quantitative evaluation of the
algorithm's success rate as a function of both SNR and the amount of
deformation. For a qualitative in vivo evaluation tagged images showing
deformations in the calf muscles and eye movement in a healthy volunteer were
acquired.
Results Both the numerical simulation and the in vivo tagged data
demonstrated the algorithm's ability for automated segmentation of single-shot
tagged MR provided that SNR of the images is above 10 and the amount of
deformation does not exceed the tag spacing. The latter constraint can be met
by adjusting the tag delay or the tag spacing.
Conclusion The scale space based algorithm for automatic segmentation of
single-shot tagged MR enables the application of tagged MR to complex
(shearing) deformation and the processing of datasets with relatively low SNR.
| no_new_dataset | 0.952042 |
1703.07473 | Niko S\"underhauf | Feras Dayoub, Niko S\"underhauf, Peter Corke | Episode-Based Active Learning with Bayesian Neural Networks | null | null | null | null | cs.CV cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate different strategies for active learning with Bayesian deep
neural networks. We focus our analysis on scenarios where new, unlabeled data
is obtained episodically, such as commonly encountered in mobile robotics
applications. An evaluation of different strategies for acquisition, updating,
and final training on the CIFAR-10 dataset shows that incremental network
updates with final training on the accumulated acquisition set are essential
for best performance, while limiting the amount of required human labeling
labor.
| [
{
"version": "v1",
"created": "Tue, 21 Mar 2017 23:56:51 GMT"
}
] | 2017-03-23T00:00:00 | [
[
"Dayoub",
"Feras",
""
],
[
"Sünderhauf",
"Niko",
""
],
[
"Corke",
"Peter",
""
]
] | TITLE: Episode-Based Active Learning with Bayesian Neural Networks
ABSTRACT: We investigate different strategies for active learning with Bayesian deep
neural networks. We focus our analysis on scenarios where new, unlabeled data
is obtained episodically, such as commonly encountered in mobile robotics
applications. An evaluation of different strategies for acquisition, updating,
and final training on the CIFAR-10 dataset shows that incremental network
updates with final training on the accumulated acquisition set are essential
for best performance, while limiting the amount of required human labeling
labor.
| no_new_dataset | 0.950273 |
1703.07506 | Marc Goessling | Marc Goessling | LogitBoost autoregressive networks | null | null | 10.1016/j.csda.2017.03.010 | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multivariate binary distributions can be decomposed into products of
univariate conditional distributions. Recently popular approaches have modeled
these conditionals through neural networks with sophisticated weight-sharing
structures. It is shown that state-of-the-art performance on several standard
benchmark datasets can actually be achieved by training separate probability
estimators for each dimension. In that case, model training can be trivially
parallelized over data dimensions. On the other hand, complexity control has to
be performed for each learned conditional distribution. Three possible methods
are considered and experimentally compared. The estimator that is employed for
each conditional is LogitBoost. Similarities and differences between the
proposed approach and autoregressive models based on neural networks are
discussed in detail.
| [
{
"version": "v1",
"created": "Wed, 22 Mar 2017 03:26:32 GMT"
}
] | 2017-03-23T00:00:00 | [
[
"Goessling",
"Marc",
""
]
] | TITLE: LogitBoost autoregressive networks
ABSTRACT: Multivariate binary distributions can be decomposed into products of
univariate conditional distributions. Recently popular approaches have modeled
these conditionals through neural networks with sophisticated weight-sharing
structures. It is shown that state-of-the-art performance on several standard
benchmark datasets can actually be achieved by training separate probability
estimators for each dimension. In that case, model training can be trivially
parallelized over data dimensions. On the other hand, complexity control has to
be performed for each learned conditional distribution. Three possible methods
are considered and experimentally compared. The estimator that is employed for
each conditional is LogitBoost. Similarities and differences between the
proposed approach and autoregressive models based on neural networks are
discussed in detail.
| no_new_dataset | 0.943556 |
1703.07579 | Zhongwen Xu | Fan Wu, Zhongwen Xu, Yi Yang | An End-to-End Approach to Natural Language Object Retrieval via
Context-Aware Deep Reinforcement Learning | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose an end-to-end approach to the natural language object retrieval
task, which localizes an object within an image according to a natural language
description, i.e., referring expression. Previous works divide this problem
into two independent stages: first, compute region proposals from the image
without the exploration of the language description; second, score the object
proposals with regard to the referring expression and choose the top-ranked
proposals. The object proposals are generated independently from the referring
expression, which makes the proposal generation redundant and even irrelevant
to the referred object. In this work, we train an agent with deep reinforcement
learning, which learns to move and reshape a bounding box to localize the
object according to the referring expression. We incorporate both the spatial
and temporal context information into the training procedure. By simultaneously
exploiting local visual information, the spatial and temporal context and the
referring language a priori, the agent selects an appropriate action to take at
each time. A special action is defined to indicate when the agent finds the
referred object, and terminate the procedure. We evaluate our model on various
datasets, and our algorithm significantly outperforms the compared algorithms.
Notably, the accuracy improvement of our method over the recent method GroundeR
and SCRC on the ReferItGame dataset are 7.67% and 18.25%, respectively.
| [
{
"version": "v1",
"created": "Wed, 22 Mar 2017 09:25:49 GMT"
}
] | 2017-03-23T00:00:00 | [
[
"Wu",
"Fan",
""
],
[
"Xu",
"Zhongwen",
""
],
[
"Yang",
"Yi",
""
]
] | TITLE: An End-to-End Approach to Natural Language Object Retrieval via
Context-Aware Deep Reinforcement Learning
ABSTRACT: We propose an end-to-end approach to the natural language object retrieval
task, which localizes an object within an image according to a natural language
description, i.e., referring expression. Previous works divide this problem
into two independent stages: first, compute region proposals from the image
without the exploration of the language description; second, score the object
proposals with regard to the referring expression and choose the top-ranked
proposals. The object proposals are generated independently from the referring
expression, which makes the proposal generation redundant and even irrelevant
to the referred object. In this work, we train an agent with deep reinforcement
learning, which learns to move and reshape a bounding box to localize the
object according to the referring expression. We incorporate both the spatial
and temporal context information into the training procedure. By simultaneously
exploiting local visual information, the spatial and temporal context and the
referring language a priori, the agent selects an appropriate action to take at
each time. A special action is defined to indicate when the agent finds the
referred object, and terminate the procedure. We evaluate our model on various
datasets, and our algorithm significantly outperforms the compared algorithms.
Notably, the accuracy improvement of our method over the recent method GroundeR
and SCRC on the ReferItGame dataset are 7.67% and 18.25%, respectively.
| no_new_dataset | 0.947381 |
1703.07617 | Jun Sun | Yuanzhen Ji, Jun Sun, Anisoara Nica, Zbigniew Jerzak, Gregor
Hackenbroich, Christof Fetzer | Quality-Driven Disorder Handling for M-way Sliding Window Stream Joins | 12 pages, 11 figures, IEEE ICDE 2016 | null | 10.1109/ICDE.2016.7498265 | null | cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Sliding window join is one of the most important operators for stream
applications. To produce high quality join results, a stream processing system
must deal with the ubiquitous disorder within input streams which is caused by
network delay, asynchronous source clocks, etc. Disorder handling involves an
inevitable tradeoff between the latency and the quality of produced join
results. To meet different requirements of stream applications, it is desirable
to provide a user-configurable result-latency vs. result-quality tradeoff.
Existing disorder handling approaches either do not provide such
configurability, or support only user-specified latency constraints.
In this work, we advocate the idea of quality-driven disorder handling, and
propose a buffer-based disorder handling approach for sliding window joins,
which minimizes sizes of input-sorting buffers, thus the result latency, while
respecting user-specified result-quality requirements. The core of our approach
is an analytical model which directly captures the relationship between sizes
of input buffers and the produced result quality. Our approach is generic. It
supports m-way sliding window joins with arbitrary join conditions. Experiments
on real-world and synthetic datasets show that, compared to the state of the
art, our approach can reduce the result latency incurred by disorder handling
by up to 95% while providing the same level of result quality.
| [
{
"version": "v1",
"created": "Wed, 22 Mar 2017 12:27:21 GMT"
}
] | 2017-03-23T00:00:00 | [
[
"Ji",
"Yuanzhen",
""
],
[
"Sun",
"Jun",
""
],
[
"Nica",
"Anisoara",
""
],
[
"Jerzak",
"Zbigniew",
""
],
[
"Hackenbroich",
"Gregor",
""
],
[
"Fetzer",
"Christof",
""
]
] | TITLE: Quality-Driven Disorder Handling for M-way Sliding Window Stream Joins
ABSTRACT: Sliding window join is one of the most important operators for stream
applications. To produce high quality join results, a stream processing system
must deal with the ubiquitous disorder within input streams which is caused by
network delay, asynchronous source clocks, etc. Disorder handling involves an
inevitable tradeoff between the latency and the quality of produced join
results. To meet different requirements of stream applications, it is desirable
to provide a user-configurable result-latency vs. result-quality tradeoff.
Existing disorder handling approaches either do not provide such
configurability, or support only user-specified latency constraints.
In this work, we advocate the idea of quality-driven disorder handling, and
propose a buffer-based disorder handling approach for sliding window joins,
which minimizes sizes of input-sorting buffers, thus the result latency, while
respecting user-specified result-quality requirements. The core of our approach
is an analytical model which directly captures the relationship between sizes
of input buffers and the produced result quality. Our approach is generic. It
supports m-way sliding window joins with arbitrary join conditions. Experiments
on real-world and synthetic datasets show that, compared to the state of the
art, our approach can reduce the result latency incurred by disorder handling
by up to 95% while providing the same level of result quality.
| no_new_dataset | 0.948775 |
1703.07625 | Joris Gu\'erin | Joris Gu\'erin, Olivier Gibaru, St\'ephane Thiery and Eric Nyiri | Clustering for Different Scales of Measurement - the Gap-Ratio Weighted
K-means Algorithm | 13 pages, 6 figures, 2 tables. This paper is under the review process
for AIAP 2017 | null | null | null | cs.LG cs.DS stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper describes a method for clustering data that are spread out over
large regions and which dimensions are on different scales of measurement. Such
an algorithm was developed to implement a robotics application consisting in
sorting and storing objects in an unsupervised way. The toy dataset used to
validate such application consists of Lego bricks of different shapes and
colors. The uncontrolled lighting conditions together with the use of RGB color
features, respectively involve data with a large spread and different levels of
measurement between data dimensions. To overcome the combination of these two
characteristics in the data, we have developed a new weighted K-means
algorithm, called gap-ratio K-means, which consists in weighting each dimension
of the feature space before running the K-means algorithm. The weight
associated with a feature is proportional to the ratio of the biggest gap
between two consecutive data points, and the average of all the other gaps.
This method is compared with two other variants of K-means on the Lego bricks
clustering problem as well as two other common classification datasets.
| [
{
"version": "v1",
"created": "Wed, 22 Mar 2017 12:50:15 GMT"
}
] | 2017-03-23T00:00:00 | [
[
"Guérin",
"Joris",
""
],
[
"Gibaru",
"Olivier",
""
],
[
"Thiery",
"Stéphane",
""
],
[
"Nyiri",
"Eric",
""
]
] | TITLE: Clustering for Different Scales of Measurement - the Gap-Ratio Weighted
K-means Algorithm
ABSTRACT: This paper describes a method for clustering data that are spread out over
large regions and which dimensions are on different scales of measurement. Such
an algorithm was developed to implement a robotics application consisting in
sorting and storing objects in an unsupervised way. The toy dataset used to
validate such application consists of Lego bricks of different shapes and
colors. The uncontrolled lighting conditions together with the use of RGB color
features, respectively involve data with a large spread and different levels of
measurement between data dimensions. To overcome the combination of these two
characteristics in the data, we have developed a new weighted K-means
algorithm, called gap-ratio K-means, which consists in weighting each dimension
of the feature space before running the K-means algorithm. The weight
associated with a feature is proportional to the ratio of the biggest gap
between two consecutive data points, and the average of all the other gaps.
This method is compared with two other variants of K-means on the Lego bricks
clustering problem as well as two other common classification datasets.
| new_dataset | 0.578322 |
1512.05742 | Iulian Vlad Serban | Iulian Vlad Serban, Ryan Lowe, Peter Henderson, Laurent Charlin,
Joelle Pineau | A Survey of Available Corpora for Building Data-Driven Dialogue Systems | 56 pages including references and appendix, 5 tables and 1 figure;
Under review for the Dialogue & Discourse journal. Update: paper has been
rewritten and now includes several new datasets | null | null | null | cs.CL cs.AI cs.HC cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | During the past decade, several areas of speech and language understanding
have witnessed substantial breakthroughs from the use of data-driven models. In
the area of dialogue systems, the trend is less obvious, and most practical
systems are still built through significant engineering and expert knowledge.
Nevertheless, several recent results suggest that data-driven approaches are
feasible and quite promising. To facilitate research in this area, we have
carried out a wide survey of publicly available datasets suitable for
data-driven learning of dialogue systems. We discuss important characteristics
of these datasets, how they can be used to learn diverse dialogue strategies,
and their other potential uses. We also examine methods for transfer learning
between datasets and the use of external knowledge. Finally, we discuss
appropriate choice of evaluation metrics for the learning objective.
| [
{
"version": "v1",
"created": "Thu, 17 Dec 2015 19:52:39 GMT"
},
{
"version": "v2",
"created": "Tue, 22 Dec 2015 04:58:05 GMT"
},
{
"version": "v3",
"created": "Tue, 21 Mar 2017 01:15:32 GMT"
}
] | 2017-03-22T00:00:00 | [
[
"Serban",
"Iulian Vlad",
""
],
[
"Lowe",
"Ryan",
""
],
[
"Henderson",
"Peter",
""
],
[
"Charlin",
"Laurent",
""
],
[
"Pineau",
"Joelle",
""
]
] | TITLE: A Survey of Available Corpora for Building Data-Driven Dialogue Systems
ABSTRACT: During the past decade, several areas of speech and language understanding
have witnessed substantial breakthroughs from the use of data-driven models. In
the area of dialogue systems, the trend is less obvious, and most practical
systems are still built through significant engineering and expert knowledge.
Nevertheless, several recent results suggest that data-driven approaches are
feasible and quite promising. To facilitate research in this area, we have
carried out a wide survey of publicly available datasets suitable for
data-driven learning of dialogue systems. We discuss important characteristics
of these datasets, how they can be used to learn diverse dialogue strategies,
and their other potential uses. We also examine methods for transfer learning
between datasets and the use of external knowledge. Finally, we discuss
appropriate choice of evaluation metrics for the learning objective.
| no_new_dataset | 0.945399 |
1603.06765 | Xiao Liu | Xiao Liu, Tian Xia, Jiang Wang, Yi Yang, Feng Zhou, Yuanqing Lin | Fully Convolutional Attention Networks for Fine-Grained Recognition | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Fine-grained recognition is challenging due to its subtle local inter-class
differences versus large intra-class variations such as poses. A key to address
this problem is to localize discriminative parts to extract pose-invariant
features. However, ground-truth part annotations can be expensive to acquire.
Moreover, it is hard to define parts for many fine-grained classes. This work
introduces Fully Convolutional Attention Networks (FCANs), a reinforcement
learning framework to optimally glimpse local discriminative regions adaptive
to different fine-grained domains. Compared to previous methods, our approach
enjoys three advantages: 1) the weakly-supervised reinforcement learning
procedure requires no expensive part annotations; 2) the fully-convolutional
architecture speeds up both training and testing; 3) the greedy reward strategy
accelerates the convergence of the learning. We demonstrate the effectiveness
of our method with extensive experiments on four challenging fine-grained
benchmark datasets, including CUB-200-2011, Stanford Dogs, Stanford Cars and
Food-101.
| [
{
"version": "v1",
"created": "Tue, 22 Mar 2016 12:45:20 GMT"
},
{
"version": "v2",
"created": "Sat, 4 Jun 2016 11:46:30 GMT"
},
{
"version": "v3",
"created": "Mon, 21 Nov 2016 11:12:45 GMT"
},
{
"version": "v4",
"created": "Tue, 21 Mar 2017 02:08:15 GMT"
}
] | 2017-03-22T00:00:00 | [
[
"Liu",
"Xiao",
""
],
[
"Xia",
"Tian",
""
],
[
"Wang",
"Jiang",
""
],
[
"Yang",
"Yi",
""
],
[
"Zhou",
"Feng",
""
],
[
"Lin",
"Yuanqing",
""
]
] | TITLE: Fully Convolutional Attention Networks for Fine-Grained Recognition
ABSTRACT: Fine-grained recognition is challenging due to its subtle local inter-class
differences versus large intra-class variations such as poses. A key to address
this problem is to localize discriminative parts to extract pose-invariant
features. However, ground-truth part annotations can be expensive to acquire.
Moreover, it is hard to define parts for many fine-grained classes. This work
introduces Fully Convolutional Attention Networks (FCANs), a reinforcement
learning framework to optimally glimpse local discriminative regions adaptive
to different fine-grained domains. Compared to previous methods, our approach
enjoys three advantages: 1) the weakly-supervised reinforcement learning
procedure requires no expensive part annotations; 2) the fully-convolutional
architecture speeds up both training and testing; 3) the greedy reward strategy
accelerates the convergence of the learning. We demonstrate the effectiveness
of our method with extensive experiments on four challenging fine-grained
benchmark datasets, including CUB-200-2011, Stanford Dogs, Stanford Cars and
Food-101.
| no_new_dataset | 0.948058 |
1606.09284 | Jacopo Grilli | Jacopo Grilli, Matteo Osella, Andrew S. Kennard, Marco Cosentino
Lagomarsino | Relevant parameters in models of cell division control | 15 pages, 5 figures | Phys. Rev. E 95, 032411 (2017) | 10.1103/PhysRevE.95.032411 | null | q-bio.CB cond-mat.stat-mech physics.bio-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A recent burst of dynamic single-cell growth-division data makes it possible
to characterize the stochastic dynamics of cell division control in bacteria.
Different modeling frameworks were used to infer specific mechanisms from such
data, but the links between frameworks are poorly explored, with relevant
consequences for how well any particular mechanism can be supported by the
data. Here, we describe a simple and generic framework in which two common
formalisms can be used interchangeably: (i) a continuous-time division process
described by a hazard function and (ii) a discrete-time equation describing
cell size across generations (where the unit of time is a cell cycle). In our
framework, this second process is a discrete-time Langevin equation with a
simple physical analogue. By perturbative expansion around the mean initial
size (or inter-division time), we show explicitly how this framework describes
a wide range of division control mechanisms, including combinations of time and
size control, as well as the constant added size mechanism recently found to
capture several aspects of the cell division behavior of different bacteria. As
we show by analytical estimates and numerical simulation, the available data
are characterized with great precision by the first-order approximation of this
expansion. Hence, a single dimensionless parameter defines the strength and the
action of the division control. However, this parameter may emerge from several
mechanisms, which are distinguished only by higher-order terms in our
perturbative expansion. An analytical estimate of the sample size needed to
distinguish between second-order effects shows that this is larger than what is
available in the current datasets. These results provide a unified framework
for future studies and clarify the relevant parameters at play in the control
of cell division.
| [
{
"version": "v1",
"created": "Wed, 29 Jun 2016 20:58:36 GMT"
}
] | 2017-03-22T00:00:00 | [
[
"Grilli",
"Jacopo",
""
],
[
"Osella",
"Matteo",
""
],
[
"Kennard",
"Andrew S.",
""
],
[
"Lagomarsino",
"Marco Cosentino",
""
]
] | TITLE: Relevant parameters in models of cell division control
ABSTRACT: A recent burst of dynamic single-cell growth-division data makes it possible
to characterize the stochastic dynamics of cell division control in bacteria.
Different modeling frameworks were used to infer specific mechanisms from such
data, but the links between frameworks are poorly explored, with relevant
consequences for how well any particular mechanism can be supported by the
data. Here, we describe a simple and generic framework in which two common
formalisms can be used interchangeably: (i) a continuous-time division process
described by a hazard function and (ii) a discrete-time equation describing
cell size across generations (where the unit of time is a cell cycle). In our
framework, this second process is a discrete-time Langevin equation with a
simple physical analogue. By perturbative expansion around the mean initial
size (or inter-division time), we show explicitly how this framework describes
a wide range of division control mechanisms, including combinations of time and
size control, as well as the constant added size mechanism recently found to
capture several aspects of the cell division behavior of different bacteria. As
we show by analytical estimates and numerical simulation, the available data
are characterized with great precision by the first-order approximation of this
expansion. Hence, a single dimensionless parameter defines the strength and the
action of the division control. However, this parameter may emerge from several
mechanisms, which are distinguished only by higher-order terms in our
perturbative expansion. An analytical estimate of the sample size needed to
distinguish between second-order effects shows that this is larger than what is
available in the current datasets. These results provide a unified framework
for future studies and clarify the relevant parameters at play in the control
of cell division.
| no_new_dataset | 0.946448 |
1607.02204 | Giuseppe Lisanti | Giuseppe Lisanti, Svebor Karaman, Iacopo Masi | Multi Channel-Kernel Canonical Correlation Analysis for Cross-View
Person Re-Identification | The latest/updated version of the manuscript with more experiments
can be found at https://doi.org/10.1145/3038916. Please cite the paper using
https://doi.org/10.1145/3038916 | ACM Transactions on Multimedia Computing, Communications, and
Applications (TOMM), Volume 13 Issue 2, March 2017 | 10.1145/3038916 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper we introduce a method to overcome one of the main challenges of
person re-identification in multi-camera networks, namely cross-view appearance
changes. The proposed solution addresses the extreme variability of person
appearance in different camera views by exploiting multiple feature
representations. For each feature, Kernel Canonical Correlation Analysis (KCCA)
with different kernels is exploited to learn several projection spaces in which
the appearance correlation between samples of the same person observed from
different cameras is maximized. An iterative logistic regression is finally
used to select and weigh the contributions of each feature projections and
perform the matching between the two views. Experimental evaluation shows that
the proposed solution obtains comparable performance on VIPeR and PRID 450s
datasets and improves on PRID and CUHK01 datasets with respect to the state of
the art.
| [
{
"version": "v1",
"created": "Fri, 8 Jul 2016 00:40:38 GMT"
},
{
"version": "v2",
"created": "Tue, 21 Mar 2017 10:14:10 GMT"
}
] | 2017-03-22T00:00:00 | [
[
"Lisanti",
"Giuseppe",
""
],
[
"Karaman",
"Svebor",
""
],
[
"Masi",
"Iacopo",
""
]
] | TITLE: Multi Channel-Kernel Canonical Correlation Analysis for Cross-View
Person Re-Identification
ABSTRACT: In this paper we introduce a method to overcome one of the main challenges of
person re-identification in multi-camera networks, namely cross-view appearance
changes. The proposed solution addresses the extreme variability of person
appearance in different camera views by exploiting multiple feature
representations. For each feature, Kernel Canonical Correlation Analysis (KCCA)
with different kernels is exploited to learn several projection spaces in which
the appearance correlation between samples of the same person observed from
different cameras is maximized. An iterative logistic regression is finally
used to select and weigh the contributions of each feature projections and
perform the matching between the two views. Experimental evaluation shows that
the proposed solution obtains comparable performance on VIPeR and PRID 450s
datasets and improves on PRID and CUHK01 datasets with respect to the state of
the art.
| no_new_dataset | 0.950641 |
1611.08906 | Yiannis Andreopoulos | Aaron Chadha and Yiannis Andreopoulos | Voronoi-based compact image descriptors: Efficient Region-of-Interest
retrieval with VLAD and deep-learning-based descriptors | IEEE Transaction on Multimedia, to appear | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate the problem of image retrieval based on visual queries when
the latter comprise arbitrary regions-of-interest (ROI) rather than entire
images. Our proposal is a compact image descriptor that combines the
state-of-the-art in content-based descriptor extraction with a multi-level,
Voronoi-based spatial partitioning of each dataset image. The proposed
multi-level Voronoi-based encoding uses a spatial hierarchical K-means over
interest-point locations, and computes a content-based descriptor over each
cell. In order to reduce the matching complexity with minimal or no sacrifice
in retrieval performance: (i) we utilize the tree structure of the spatial
hierarchical K-means to perform a top-to-bottom pruning for local similarity
maxima; (ii) we propose a new image similarity score that combines relevant
information from all partition levels into a single measure for similarity;
(iii) we combine our proposal with a novel and efficient approach for optimal
bit allocation within quantized descriptor representations. By deriving both a
Voronoi-based VLAD descriptor (termed as Fast-VVLAD) and a Voronoi-based deep
convolutional neural network (CNN) descriptor (termed as Fast-VDCNN), we
demonstrate that our Voronoi-based framework is agnostic to the descriptor
basis, and can easily be slotted into existing frameworks. Via a range of ROI
queries in two standard datasets, it is shown that the Voronoi-based
descriptors achieve comparable or higher mean Average Precision against
conventional grid-based spatial search, while offering more than two-fold
reduction in complexity. Finally, beyond ROI queries, we show that Voronoi
partitioning improves the geometric invariance of compact CNN descriptors,
thereby resulting in competitive performance to the current state-of-the-art on
whole image retrieval.
| [
{
"version": "v1",
"created": "Sun, 27 Nov 2016 20:35:48 GMT"
},
{
"version": "v2",
"created": "Mon, 20 Mar 2017 18:37:56 GMT"
}
] | 2017-03-22T00:00:00 | [
[
"Chadha",
"Aaron",
""
],
[
"Andreopoulos",
"Yiannis",
""
]
] | TITLE: Voronoi-based compact image descriptors: Efficient Region-of-Interest
retrieval with VLAD and deep-learning-based descriptors
ABSTRACT: We investigate the problem of image retrieval based on visual queries when
the latter comprise arbitrary regions-of-interest (ROI) rather than entire
images. Our proposal is a compact image descriptor that combines the
state-of-the-art in content-based descriptor extraction with a multi-level,
Voronoi-based spatial partitioning of each dataset image. The proposed
multi-level Voronoi-based encoding uses a spatial hierarchical K-means over
interest-point locations, and computes a content-based descriptor over each
cell. In order to reduce the matching complexity with minimal or no sacrifice
in retrieval performance: (i) we utilize the tree structure of the spatial
hierarchical K-means to perform a top-to-bottom pruning for local similarity
maxima; (ii) we propose a new image similarity score that combines relevant
information from all partition levels into a single measure for similarity;
(iii) we combine our proposal with a novel and efficient approach for optimal
bit allocation within quantized descriptor representations. By deriving both a
Voronoi-based VLAD descriptor (termed as Fast-VVLAD) and a Voronoi-based deep
convolutional neural network (CNN) descriptor (termed as Fast-VDCNN), we
demonstrate that our Voronoi-based framework is agnostic to the descriptor
basis, and can easily be slotted into existing frameworks. Via a range of ROI
queries in two standard datasets, it is shown that the Voronoi-based
descriptors achieve comparable or higher mean Average Precision against
conventional grid-based spatial search, while offering more than two-fold
reduction in complexity. Finally, beyond ROI queries, we show that Voronoi
partitioning improves the geometric invariance of compact CNN descriptors,
thereby resulting in competitive performance to the current state-of-the-art on
whole image retrieval.
| no_new_dataset | 0.956022 |
1703.05161 | Christian Reinbacher | Christian Reinbacher and Gottfried Munda and Thomas Pock | Real-Time Panoramic Tracking for Event Cameras | Accepted to International Conference on Computational Photography
2017 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Event cameras are a paradigm shift in camera technology. Instead of full
frames, the sensor captures a sparse set of events caused by intensity changes.
Since only the changes are transferred, those cameras are able to capture quick
movements of objects in the scene or of the camera itself. In this work we
propose a novel method to perform camera tracking of event cameras in a
panoramic setting with three degrees of freedom. We propose a direct camera
tracking formulation, similar to state-of-the-art in visual odometry. We show
that the minimal information needed for simultaneous tracking and mapping is
the spatial position of events, without using the appearance of the imaged
scene point. We verify the robustness to fast camera movements and dynamic
objects in the scene on a recently proposed dataset and self-recorded
sequences.
| [
{
"version": "v1",
"created": "Wed, 15 Mar 2017 14:03:47 GMT"
},
{
"version": "v2",
"created": "Tue, 21 Mar 2017 13:08:49 GMT"
}
] | 2017-03-22T00:00:00 | [
[
"Reinbacher",
"Christian",
""
],
[
"Munda",
"Gottfried",
""
],
[
"Pock",
"Thomas",
""
]
] | TITLE: Real-Time Panoramic Tracking for Event Cameras
ABSTRACT: Event cameras are a paradigm shift in camera technology. Instead of full
frames, the sensor captures a sparse set of events caused by intensity changes.
Since only the changes are transferred, those cameras are able to capture quick
movements of objects in the scene or of the camera itself. In this work we
propose a novel method to perform camera tracking of event cameras in a
panoramic setting with three degrees of freedom. We propose a direct camera
tracking formulation, similar to state-of-the-art in visual odometry. We show
that the minimal information needed for simultaneous tracking and mapping is
the spatial position of events, without using the appearance of the imaged
scene point. We verify the robustness to fast camera movements and dynamic
objects in the scene on a recently proposed dataset and self-recorded
sequences.
| new_dataset | 0.956594 |
1703.06585 | Abhishek Das | Abhishek Das, Satwik Kottur, Jos\'e M. F. Moura, Stefan Lee, Dhruv
Batra | Learning Cooperative Visual Dialog Agents with Deep Reinforcement
Learning | 11 pages, 4 figures, 2 tables, webpage: http://visualdialog.org/ | null | null | null | cs.CV cs.AI cs.CL cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce the first goal-driven training for visual question answering and
dialog agents. Specifically, we pose a cooperative 'image guessing' game
between two agents -- Qbot and Abot -- who communicate in natural language
dialog so that Qbot can select an unseen image from a lineup of images. We use
deep reinforcement learning (RL) to learn the policies of these agents
end-to-end -- from pixels to multi-agent multi-round dialog to game reward.
We demonstrate two experimental results.
First, as a 'sanity check' demonstration of pure RL (from scratch), we show
results on a synthetic world, where the agents communicate in ungrounded
vocabulary, i.e., symbols with no pre-specified meanings (X, Y, Z). We find
that two bots invent their own communication protocol and start using certain
symbols to ask/answer about certain visual attributes (shape/color/style).
Thus, we demonstrate the emergence of grounded language and communication among
'visual' dialog agents with no human supervision.
Second, we conduct large-scale real-image experiments on the VisDial dataset,
where we pretrain with supervised dialog data and show that the RL 'fine-tuned'
agents significantly outperform SL agents. Interestingly, the RL Qbot learns to
ask questions that Abot is good at, ultimately resulting in more informative
dialog and a better team.
| [
{
"version": "v1",
"created": "Mon, 20 Mar 2017 03:50:57 GMT"
},
{
"version": "v2",
"created": "Tue, 21 Mar 2017 17:41:23 GMT"
}
] | 2017-03-22T00:00:00 | [
[
"Das",
"Abhishek",
""
],
[
"Kottur",
"Satwik",
""
],
[
"Moura",
"José M. F.",
""
],
[
"Lee",
"Stefan",
""
],
[
"Batra",
"Dhruv",
""
]
] | TITLE: Learning Cooperative Visual Dialog Agents with Deep Reinforcement
Learning
ABSTRACT: We introduce the first goal-driven training for visual question answering and
dialog agents. Specifically, we pose a cooperative 'image guessing' game
between two agents -- Qbot and Abot -- who communicate in natural language
dialog so that Qbot can select an unseen image from a lineup of images. We use
deep reinforcement learning (RL) to learn the policies of these agents
end-to-end -- from pixels to multi-agent multi-round dialog to game reward.
We demonstrate two experimental results.
First, as a 'sanity check' demonstration of pure RL (from scratch), we show
results on a synthetic world, where the agents communicate in ungrounded
vocabulary, i.e., symbols with no pre-specified meanings (X, Y, Z). We find
that two bots invent their own communication protocol and start using certain
symbols to ask/answer about certain visual attributes (shape/color/style).
Thus, we demonstrate the emergence of grounded language and communication among
'visual' dialog agents with no human supervision.
Second, we conduct large-scale real-image experiments on the VisDial dataset,
where we pretrain with supervised dialog data and show that the RL 'fine-tuned'
agents significantly outperform SL agents. Interestingly, the RL Qbot learns to
ask questions that Abot is good at, ultimately resulting in more informative
dialog and a better team.
| no_new_dataset | 0.939969 |
1703.06902 | Juncheng Li | Juncheng Li, Wei Dai, Florian Metze, Shuhui Qu, Samarjit Das | A Comparison of deep learning methods for environmental sound | 5 pages including reference | published at ICASSP 2017 | null | null | cs.SD cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Environmental sound detection is a challenging application of machine
learning because of the noisy nature of the signal, and the small amount of
(labeled) data that is typically available. This work thus presents a
comparison of several state-of-the-art Deep Learning models on the IEEE
challenge on Detection and Classification of Acoustic Scenes and Events (DCASE)
2016 challenge task and data, classifying sounds into one of fifteen common
indoor and outdoor acoustic scenes, such as bus, cafe, car, city center, forest
path, library, train, etc. In total, 13 hours of stereo audio recordings are
available, making this one of the largest datasets available. We perform
experiments on six sets of features, including standard Mel-frequency cepstral
coefficients (MFCC), Binaural MFCC, log Mel-spectrum and two different large-
scale temporal pooling features extracted using OpenSMILE. On these features,
we apply five models: Gaussian Mixture Model (GMM), Deep Neural Network (DNN),
Recurrent Neural Network (RNN), Convolutional Deep Neural Net- work (CNN) and
i-vector. Using the late-fusion approach, we improve the performance of the
baseline 72.5% by 15.6% in 4-fold Cross Validation (CV) avg. accuracy and 11%
in test accuracy, which matches the best result of the DCASE 2016 challenge.
With large feature sets, deep neural network models out- perform traditional
methods and achieve the best performance among all the studied methods.
Consistent with other work, the best performing single model is the
non-temporal DNN model, which we take as evidence that sounds in the DCASE
challenge do not exhibit strong temporal dynamics.
| [
{
"version": "v1",
"created": "Mon, 20 Mar 2017 18:11:47 GMT"
}
] | 2017-03-22T00:00:00 | [
[
"Li",
"Juncheng",
""
],
[
"Dai",
"Wei",
""
],
[
"Metze",
"Florian",
""
],
[
"Qu",
"Shuhui",
""
],
[
"Das",
"Samarjit",
""
]
] | TITLE: A Comparison of deep learning methods for environmental sound
ABSTRACT: Environmental sound detection is a challenging application of machine
learning because of the noisy nature of the signal, and the small amount of
(labeled) data that is typically available. This work thus presents a
comparison of several state-of-the-art Deep Learning models on the IEEE
challenge on Detection and Classification of Acoustic Scenes and Events (DCASE)
2016 challenge task and data, classifying sounds into one of fifteen common
indoor and outdoor acoustic scenes, such as bus, cafe, car, city center, forest
path, library, train, etc. In total, 13 hours of stereo audio recordings are
available, making this one of the largest datasets available. We perform
experiments on six sets of features, including standard Mel-frequency cepstral
coefficients (MFCC), Binaural MFCC, log Mel-spectrum and two different large-
scale temporal pooling features extracted using OpenSMILE. On these features,
we apply five models: Gaussian Mixture Model (GMM), Deep Neural Network (DNN),
Recurrent Neural Network (RNN), Convolutional Deep Neural Net- work (CNN) and
i-vector. Using the late-fusion approach, we improve the performance of the
baseline 72.5% by 15.6% in 4-fold Cross Validation (CV) avg. accuracy and 11%
in test accuracy, which matches the best result of the DCASE 2016 challenge.
With large feature sets, deep neural network models out- perform traditional
methods and achieve the best performance among all the studied methods.
Consistent with other work, the best performing single model is the
non-temporal DNN model, which we take as evidence that sounds in the DCASE
challenge do not exhibit strong temporal dynamics.
| no_new_dataset | 0.946794 |
1703.07004 | Marzyeh Ghassemi | Harini Suresh, Peter Szolovits, Marzyeh Ghassemi | The Use of Autoencoders for Discovering Patient Phenotypes | null | NIPS Workshop on Machine Learning for Healthcare (NIPS ML4HC) 2016 | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We use autoencoders to create low-dimensional embeddings of underlying
patient phenotypes that we hypothesize are a governing factor in determining
how different patients will react to different interventions. We compare the
performance of autoencoders that take fixed length sequences of concatenated
timesteps as input with a recurrent sequence-to-sequence autoencoder. We
evaluate our methods on around 35,500 patients from the latest MIMIC III
dataset from Beth Israel Deaconess Hospital.
| [
{
"version": "v1",
"created": "Mon, 20 Mar 2017 23:30:40 GMT"
}
] | 2017-03-22T00:00:00 | [
[
"Suresh",
"Harini",
""
],
[
"Szolovits",
"Peter",
""
],
[
"Ghassemi",
"Marzyeh",
""
]
] | TITLE: The Use of Autoencoders for Discovering Patient Phenotypes
ABSTRACT: We use autoencoders to create low-dimensional embeddings of underlying
patient phenotypes that we hypothesize are a governing factor in determining
how different patients will react to different interventions. We compare the
performance of autoencoders that take fixed length sequences of concatenated
timesteps as input with a recurrent sequence-to-sequence autoencoder. We
evaluate our methods on around 35,500 patients from the latest MIMIC III
dataset from Beth Israel Deaconess Hospital.
| no_new_dataset | 0.945651 |
1703.07090 | Jun Zhang | Xu Tian, Jun Zhang, Zejun Ma, Yi He, Juan Wei, Peihao Wu, Wenchang
Situ, Shuai Li, Yang Zhang | Deep LSTM for Large Vocabulary Continuous Speech Recognition | 8 pages. arXiv admin note: text overlap with arXiv:1703.01024 | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recurrent neural networks (RNNs), especially long short-term memory (LSTM)
RNNs, are effective network for sequential task like speech recognition. Deeper
LSTM models perform well on large vocabulary continuous speech recognition,
because of their impressive learning ability. However, it is more difficult to
train a deeper network. We introduce a training framework with layer-wise
training and exponential moving average methods for deeper LSTM models. It is a
competitive framework that LSTM models of more than 7 layers are successfully
trained on Shenma voice search data in Mandarin and they outperform the deep
LSTM models trained by conventional approach. Moreover, in order for online
streaming speech recognition applications, the shallow model with low real time
factor is distilled from the very deep model. The recognition accuracy have
little loss in the distillation process. Therefore, the model trained with the
proposed training framework reduces relative 14\% character error rate,
compared to original model which has the similar real-time capability.
Furthermore, the novel transfer learning strategy with segmental Minimum
Bayes-Risk is also introduced in the framework. The strategy makes it possible
that training with only a small part of dataset could outperform full dataset
training from the beginning.
| [
{
"version": "v1",
"created": "Tue, 21 Mar 2017 08:24:50 GMT"
}
] | 2017-03-22T00:00:00 | [
[
"Tian",
"Xu",
""
],
[
"Zhang",
"Jun",
""
],
[
"Ma",
"Zejun",
""
],
[
"He",
"Yi",
""
],
[
"Wei",
"Juan",
""
],
[
"Wu",
"Peihao",
""
],
[
"Situ",
"Wenchang",
""
],
[
"Li",
"Shuai",
""
],
[
"Zhang",
"Yang",
""
]
] | TITLE: Deep LSTM for Large Vocabulary Continuous Speech Recognition
ABSTRACT: Recurrent neural networks (RNNs), especially long short-term memory (LSTM)
RNNs, are effective network for sequential task like speech recognition. Deeper
LSTM models perform well on large vocabulary continuous speech recognition,
because of their impressive learning ability. However, it is more difficult to
train a deeper network. We introduce a training framework with layer-wise
training and exponential moving average methods for deeper LSTM models. It is a
competitive framework that LSTM models of more than 7 layers are successfully
trained on Shenma voice search data in Mandarin and they outperform the deep
LSTM models trained by conventional approach. Moreover, in order for online
streaming speech recognition applications, the shallow model with low real time
factor is distilled from the very deep model. The recognition accuracy have
little loss in the distillation process. Therefore, the model trained with the
proposed training framework reduces relative 14\% character error rate,
compared to original model which has the similar real-time capability.
Furthermore, the novel transfer learning strategy with segmental Minimum
Bayes-Risk is also introduced in the framework. The strategy makes it possible
that training with only a small part of dataset could outperform full dataset
training from the beginning.
| no_new_dataset | 0.952042 |
1703.07115 | Mandar Kulkarni Mr. | Mandar Kulkarni, Shirish Karande | Layer-wise training of deep networks using kernel similarity | null | Deep Learning for Pattern Recognition (DLPR) workshop at ICPR 2016 | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deep learning has shown promising results in many machine learning
applications. The hierarchical feature representation built by deep networks
enable compact and precise encoding of the data. A kernel analysis of the
trained deep networks demonstrated that with deeper layers, more simple and
more accurate data representations are obtained. In this paper, we propose an
approach for layer-wise training of a deep network for the supervised
classification task. A transformation matrix of each layer is obtained by
solving an optimization aimed at a better representation where a subsequent
layer builds its representation on the top of the features produced by a
previous layer. We compared the performance of our approach with a DNN trained
using back-propagation which has same architecture as ours. Experimental
results on the real image datasets demonstrate efficacy of our approach. We
also performed kernel analysis of layer representations to validate the claim
of better feature encoding.
| [
{
"version": "v1",
"created": "Tue, 21 Mar 2017 09:53:51 GMT"
}
] | 2017-03-22T00:00:00 | [
[
"Kulkarni",
"Mandar",
""
],
[
"Karande",
"Shirish",
""
]
] | TITLE: Layer-wise training of deep networks using kernel similarity
ABSTRACT: Deep learning has shown promising results in many machine learning
applications. The hierarchical feature representation built by deep networks
enable compact and precise encoding of the data. A kernel analysis of the
trained deep networks demonstrated that with deeper layers, more simple and
more accurate data representations are obtained. In this paper, we propose an
approach for layer-wise training of a deep network for the supervised
classification task. A transformation matrix of each layer is obtained by
solving an optimization aimed at a better representation where a subsequent
layer builds its representation on the top of the features produced by a
previous layer. We compared the performance of our approach with a DNN trained
using back-propagation which has same architecture as ours. Experimental
results on the real image datasets demonstrate efficacy of our approach. We
also performed kernel analysis of layer representations to validate the claim
of better feature encoding.
| no_new_dataset | 0.951953 |
1703.07131 | Mandar Kulkarni Mr. | Mandar Kulkarni, Kalpesh Patil, Shirish Karande | Knowledge distillation using unlabeled mismatched images | null | null | null | null | cs.CV cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Current approaches for Knowledge Distillation (KD) either directly use
training data or sample from the training data distribution. In this paper, we
demonstrate effectiveness of 'mismatched' unlabeled stimulus to perform KD for
image classification networks. For illustration, we consider scenarios where
this is a complete absence of training data, or mismatched stimulus has to be
used for augmenting a small amount of training data. We demonstrate that
stimulus complexity is a key factor for distillation's good performance. Our
examples include use of various datasets for stimulating MNIST and CIFAR
teachers.
| [
{
"version": "v1",
"created": "Tue, 21 Mar 2017 10:34:59 GMT"
}
] | 2017-03-22T00:00:00 | [
[
"Kulkarni",
"Mandar",
""
],
[
"Patil",
"Kalpesh",
""
],
[
"Karande",
"Shirish",
""
]
] | TITLE: Knowledge distillation using unlabeled mismatched images
ABSTRACT: Current approaches for Knowledge Distillation (KD) either directly use
training data or sample from the training data distribution. In this paper, we
demonstrate effectiveness of 'mismatched' unlabeled stimulus to perform KD for
image classification networks. For illustration, we consider scenarios where
this is a complete absence of training data, or mismatched stimulus has to be
used for augmenting a small amount of training data. We demonstrate that
stimulus complexity is a key factor for distillation's good performance. Our
examples include use of various datasets for stimulating MNIST and CIFAR
teachers.
| no_new_dataset | 0.950549 |
1703.07144 | Bumsub Ham | Bumsub Ham, Minsu Cho, Cordelia Schmid, Jean Ponce | Proposal Flow: Semantic Correspondences from Object Proposals | arXiv admin note: text overlap with arXiv:1511.05065 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Finding image correspondences remains a challenging problem in the presence
of intra-class variations and large changes in scene layout. Semantic flow
methods are designed to handle images depicting different instances of the same
object or scene category. We introduce a novel approach to semantic flow,
dubbed proposal flow, that establishes reliable correspondences using object
proposals. Unlike prevailing semantic flow approaches that operate on pixels or
regularly sampled local regions, proposal flow benefits from the
characteristics of modern object proposals, that exhibit high repeatability at
multiple scales, and can take advantage of both local and geometric consistency
constraints among proposals. We also show that the corresponding sparse
proposal flow can effectively be transformed into a conventional dense flow
field. We introduce two new challenging datasets that can be used to evaluate
both general semantic flow techniques and region-based approaches such as
proposal flow. We use these benchmarks to compare different matching
algorithms, object proposals, and region features within proposal flow, to the
state of the art in semantic flow. This comparison, along with experiments on
standard datasets, demonstrates that proposal flow significantly outperforms
existing semantic flow methods in various settings.
| [
{
"version": "v1",
"created": "Tue, 21 Mar 2017 10:57:27 GMT"
}
] | 2017-03-22T00:00:00 | [
[
"Ham",
"Bumsub",
""
],
[
"Cho",
"Minsu",
""
],
[
"Schmid",
"Cordelia",
""
],
[
"Ponce",
"Jean",
""
]
] | TITLE: Proposal Flow: Semantic Correspondences from Object Proposals
ABSTRACT: Finding image correspondences remains a challenging problem in the presence
of intra-class variations and large changes in scene layout. Semantic flow
methods are designed to handle images depicting different instances of the same
object or scene category. We introduce a novel approach to semantic flow,
dubbed proposal flow, that establishes reliable correspondences using object
proposals. Unlike prevailing semantic flow approaches that operate on pixels or
regularly sampled local regions, proposal flow benefits from the
characteristics of modern object proposals, that exhibit high repeatability at
multiple scales, and can take advantage of both local and geometric consistency
constraints among proposals. We also show that the corresponding sparse
proposal flow can effectively be transformed into a conventional dense flow
field. We introduce two new challenging datasets that can be used to evaluate
both general semantic flow techniques and region-based approaches such as
proposal flow. We use these benchmarks to compare different matching
algorithms, object proposals, and region features within proposal flow, to the
state of the art in semantic flow. This comparison, along with experiments on
standard datasets, demonstrates that proposal flow significantly outperforms
existing semantic flow methods in various settings.
| new_dataset | 0.958187 |
1703.07334 | Shichao Yang | Shichao Yang, Yu Song, Michael Kaess, Sebastian Scherer | Pop-up SLAM: Semantic Monocular Plane SLAM for Low-texture Environments | International Conference on Intelligent Robots and Systems (IROS)
2016 | null | null | null | cs.CV cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Existing simultaneous localization and mapping (SLAM) algorithms are not
robust in challenging low-texture environments because there are only few
salient features. The resulting sparse or semi-dense map also conveys little
information for motion planning. Though some work utilize plane or scene layout
for dense map regularization, they require decent state estimation from other
sources. In this paper, we propose real-time monocular plane SLAM to
demonstrate that scene understanding could improve both state estimation and
dense mapping especially in low-texture environments. The plane measurements
come from a pop-up 3D plane model applied to each single image. We also combine
planes with point based SLAM to improve robustness. On a public TUM dataset,
our algorithm generates a dense semantic 3D model with pixel depth error of 6.2
cm while existing SLAM algorithms fail. On a 60 m long dataset with loops, our
method creates a much better 3D model with state estimation error of 0.67%.
| [
{
"version": "v1",
"created": "Tue, 21 Mar 2017 17:41:46 GMT"
}
] | 2017-03-22T00:00:00 | [
[
"Yang",
"Shichao",
""
],
[
"Song",
"Yu",
""
],
[
"Kaess",
"Michael",
""
],
[
"Scherer",
"Sebastian",
""
]
] | TITLE: Pop-up SLAM: Semantic Monocular Plane SLAM for Low-texture Environments
ABSTRACT: Existing simultaneous localization and mapping (SLAM) algorithms are not
robust in challenging low-texture environments because there are only few
salient features. The resulting sparse or semi-dense map also conveys little
information for motion planning. Though some work utilize plane or scene layout
for dense map regularization, they require decent state estimation from other
sources. In this paper, we propose real-time monocular plane SLAM to
demonstrate that scene understanding could improve both state estimation and
dense mapping especially in low-texture environments. The plane measurements
come from a pop-up 3D plane model applied to each single image. We also combine
planes with point based SLAM to improve robustness. On a public TUM dataset,
our algorithm generates a dense semantic 3D model with pixel depth error of 6.2
cm while existing SLAM algorithms fail. On a 60 m long dataset with loops, our
method creates a much better 3D model with state estimation error of 0.67%.
| no_new_dataset | 0.951459 |
1511.03376 | Yue Wang | Yue Wang and Jaewoo Lee and Daniel Kifer | Revisiting Differentially Private Hypothesis Tests for Categorical Data | null | null | null | null | cs.CR stat.ME | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we consider methods for performing hypothesis tests on data
protected by a statistical disclosure control technology known as differential
privacy. Previous approaches to differentially private hypothesis testing
either perturbed the test statistic with random noise having large variance
(and resulted in a significant loss of power) or added smaller amounts of noise
directly to the data but failed to adjust the test in response to the added
noise (resulting in biased, unreliable $p$-values). In this paper, we develop a
variety of practical hypothesis tests that address these problems. Using a
different asymptotic regime that is more suited to hypothesis testing with
privacy, we show a modified equivalence between chi-squared tests and
likelihood ratio tests. We then develop differentially private likelihood ratio
and chi-squared tests for a variety of applications on tabular data (i.e.,
independence, sample proportions, and goodness-of-fit tests). Experimental
evaluations on small and large datasets using a wide variety of privacy
settings demonstrate the practicality and reliability of our methods.
| [
{
"version": "v1",
"created": "Wed, 11 Nov 2015 03:36:38 GMT"
},
{
"version": "v2",
"created": "Sat, 13 Feb 2016 03:19:19 GMT"
},
{
"version": "v3",
"created": "Fri, 2 Dec 2016 04:09:27 GMT"
},
{
"version": "v4",
"created": "Sat, 18 Mar 2017 06:55:30 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Wang",
"Yue",
""
],
[
"Lee",
"Jaewoo",
""
],
[
"Kifer",
"Daniel",
""
]
] | TITLE: Revisiting Differentially Private Hypothesis Tests for Categorical Data
ABSTRACT: In this paper, we consider methods for performing hypothesis tests on data
protected by a statistical disclosure control technology known as differential
privacy. Previous approaches to differentially private hypothesis testing
either perturbed the test statistic with random noise having large variance
(and resulted in a significant loss of power) or added smaller amounts of noise
directly to the data but failed to adjust the test in response to the added
noise (resulting in biased, unreliable $p$-values). In this paper, we develop a
variety of practical hypothesis tests that address these problems. Using a
different asymptotic regime that is more suited to hypothesis testing with
privacy, we show a modified equivalence between chi-squared tests and
likelihood ratio tests. We then develop differentially private likelihood ratio
and chi-squared tests for a variety of applications on tabular data (i.e.,
independence, sample proportions, and goodness-of-fit tests). Experimental
evaluations on small and large datasets using a wide variety of privacy
settings demonstrate the practicality and reliability of our methods.
| no_new_dataset | 0.953837 |
1606.02185 | Harrison Edwards | Harrison Edwards, Amos Storkey | Towards a Neural Statistician | Updated to camera ready version for ICLR 2017 | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | An efficient learner is one who reuses what they already know to tackle a new
problem. For a machine learner, this means understanding the similarities
amongst datasets. In order to do this, one must take seriously the idea of
working with datasets, rather than datapoints, as the key objects to model.
Towards this goal, we demonstrate an extension of a variational autoencoder
that can learn a method for computing representations, or statistics, of
datasets in an unsupervised fashion. The network is trained to produce
statistics that encapsulate a generative model for each dataset. Hence the
network enables efficient learning from new datasets for both unsupervised and
supervised tasks. We show that we are able to learn statistics that can be used
for: clustering datasets, transferring generative models to new datasets,
selecting representative samples of datasets and classifying previously unseen
classes. We refer to our model as a neural statistician, and by this we mean a
neural network that can learn to compute summary statistics of datasets without
supervision.
| [
{
"version": "v1",
"created": "Tue, 7 Jun 2016 15:36:39 GMT"
},
{
"version": "v2",
"created": "Mon, 20 Mar 2017 17:18:16 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Edwards",
"Harrison",
""
],
[
"Storkey",
"Amos",
""
]
] | TITLE: Towards a Neural Statistician
ABSTRACT: An efficient learner is one who reuses what they already know to tackle a new
problem. For a machine learner, this means understanding the similarities
amongst datasets. In order to do this, one must take seriously the idea of
working with datasets, rather than datapoints, as the key objects to model.
Towards this goal, we demonstrate an extension of a variational autoencoder
that can learn a method for computing representations, or statistics, of
datasets in an unsupervised fashion. The network is trained to produce
statistics that encapsulate a generative model for each dataset. Hence the
network enables efficient learning from new datasets for both unsupervised and
supervised tasks. We show that we are able to learn statistics that can be used
for: clustering datasets, transferring generative models to new datasets,
selecting representative samples of datasets and classifying previously unseen
classes. We refer to our model as a neural statistician, and by this we mean a
neural network that can learn to compute summary statistics of datasets without
supervision.
| no_new_dataset | 0.946001 |
1609.09747 | Antoine Deleforge | Saurabh Kataria (IIT Kanpur, Panama), Cl\'ement Gaultier (Panama),
Antoine Deleforge (Panama) | Hearing in a shoe-box : binaural source position and wall absorption
estimation using virtually supervised learning | 2017 IEEE International Conference on Acoustics, Speech and Signal
Processing (ICASSP), Mar 2017, New-Orleans, United States | null | null | hal-01372435 | cs.SD | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper introduces a new framework for supervised sound source
localization referred to as virtually-supervised learning. An acoustic shoe-box
room simulator is used to generate a large number of binaural single-source
audio scenes. These scenes are used to build a dataset of spatial binaural
features annotated with acoustic properties such as the 3D source position and
the walls' absorption coefficients. A probabilistic high- to low-dimensional
regression framework is used to learn a mapping from these features to the
acoustic properties. Results indicate that this mapping successfully estimates
the azimuth and elevation of new sources, but also their range and even the
walls' absorption coefficients solely based on binaural signals. Results also
reveal that incorporating random-diffusion effects in the data significantly
improves the estimation of all parameters.
| [
{
"version": "v1",
"created": "Fri, 30 Sep 2016 14:20:56 GMT"
},
{
"version": "v2",
"created": "Mon, 20 Mar 2017 13:39:49 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Kataria",
"Saurabh",
"",
"IIT Kanpur, Panama"
],
[
"Gaultier",
"Clément",
"",
"Panama"
],
[
"Deleforge",
"Antoine",
"",
"Panama"
]
] | TITLE: Hearing in a shoe-box : binaural source position and wall absorption
estimation using virtually supervised learning
ABSTRACT: This paper introduces a new framework for supervised sound source
localization referred to as virtually-supervised learning. An acoustic shoe-box
room simulator is used to generate a large number of binaural single-source
audio scenes. These scenes are used to build a dataset of spatial binaural
features annotated with acoustic properties such as the 3D source position and
the walls' absorption coefficients. A probabilistic high- to low-dimensional
regression framework is used to learn a mapping from these features to the
acoustic properties. Results indicate that this mapping successfully estimates
the azimuth and elevation of new sources, but also their range and even the
walls' absorption coefficients solely based on binaural signals. Results also
reveal that incorporating random-diffusion effects in the data significantly
improves the estimation of all parameters.
| no_new_dataset | 0.809276 |
1611.01436 | Tom Kwiatkowski | Kenton Lee, Shimi Salant, Tom Kwiatkowski, Ankur Parikh, Dipanjan Das,
Jonathan Berant | Learning Recurrent Span Representations for Extractive Question
Answering | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The reading comprehension task, that asks questions about a given evidence
document, is a central problem in natural language understanding. Recent
formulations of this task have typically focused on answer selection from a set
of candidates pre-defined manually or through the use of an external NLP
pipeline. However, Rajpurkar et al. (2016) recently released the SQuAD dataset
in which the answers can be arbitrary strings from the supplied text. In this
paper, we focus on this answer extraction task, presenting a novel model
architecture that efficiently builds fixed length representations of all spans
in the evidence document with a recurrent network. We show that scoring
explicit span representations significantly improves performance over other
approaches that factor the prediction into separate predictions about words or
start and end markers. Our approach improves upon the best published results of
Wang & Jiang (2016) by 5% and decreases the error of Rajpurkar et al.'s
baseline by > 50%.
| [
{
"version": "v1",
"created": "Fri, 4 Nov 2016 16:12:46 GMT"
},
{
"version": "v2",
"created": "Fri, 17 Mar 2017 18:11:12 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Lee",
"Kenton",
""
],
[
"Salant",
"Shimi",
""
],
[
"Kwiatkowski",
"Tom",
""
],
[
"Parikh",
"Ankur",
""
],
[
"Das",
"Dipanjan",
""
],
[
"Berant",
"Jonathan",
""
]
] | TITLE: Learning Recurrent Span Representations for Extractive Question
Answering
ABSTRACT: The reading comprehension task, that asks questions about a given evidence
document, is a central problem in natural language understanding. Recent
formulations of this task have typically focused on answer selection from a set
of candidates pre-defined manually or through the use of an external NLP
pipeline. However, Rajpurkar et al. (2016) recently released the SQuAD dataset
in which the answers can be arbitrary strings from the supplied text. In this
paper, we focus on this answer extraction task, presenting a novel model
architecture that efficiently builds fixed length representations of all spans
in the evidence document with a recurrent network. We show that scoring
explicit span representations significantly improves performance over other
approaches that factor the prediction into separate predictions about words or
start and end markers. Our approach improves upon the best published results of
Wang & Jiang (2016) by 5% and decreases the error of Rajpurkar et al.'s
baseline by > 50%.
| new_dataset | 0.961461 |
1611.01560 | R\'obert Beck | R\'obert Beck, L\'aszl\'o Dobos, Tam\'as Budav\'ari, Alexander S.
Szalay, Istv\'an Csabai | Photo-z-SQL: integrated, flexible photometric redshift computation in a
database | 14 pages, 5 figures. Minor revision accepted by Astronomy & Computing
on 2017 March 11 | null | 10.1016/j.ascom.2017.03.002 | null | astro-ph.GA astro-ph.IM cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a flexible template-based photometric redshift estimation
framework, implemented in C#, that can be seamlessly integrated into a SQL
database (or DB) server and executed on-demand in SQL. The DB integration
eliminates the need to move large photometric datasets outside a database for
redshift estimation, and utilizes the computational capabilities of DB
hardware. The code is able to perform both maximum likelihood and Bayesian
estimation, and can handle inputs of variable photometric filter sets and
corresponding broad-band magnitudes. It is possible to take into account the
full covariance matrix between filters, and filter zero points can be
empirically calibrated using measurements with given redshifts. The list of
spectral templates and the prior can be specified flexibly, and the expensive
synthetic magnitude computations are done via lazy evaluation, coupled with a
caching of results. Parallel execution is fully supported. For large upcoming
photometric surveys such as the LSST, the ability to perform in-place photo-z
calculation would be a significant advantage. Also, the efficient handling of
variable filter sets is a necessity for heterogeneous databases, for example
the Hubble Source Catalog, and for cross-match services such as SkyQuery. We
illustrate the performance of our code on two reference photo-z estimation
testing datasets, and provide an analysis of execution time and scalability
with respect to different configurations. The code is available for download at
https://github.com/beckrob/Photo-z-SQL.
| [
{
"version": "v1",
"created": "Fri, 4 Nov 2016 22:48:06 GMT"
},
{
"version": "v2",
"created": "Mon, 20 Mar 2017 11:46:57 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Beck",
"Róbert",
""
],
[
"Dobos",
"László",
""
],
[
"Budavári",
"Tamás",
""
],
[
"Szalay",
"Alexander S.",
""
],
[
"Csabai",
"István",
""
]
] | TITLE: Photo-z-SQL: integrated, flexible photometric redshift computation in a
database
ABSTRACT: We present a flexible template-based photometric redshift estimation
framework, implemented in C#, that can be seamlessly integrated into a SQL
database (or DB) server and executed on-demand in SQL. The DB integration
eliminates the need to move large photometric datasets outside a database for
redshift estimation, and utilizes the computational capabilities of DB
hardware. The code is able to perform both maximum likelihood and Bayesian
estimation, and can handle inputs of variable photometric filter sets and
corresponding broad-band magnitudes. It is possible to take into account the
full covariance matrix between filters, and filter zero points can be
empirically calibrated using measurements with given redshifts. The list of
spectral templates and the prior can be specified flexibly, and the expensive
synthetic magnitude computations are done via lazy evaluation, coupled with a
caching of results. Parallel execution is fully supported. For large upcoming
photometric surveys such as the LSST, the ability to perform in-place photo-z
calculation would be a significant advantage. Also, the efficient handling of
variable filter sets is a necessity for heterogeneous databases, for example
the Hubble Source Catalog, and for cross-match services such as SkyQuery. We
illustrate the performance of our code on two reference photo-z estimation
testing datasets, and provide an analysis of execution time and scalability
with respect to different configurations. The code is available for download at
https://github.com/beckrob/Photo-z-SQL.
| no_new_dataset | 0.945751 |
1611.05216 | Yemin Shi Shi | Yemin Shi and Yonghong Tian and Yaowei Wang and Tiejun Huang | Learning long-term dependencies for action recognition with a
biologically-inspired deep network | 9 pages, 4 figures | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Despite a lot of research efforts devoted in recent years, how to efficiently
learn long-term dependencies from sequences still remains a pretty challenging
task. As one of the key models for sequence learning, recurrent neural network
(RNN) and its variants such as long short term memory (LSTM) and gated
recurrent unit (GRU) are still not powerful enough in practice. One possible
reason is that they have only feedforward connections, which is different from
the biological neural system that is typically composed of both feedforward and
feedback connections. To address this problem, this paper proposes a
biologically-inspired deep network, called shuttleNet\footnote{Our code is
available at \url{https://github.com/shiyemin/shuttlenet}}. Technologically,
the shuttleNet consists of several processors, each of which is a GRU while
associated with multiple groups of cells and states. Unlike traditional RNNs,
all processors inside shuttleNet are loop connected to mimic the brain's
feedforward and feedback connections, in which they are shared across multiple
pathways in the loop connection. Attention mechanism is then employed to select
the best information flow pathway. Extensive experiments conducted on two
benchmark datasets (i.e UCF101 and HMDB51) show that we can beat
state-of-the-art methods by simply embedding shuttleNet into a CNN-RNN
framework.
| [
{
"version": "v1",
"created": "Wed, 16 Nov 2016 10:49:43 GMT"
},
{
"version": "v2",
"created": "Thu, 16 Mar 2017 15:55:14 GMT"
},
{
"version": "v3",
"created": "Sun, 19 Mar 2017 08:27:24 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Shi",
"Yemin",
""
],
[
"Tian",
"Yonghong",
""
],
[
"Wang",
"Yaowei",
""
],
[
"Huang",
"Tiejun",
""
]
] | TITLE: Learning long-term dependencies for action recognition with a
biologically-inspired deep network
ABSTRACT: Despite a lot of research efforts devoted in recent years, how to efficiently
learn long-term dependencies from sequences still remains a pretty challenging
task. As one of the key models for sequence learning, recurrent neural network
(RNN) and its variants such as long short term memory (LSTM) and gated
recurrent unit (GRU) are still not powerful enough in practice. One possible
reason is that they have only feedforward connections, which is different from
the biological neural system that is typically composed of both feedforward and
feedback connections. To address this problem, this paper proposes a
biologically-inspired deep network, called shuttleNet\footnote{Our code is
available at \url{https://github.com/shiyemin/shuttlenet}}. Technologically,
the shuttleNet consists of several processors, each of which is a GRU while
associated with multiple groups of cells and states. Unlike traditional RNNs,
all processors inside shuttleNet are loop connected to mimic the brain's
feedforward and feedback connections, in which they are shared across multiple
pathways in the loop connection. Attention mechanism is then employed to select
the best information flow pathway. Extensive experiments conducted on two
benchmark datasets (i.e UCF101 and HMDB51) show that we can beat
state-of-the-art methods by simply embedding shuttleNet into a CNN-RNN
framework.
| no_new_dataset | 0.943712 |
1611.06013 | Kui Jia | Kui Jia | Improving training of deep neural networks via Singular Value Bounding | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deep learning methods achieve great success recently on many computer vision
problems, with image classification and object detection as the prominent
examples. In spite of these practical successes, optimization of deep networks
remains an active topic in deep learning research. In this work, we focus on
investigation of the network solution properties that can potentially lead to
good performance. Our research is inspired by theoretical and empirical results
that use orthogonal matrices to initialize networks, but we are interested in
investigating how orthogonal weight matrices perform when network training
converges. To this end, we propose to constrain the solutions of weight
matrices in the orthogonal feasible set during the whole process of network
training, and achieve this by a simple yet effective method called Singular
Value Bounding (SVB). In SVB, all singular values of each weight matrix are
simply bounded in a narrow band around the value of 1. Based on the same
motivation, we also propose Bounded Batch Normalization (BBN), which improves
Batch Normalization by removing its potential risk of ill-conditioned layer
transform. We present both theoretical and empirical results to justify our
proposed methods. Experiments on benchmark image classification datasets show
the efficacy of our proposed SVB and BBN. In particular, we achieve the
state-of-the-art results of 3.06% error rate on CIFAR10 and 16.90% on CIFAR100,
using off-the-shelf network architectures (Wide ResNets). Our preliminary
results on ImageNet also show the promise in large-scale learning.
| [
{
"version": "v1",
"created": "Fri, 18 Nov 2016 09:09:56 GMT"
},
{
"version": "v2",
"created": "Tue, 29 Nov 2016 09:49:11 GMT"
},
{
"version": "v3",
"created": "Sat, 18 Mar 2017 07:27:09 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Jia",
"Kui",
""
]
] | TITLE: Improving training of deep neural networks via Singular Value Bounding
ABSTRACT: Deep learning methods achieve great success recently on many computer vision
problems, with image classification and object detection as the prominent
examples. In spite of these practical successes, optimization of deep networks
remains an active topic in deep learning research. In this work, we focus on
investigation of the network solution properties that can potentially lead to
good performance. Our research is inspired by theoretical and empirical results
that use orthogonal matrices to initialize networks, but we are interested in
investigating how orthogonal weight matrices perform when network training
converges. To this end, we propose to constrain the solutions of weight
matrices in the orthogonal feasible set during the whole process of network
training, and achieve this by a simple yet effective method called Singular
Value Bounding (SVB). In SVB, all singular values of each weight matrix are
simply bounded in a narrow band around the value of 1. Based on the same
motivation, we also propose Bounded Batch Normalization (BBN), which improves
Batch Normalization by removing its potential risk of ill-conditioned layer
transform. We present both theoretical and empirical results to justify our
proposed methods. Experiments on benchmark image classification datasets show
the efficacy of our proposed SVB and BBN. In particular, we achieve the
state-of-the-art results of 3.06% error rate on CIFAR10 and 16.90% on CIFAR100,
using off-the-shelf network architectures (Wide ResNets). Our preliminary
results on ImageNet also show the promise in large-scale learning.
| no_new_dataset | 0.946941 |
1611.09340 | Adriana Romero | Adriana Romero, Pierre Luc Carrier, Akram Erraqabi, Tristan Sylvain,
Alex Auvolat, Etienne Dejoie, Marc-Andr\'e Legault, Marie-Pierre Dub\'e,
Julie G. Hussin, Yoshua Bengio | Diet Networks: Thin Parameters for Fat Genomics | null | ICLR 2017 | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Learning tasks such as those involving genomic data often poses a serious
challenge: the number of input features can be orders of magnitude larger than
the number of training examples, making it difficult to avoid overfitting, even
when using the known regularization techniques. We focus here on tasks in which
the input is a description of the genetic variation specific to a patient, the
single nucleotide polymorphisms (SNPs), yielding millions of ternary inputs.
Improving the ability of deep learning to handle such datasets could have an
important impact in precision medicine, where high-dimensional data regarding a
particular patient is used to make predictions of interest. Even though the
amount of data for such tasks is increasing, this mismatch between the number
of examples and the number of inputs remains a concern. Naive implementations
of classifier neural networks involve a huge number of free parameters in their
first layer: each input feature is associated with as many parameters as there
are hidden units. We propose a novel neural network parametrization which
considerably reduces the number of free parameters. It is based on the idea
that we can first learn or provide a distributed representation for each input
feature (e.g. for each position in the genome where variations are observed),
and then learn (with another neural network called the parameter prediction
network) how to map a feature's distributed representation to the vector of
parameters specific to that feature in the classifier neural network (the
weights which link the value of the feature to each of the hidden units). We
show experimentally on a population stratification task of interest to medical
studies that the proposed approach can significantly reduce both the number of
parameters and the error rate of the classifier.
| [
{
"version": "v1",
"created": "Mon, 28 Nov 2016 20:50:32 GMT"
},
{
"version": "v2",
"created": "Fri, 6 Jan 2017 18:51:52 GMT"
},
{
"version": "v3",
"created": "Thu, 16 Mar 2017 21:09:28 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Romero",
"Adriana",
""
],
[
"Carrier",
"Pierre Luc",
""
],
[
"Erraqabi",
"Akram",
""
],
[
"Sylvain",
"Tristan",
""
],
[
"Auvolat",
"Alex",
""
],
[
"Dejoie",
"Etienne",
""
],
[
"Legault",
"Marc-André",
""
],
[
"Dubé",
"Marie-Pierre",
""
],
[
"Hussin",
"Julie G.",
""
],
[
"Bengio",
"Yoshua",
""
]
] | TITLE: Diet Networks: Thin Parameters for Fat Genomics
ABSTRACT: Learning tasks such as those involving genomic data often poses a serious
challenge: the number of input features can be orders of magnitude larger than
the number of training examples, making it difficult to avoid overfitting, even
when using the known regularization techniques. We focus here on tasks in which
the input is a description of the genetic variation specific to a patient, the
single nucleotide polymorphisms (SNPs), yielding millions of ternary inputs.
Improving the ability of deep learning to handle such datasets could have an
important impact in precision medicine, where high-dimensional data regarding a
particular patient is used to make predictions of interest. Even though the
amount of data for such tasks is increasing, this mismatch between the number
of examples and the number of inputs remains a concern. Naive implementations
of classifier neural networks involve a huge number of free parameters in their
first layer: each input feature is associated with as many parameters as there
are hidden units. We propose a novel neural network parametrization which
considerably reduces the number of free parameters. It is based on the idea
that we can first learn or provide a distributed representation for each input
feature (e.g. for each position in the genome where variations are observed),
and then learn (with another neural network called the parameter prediction
network) how to map a feature's distributed representation to the vector of
parameters specific to that feature in the classifier neural network (the
weights which link the value of the feature to each of the hidden units). We
show experimentally on a population stratification task of interest to medical
studies that the proposed approach can significantly reduce both the number of
parameters and the error rate of the classifier.
| no_new_dataset | 0.951594 |
1701.00180 | Hamid Hamraz | Hamid Hamraz, Marco A. Contreras, and Jun Zhang | A scalable approach for tree segmentation within small-footprint
airborne LiDAR data | The replacement version is exactly the same and only the journal
biblio information and the DOI of the published version was added | Computers and Geosciences 102 (pp. 139-147): Elsevier (2017) | 10.1016/j.cageo.2017.02.017 | null | cs.DC cs.CE | http://creativecommons.org/licenses/by/4.0/ | This paper presents a distributed approach that scales up to segment tree
crowns within a LiDAR point cloud representing an arbitrarily large forested
area. The approach uses a single-processor tree segmentation algorithm as a
building block in order to process the data delivered in the shape of tiles in
parallel. The distributed processing is performed in a master-slave manner, in
which the master maintains the global map of the tiles and coordinates the
slaves that segment tree crowns within and across the boundaries of the tiles.
A minimal bias was introduced to the number of detected trees because of trees
lying across the tile boundaries, which was quantified and adjusted for.
Theoretical and experimental analyses of the runtime of the approach revealed a
near linear speedup. The estimated number of trees categorized by crown class
and the associated error margins as well as the height distribution of the
detected trees aligned well with field estimations, verifying that the
distributed approach works correctly. The approach enables providing
information of individual tree locations and point cloud segments for a
forest-level area in a timely manner, which can be used to create detailed
remotely sensed forest inventories. Although the approach was presented for
tree segmentation within LiDAR point clouds, the idea can also be generalized
to scale up processing other big spatial datasets.
Highlights: - A scalable distributed approach for tree segmentation was
developed and theoretically analyzed. - ~2 million trees in a 7440 ha forest
was segmented in 2.5 hours using 192 cores. - 2% false positive trees were
identified as a result of the distributed run. - The approach can be used to
scale up processing other big spatial data
| [
{
"version": "v1",
"created": "Sun, 1 Jan 2017 00:10:42 GMT"
},
{
"version": "v2",
"created": "Sun, 19 Mar 2017 21:13:31 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Hamraz",
"Hamid",
""
],
[
"Contreras",
"Marco A.",
""
],
[
"Zhang",
"Jun",
""
]
] | TITLE: A scalable approach for tree segmentation within small-footprint
airborne LiDAR data
ABSTRACT: This paper presents a distributed approach that scales up to segment tree
crowns within a LiDAR point cloud representing an arbitrarily large forested
area. The approach uses a single-processor tree segmentation algorithm as a
building block in order to process the data delivered in the shape of tiles in
parallel. The distributed processing is performed in a master-slave manner, in
which the master maintains the global map of the tiles and coordinates the
slaves that segment tree crowns within and across the boundaries of the tiles.
A minimal bias was introduced to the number of detected trees because of trees
lying across the tile boundaries, which was quantified and adjusted for.
Theoretical and experimental analyses of the runtime of the approach revealed a
near linear speedup. The estimated number of trees categorized by crown class
and the associated error margins as well as the height distribution of the
detected trees aligned well with field estimations, verifying that the
distributed approach works correctly. The approach enables providing
information of individual tree locations and point cloud segments for a
forest-level area in a timely manner, which can be used to create detailed
remotely sensed forest inventories. Although the approach was presented for
tree segmentation within LiDAR point clouds, the idea can also be generalized
to scale up processing other big spatial datasets.
Highlights: - A scalable distributed approach for tree segmentation was
developed and theoretically analyzed. - ~2 million trees in a 7440 ha forest
was segmented in 2.5 hours using 192 cores. - 2% false positive trees were
identified as a result of the distributed run. - The approach can be used to
scale up processing other big spatial data
| no_new_dataset | 0.954563 |
1701.05524 | Xingchao Peng | Xingchao Peng, Kate Saenko | Synthetic to Real Adaptation with Generative Correlation Alignment
Networks | 13 pages, 8 figures | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Synthetic images rendered from 3D CAD models are useful for augmenting
training data for object recognition algorithms. However, the generated images
are non-photorealistic and do not match real image statistics. This leads to a
large domain discrepancy, causing models trained on synthetic data to perform
poorly on real domains. Recent work has shown the great potential of deep
convolutional neural networks to generate realistic images, but has not
utilized generative models to address synthetic-to-real domain adaptation. In
this work, we propose a Deep Generative Correlation Alignment Network (DGCAN)
to synthesize images using a novel domain adaption algorithm. DGCAN leverages a
shape preserving loss and a low level statistic matching loss to minimize the
domain discrepancy between synthetic and real images in deep feature space.
Experimentally, we show training off-the-shelf classifiers on the newly
generated data can significantly boost performance when testing on the real
image domains (PASCAL VOC 2007 benchmark and Office dataset), improving upon
several existing methods.
| [
{
"version": "v1",
"created": "Thu, 19 Jan 2017 17:42:00 GMT"
},
{
"version": "v2",
"created": "Thu, 9 Mar 2017 20:41:32 GMT"
},
{
"version": "v3",
"created": "Sat, 18 Mar 2017 12:56:45 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Peng",
"Xingchao",
""
],
[
"Saenko",
"Kate",
""
]
] | TITLE: Synthetic to Real Adaptation with Generative Correlation Alignment
Networks
ABSTRACT: Synthetic images rendered from 3D CAD models are useful for augmenting
training data for object recognition algorithms. However, the generated images
are non-photorealistic and do not match real image statistics. This leads to a
large domain discrepancy, causing models trained on synthetic data to perform
poorly on real domains. Recent work has shown the great potential of deep
convolutional neural networks to generate realistic images, but has not
utilized generative models to address synthetic-to-real domain adaptation. In
this work, we propose a Deep Generative Correlation Alignment Network (DGCAN)
to synthesize images using a novel domain adaption algorithm. DGCAN leverages a
shape preserving loss and a low level statistic matching loss to minimize the
domain discrepancy between synthetic and real images in deep feature space.
Experimentally, we show training off-the-shelf classifiers on the newly
generated data can significantly boost performance when testing on the real
image domains (PASCAL VOC 2007 benchmark and Office dataset), improving upon
several existing methods.
| no_new_dataset | 0.952175 |
1703.04566 | Mohammad Azzeh | Mohammad Azzeh | Model tree based adaption strategy for software effort estimation by
analogy | null | null | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Background: Adaptation technique is a crucial task for analogy based
estimation. Current adaptation techniques often use linear size or linear
similarity adjustment mechanisms which are often not suitable for datasets that
have complex structure with many categorical attributes. Furthermore, the use
of nonlinear adaptation technique such as neural network and genetic algorithms
needs many user interactions and parameters optimization for configuring them
(such as network model, number of neurons, activation functions, training
functions, mutation, selection, crossover, ... etc.). Aims: In response to the
abovementioned challenges, the present paper proposes a new adaptation strategy
using Model Tree based attribute distance to adjust estimation by analogy and
derive new estimates. Using Model Tree has an advantage to deal with
categorical attributes, minimize user interaction and improve efficiency of
model learning through classification. Method: Seven well known datasets have
been used with 3-Fold cross validation to empirically validate the proposed
approach. The proposed method has been investigated using various K analogies
from 1 to 3. Results: Experimental results showed that the proposed approach
produced better results when compared with those obtained by using estimation
by analogy based linear size adaptation, linear similarity adaptation,
'regression towards the mean' and null adaptation. Conclusions: Model Tree
could form a useful extension for estimation by analogy especially for complex
data sets with large number of categorical attributes.
| [
{
"version": "v1",
"created": "Sat, 11 Mar 2017 20:22:34 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Azzeh",
"Mohammad",
""
]
] | TITLE: Model tree based adaption strategy for software effort estimation by
analogy
ABSTRACT: Background: Adaptation technique is a crucial task for analogy based
estimation. Current adaptation techniques often use linear size or linear
similarity adjustment mechanisms which are often not suitable for datasets that
have complex structure with many categorical attributes. Furthermore, the use
of nonlinear adaptation technique such as neural network and genetic algorithms
needs many user interactions and parameters optimization for configuring them
(such as network model, number of neurons, activation functions, training
functions, mutation, selection, crossover, ... etc.). Aims: In response to the
abovementioned challenges, the present paper proposes a new adaptation strategy
using Model Tree based attribute distance to adjust estimation by analogy and
derive new estimates. Using Model Tree has an advantage to deal with
categorical attributes, minimize user interaction and improve efficiency of
model learning through classification. Method: Seven well known datasets have
been used with 3-Fold cross validation to empirically validate the proposed
approach. The proposed method has been investigated using various K analogies
from 1 to 3. Results: Experimental results showed that the proposed approach
produced better results when compared with those obtained by using estimation
by analogy based linear size adaptation, linear similarity adaptation,
'regression towards the mean' and null adaptation. Conclusions: Model Tree
could form a useful extension for estimation by analogy especially for complex
data sets with large number of categorical attributes.
| no_new_dataset | 0.952442 |
1703.04575 | Mohammad Azzeh | Mohammad Azzeh | Dataset Quality Assessment: An extension for analogy based effort
estimation | null | null | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Estimation by Analogy (EBA) is an increasingly active research method in the
area of software engineering. The fundamental assumption of this method is that
the similar projects in terms of attribute values will also be similar in terms
of effort values. It is well recognized that the quality of software datasets
has a considerable impact on the reliability and accuracy of such method.
Therefore, if the software dataset does not satisfy the aforementioned
assumption then it is not rather useful for EBA method. This paper presents a
new method based on Kendall's row-wise rank correlation that enables data
quality evaluation and providing a data preprocessing stage for EBA. The
proposed method provides sound statistical basis and justification for the
process of data quality evaluation. Unlike Analogy-X, our method has the
ability to deal with categorical attributes individually without the need for
partitioning the dataset. Experimental results showed that the proposed method
could form a useful extension for EBA as it enables: dataset quality
evaluation, attribute selection and identifying abnormal observations.
| [
{
"version": "v1",
"created": "Sat, 11 Mar 2017 20:36:17 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Azzeh",
"Mohammad",
""
]
] | TITLE: Dataset Quality Assessment: An extension for analogy based effort
estimation
ABSTRACT: Estimation by Analogy (EBA) is an increasingly active research method in the
area of software engineering. The fundamental assumption of this method is that
the similar projects in terms of attribute values will also be similar in terms
of effort values. It is well recognized that the quality of software datasets
has a considerable impact on the reliability and accuracy of such method.
Therefore, if the software dataset does not satisfy the aforementioned
assumption then it is not rather useful for EBA method. This paper presents a
new method based on Kendall's row-wise rank correlation that enables data
quality evaluation and providing a data preprocessing stage for EBA. The
proposed method provides sound statistical basis and justification for the
process of data quality evaluation. Unlike Analogy-X, our method has the
ability to deal with categorical attributes individually without the need for
partitioning the dataset. Experimental results showed that the proposed method
could form a useful extension for EBA as it enables: dataset quality
evaluation, attribute selection and identifying abnormal observations.
| no_new_dataset | 0.945298 |
1703.05002 | Donghui Wang | Yanan Li, Donghui Wang, Huanhang Hu, Yuetan Lin, Yueting Zhuang | Zero-Shot Recognition using Dual Visual-Semantic Mapping Paths | Accepted as a full paper in IEEE Computer Vision and Pattern
Recognition (CVPR) 2017 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Zero-shot recognition aims to accurately recognize objects of unseen classes
by using a shared visual-semantic mapping between the image feature space and
the semantic embedding space. This mapping is learned on training data of seen
classes and is expected to have transfer ability to unseen classes. In this
paper, we tackle this problem by exploiting the intrinsic relationship between
the semantic space manifold and the transfer ability of visual-semantic
mapping. We formalize their connection and cast zero-shot recognition as a
joint optimization problem. Motivated by this, we propose a novel framework for
zero-shot recognition, which contains dual visual-semantic mapping paths. Our
analysis shows this framework can not only apply prior semantic knowledge to
infer underlying semantic manifold in the image feature space, but also
generate optimized semantic embedding space, which can enhance the transfer
ability of the visual-semantic mapping to unseen classes. The proposed method
is evaluated for zero-shot recognition on four benchmark datasets, achieving
outstanding results.
| [
{
"version": "v1",
"created": "Wed, 15 Mar 2017 08:28:58 GMT"
},
{
"version": "v2",
"created": "Mon, 20 Mar 2017 01:46:27 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Li",
"Yanan",
""
],
[
"Wang",
"Donghui",
""
],
[
"Hu",
"Huanhang",
""
],
[
"Lin",
"Yuetan",
""
],
[
"Zhuang",
"Yueting",
""
]
] | TITLE: Zero-Shot Recognition using Dual Visual-Semantic Mapping Paths
ABSTRACT: Zero-shot recognition aims to accurately recognize objects of unseen classes
by using a shared visual-semantic mapping between the image feature space and
the semantic embedding space. This mapping is learned on training data of seen
classes and is expected to have transfer ability to unseen classes. In this
paper, we tackle this problem by exploiting the intrinsic relationship between
the semantic space manifold and the transfer ability of visual-semantic
mapping. We formalize their connection and cast zero-shot recognition as a
joint optimization problem. Motivated by this, we propose a novel framework for
zero-shot recognition, which contains dual visual-semantic mapping paths. Our
analysis shows this framework can not only apply prior semantic knowledge to
infer underlying semantic manifold in the image feature space, but also
generate optimized semantic embedding space, which can enhance the transfer
ability of the visual-semantic mapping to unseen classes. The proposed method
is evaluated for zero-shot recognition on four benchmark datasets, achieving
outstanding results.
| no_new_dataset | 0.949529 |
1703.05908 | Yao-Hung Tsai | Yao-Hung Hubert Tsai and Liang-Kang Huang and Ruslan Salakhutdinov | Learning Robust Visual-Semantic Embeddings | 12 pages | null | null | null | cs.CV cs.CL cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Many of the existing methods for learning joint embedding of images and text
use only supervised information from paired images and its textual attributes.
Taking advantage of the recent success of unsupervised learning in deep neural
networks, we propose an end-to-end learning framework that is able to extract
more robust multi-modal representations across domains. The proposed method
combines representation learning models (i.e., auto-encoders) together with
cross-domain learning criteria (i.e., Maximum Mean Discrepancy loss) to learn
joint embeddings for semantic and visual features. A novel technique of
unsupervised-data adaptation inference is introduced to construct more
comprehensive embeddings for both labeled and unlabeled data. We evaluate our
method on Animals with Attributes and Caltech-UCSD Birds 200-2011 dataset with
a wide range of applications, including zero and few-shot image recognition and
retrieval, from inductive to transductive settings. Empirically, we show that
our framework improves over the current state of the art on many of the
considered tasks.
| [
{
"version": "v1",
"created": "Fri, 17 Mar 2017 06:59:51 GMT"
},
{
"version": "v2",
"created": "Mon, 20 Mar 2017 00:28:07 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Tsai",
"Yao-Hung Hubert",
""
],
[
"Huang",
"Liang-Kang",
""
],
[
"Salakhutdinov",
"Ruslan",
""
]
] | TITLE: Learning Robust Visual-Semantic Embeddings
ABSTRACT: Many of the existing methods for learning joint embedding of images and text
use only supervised information from paired images and its textual attributes.
Taking advantage of the recent success of unsupervised learning in deep neural
networks, we propose an end-to-end learning framework that is able to extract
more robust multi-modal representations across domains. The proposed method
combines representation learning models (i.e., auto-encoders) together with
cross-domain learning criteria (i.e., Maximum Mean Discrepancy loss) to learn
joint embeddings for semantic and visual features. A novel technique of
unsupervised-data adaptation inference is introduced to construct more
comprehensive embeddings for both labeled and unlabeled data. We evaluate our
method on Animals with Attributes and Caltech-UCSD Birds 200-2011 dataset with
a wide range of applications, including zero and few-shot image recognition and
retrieval, from inductive to transductive settings. Empirically, we show that
our framework improves over the current state of the art on many of the
considered tasks.
| no_new_dataset | 0.940844 |
1703.06151 | Sheng Zou | Sheng Zou, Hao Sun, Alina Zare | Hyperspectral Unmixing with Endmember Variability using Semi-supervised
Partial Membership Latent Dirichlet Allocation | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A semi-supervised Partial Membership Latent Dirichlet Allocation approach is
developed for hyperspectral unmixing and endmember estimation while accounting
for spectral variability and spatial information. Partial Membership Latent
Dirichlet Allocation is an effective approach for spectral unmixing while
representing spectral variability and leveraging spatial information. In this
work, we extend Partial Membership Latent Dirichlet Allocation to incorporate
any available (imprecise) label information to help guide unmixing.
Experimental results on two hyperspectral datasets show that the proposed
semi-supervised PM-LDA can yield improved hyperspectral unmixing and endmember
estimation results.
| [
{
"version": "v1",
"created": "Fri, 17 Mar 2017 18:13:59 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Zou",
"Sheng",
""
],
[
"Sun",
"Hao",
""
],
[
"Zare",
"Alina",
""
]
] | TITLE: Hyperspectral Unmixing with Endmember Variability using Semi-supervised
Partial Membership Latent Dirichlet Allocation
ABSTRACT: A semi-supervised Partial Membership Latent Dirichlet Allocation approach is
developed for hyperspectral unmixing and endmember estimation while accounting
for spectral variability and spatial information. Partial Membership Latent
Dirichlet Allocation is an effective approach for spectral unmixing while
representing spectral variability and leveraging spatial information. In this
work, we extend Partial Membership Latent Dirichlet Allocation to incorporate
any available (imprecise) label information to help guide unmixing.
Experimental results on two hyperspectral datasets show that the proposed
semi-supervised PM-LDA can yield improved hyperspectral unmixing and endmember
estimation results.
| no_new_dataset | 0.949435 |
1703.06300 | Lech Madeyski | Jaros{\l}aw Hryszko and Lech Madeyski and Marta D\k{a}browska and
Piotr Konopka | Defect prediction with bad smells in code | Chapter 10 in Software Engineering: Improving Practice through
Research (B. Hnatkowska and M. \'Smia{\l}ek, eds.), pp. 163-176, 2016 | null | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Background: Defect prediction in software can be highly beneficial for
development projects, when prediction is highly effective and defect-prone
areas are predicted correctly. One of the key elements to gain effective
software defect prediction is proper selection of metrics used for dataset
preparation. Objective: The purpose of this research is to verify, whether code
smells metrics, collected using Microsoft CodeAnalysis tool, added to basic
metric set, can improve defect prediction in industrial software development
project. Results: We verified, if dataset extension by the code smells sourced
metrics, change the effectiveness of the defect prediction by comparing
prediction results for datasets with and without code smells-oriented metrics.
In a result, we observed only small improvement of effectiveness of defect
prediction when dataset extended with bad smells metrics was used: average
accuracy value increased by 0.0091 and stayed within the margin of error.
However, when only use of code smells based metrics were used for prediction
(without basic set of metrics), such process resulted with surprisingly high
accuracy (0.8249) and F-measure (0.8286) results. We also elaborated data
anomalies and problems we observed when two different metric sources were used
to prepare one, consistent set of data. Conclusion: Extending the dataset by
the code smells sourced metric does not significantly improve the prediction
effectiveness. Achieved result did not compensate effort needed to collect
additional metrics. However, we observed that defect prediction based on the
code smells only is still highly effective and can be used especially where
other metrics hardly be used.
| [
{
"version": "v1",
"created": "Sat, 18 Mar 2017 13:55:50 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Hryszko",
"Jarosław",
""
],
[
"Madeyski",
"Lech",
""
],
[
"Dąbrowska",
"Marta",
""
],
[
"Konopka",
"Piotr",
""
]
] | TITLE: Defect prediction with bad smells in code
ABSTRACT: Background: Defect prediction in software can be highly beneficial for
development projects, when prediction is highly effective and defect-prone
areas are predicted correctly. One of the key elements to gain effective
software defect prediction is proper selection of metrics used for dataset
preparation. Objective: The purpose of this research is to verify, whether code
smells metrics, collected using Microsoft CodeAnalysis tool, added to basic
metric set, can improve defect prediction in industrial software development
project. Results: We verified, if dataset extension by the code smells sourced
metrics, change the effectiveness of the defect prediction by comparing
prediction results for datasets with and without code smells-oriented metrics.
In a result, we observed only small improvement of effectiveness of defect
prediction when dataset extended with bad smells metrics was used: average
accuracy value increased by 0.0091 and stayed within the margin of error.
However, when only use of code smells based metrics were used for prediction
(without basic set of metrics), such process resulted with surprisingly high
accuracy (0.8249) and F-measure (0.8286) results. We also elaborated data
anomalies and problems we observed when two different metric sources were used
to prepare one, consistent set of data. Conclusion: Extending the dataset by
the code smells sourced metric does not significantly improve the prediction
effectiveness. Achieved result did not compensate effort needed to collect
additional metrics. However, we observed that defect prediction based on the
code smells only is still highly effective and can be used especially where
other metrics hardly be used.
| no_new_dataset | 0.952131 |
1703.06361 | Lewis Mitchell | James P. Bagrow, Christopher M. Danforth, Lewis Mitchell | Which friends are more popular than you? Contact strength and the
friendship paradox in social networks | null | null | null | null | cs.SI cs.CY physics.soc-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The friendship paradox states that in a social network, egos tend to have
lower degree than their alters, or, "your friends have more friends than you
do". Most research has focused on the friendship paradox and its implications
for information transmission, but treating the network as static and
unweighted. Yet, people can dedicate only a finite fraction of their attention
budget to each social interaction: a high-degree individual may have less time
to dedicate to individual social links, forcing them to modulate the quantities
of contact made to their different social ties. Here we study the friendship
paradox in the context of differing contact volumes between egos and alters,
finding a connection between contact volume and the strength of the friendship
paradox. The most frequently contacted alters exhibit a less pronounced
friendship paradox compared with the ego, whereas less-frequently contacted
alters are more likely to be high degree and give rise to the paradox. We argue
therefore for a more nuanced version of the friendship paradox: "your closest
friends have slightly more friends than you do", and in certain networks even:
"your best friend has no more friends than you do". We demonstrate that this
relationship is robust, holding in both a social media and a mobile phone
dataset. These results have implications for information transfer and influence
in social networks, which we explore using a simple dynamical model.
| [
{
"version": "v1",
"created": "Sat, 18 Mar 2017 22:50:02 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Bagrow",
"James P.",
""
],
[
"Danforth",
"Christopher M.",
""
],
[
"Mitchell",
"Lewis",
""
]
] | TITLE: Which friends are more popular than you? Contact strength and the
friendship paradox in social networks
ABSTRACT: The friendship paradox states that in a social network, egos tend to have
lower degree than their alters, or, "your friends have more friends than you
do". Most research has focused on the friendship paradox and its implications
for information transmission, but treating the network as static and
unweighted. Yet, people can dedicate only a finite fraction of their attention
budget to each social interaction: a high-degree individual may have less time
to dedicate to individual social links, forcing them to modulate the quantities
of contact made to their different social ties. Here we study the friendship
paradox in the context of differing contact volumes between egos and alters,
finding a connection between contact volume and the strength of the friendship
paradox. The most frequently contacted alters exhibit a less pronounced
friendship paradox compared with the ego, whereas less-frequently contacted
alters are more likely to be high degree and give rise to the paradox. We argue
therefore for a more nuanced version of the friendship paradox: "your closest
friends have slightly more friends than you do", and in certain networks even:
"your best friend has no more friends than you do". We demonstrate that this
relationship is robust, holding in both a social media and a mobile phone
dataset. These results have implications for information transfer and influence
in social networks, which we explore using a simple dynamical model.
| no_new_dataset | 0.946794 |
1703.06380 | Shichao Yang | Shichao Yang, Sebastian Scherer | Direct Monocular Odometry Using Points and Lines | ICRA 2017 | null | null | null | cs.CV cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Most visual odometry algorithm for a monocular camera focuses on points,
either by feature matching, or direct alignment of pixel intensity, while
ignoring a common but important geometry entity: edges. In this paper, we
propose an odometry algorithm that combines points and edges to benefit from
the advantages of both direct and feature based methods. It works better in
texture-less environments and is also more robust to lighting changes and fast
motion by increasing the convergence basin. We maintain a depth map for the
keyframe then in the tracking part, the camera pose is recovered by minimizing
both the photometric error and geometric error to the matched edge in a
probabilistic framework. In the mapping part, edge is used to speed up and
increase stereo matching accuracy. On various public datasets, our algorithm
achieves better or comparable performance than state-of-the-art monocular
odometry methods. In some challenging texture-less environments, our algorithm
reduces the state estimation error over 50%.
| [
{
"version": "v1",
"created": "Sun, 19 Mar 2017 01:59:53 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Yang",
"Shichao",
""
],
[
"Scherer",
"Sebastian",
""
]
] | TITLE: Direct Monocular Odometry Using Points and Lines
ABSTRACT: Most visual odometry algorithm for a monocular camera focuses on points,
either by feature matching, or direct alignment of pixel intensity, while
ignoring a common but important geometry entity: edges. In this paper, we
propose an odometry algorithm that combines points and edges to benefit from
the advantages of both direct and feature based methods. It works better in
texture-less environments and is also more robust to lighting changes and fast
motion by increasing the convergence basin. We maintain a depth map for the
keyframe then in the tracking part, the camera pose is recovered by minimizing
both the photometric error and geometric error to the matched edge in a
probabilistic framework. In the mapping part, edge is used to speed up and
increase stereo matching accuracy. On various public datasets, our algorithm
achieves better or comparable performance than state-of-the-art monocular
odometry methods. In some challenging texture-less environments, our algorithm
reduces the state estimation error over 50%.
| no_new_dataset | 0.955402 |
1703.06541 | Shervin Malmasi Ph.D. | Shervin Malmasi and Mark Dras | Native Language Identification using Stacked Generalization | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Ensemble methods using multiple classifiers have proven to be the most
successful approach for the task of Native Language Identification (NLI),
achieving the current state of the art. However, a systematic examination of
ensemble methods for NLI has yet to be conducted. Additionally, deeper ensemble
architectures such as classifier stacking have not been closely evaluated. We
present a set of experiments using three ensemble-based models, testing each
with multiple configurations and algorithms. This includes a rigorous
application of meta-classification models for NLI, achieving state-of-the-art
results on three datasets from different languages. We also present the first
use of statistical significance testing for comparing NLI systems, showing that
our results are significantly better than the previous state of the art. We
make available a collection of test set predictions to facilitate future
statistical tests.
| [
{
"version": "v1",
"created": "Sun, 19 Mar 2017 23:42:28 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Malmasi",
"Shervin",
""
],
[
"Dras",
"Mark",
""
]
] | TITLE: Native Language Identification using Stacked Generalization
ABSTRACT: Ensemble methods using multiple classifiers have proven to be the most
successful approach for the task of Native Language Identification (NLI),
achieving the current state of the art. However, a systematic examination of
ensemble methods for NLI has yet to be conducted. Additionally, deeper ensemble
architectures such as classifier stacking have not been closely evaluated. We
present a set of experiments using three ensemble-based models, testing each
with multiple configurations and algorithms. This includes a rigorous
application of meta-classification models for NLI, achieving state-of-the-art
results on three datasets from different languages. We also present the first
use of statistical significance testing for comparing NLI systems, showing that
our results are significantly better than the previous state of the art. We
make available a collection of test set predictions to facilitate future
statistical tests.
| no_new_dataset | 0.903465 |
1703.06618 | Yuting Hu | Yuting Hu, Liang Zheng, Yi Yang, and Yongfeng Huang | Twitter100k: A Real-world Dataset for Weakly Supervised Cross-Media
Retrieval | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper contributes a new large-scale dataset for weakly supervised
cross-media retrieval, named Twitter100k. Current datasets, such as Wikipedia,
NUS Wide and Flickr30k, have two major limitations. First, these datasets are
lacking in content diversity, i.e., only some pre-defined classes are covered.
Second, texts in these datasets are written in well-organized language, leading
to inconsistency with realistic applications. To overcome these drawbacks, the
proposed Twitter100k dataset is characterized by two aspects: 1) it has 100,000
image-text pairs randomly crawled from Twitter and thus has no constraint in
the image categories; 2) text in Twitter100k is written in informal language by
the users.
Since strongly supervised methods leverage the class labels that may be
missing in practice, this paper focuses on weakly supervised learning for
cross-media retrieval, in which only text-image pairs are exploited during
training. We extensively benchmark the performance of four subspace learning
methods and three variants of the Correspondence AutoEncoder, along with
various text features on Wikipedia, Flickr30k and Twitter100k. Novel insights
are provided. As a minor contribution, inspired by the characteristic of
Twitter100k, we propose an OCR-based cross-media retrieval method. In
experiment, we show that the proposed OCR-based method improves the baseline
performance.
| [
{
"version": "v1",
"created": "Mon, 20 Mar 2017 06:56:33 GMT"
}
] | 2017-03-21T00:00:00 | [
[
"Hu",
"Yuting",
""
],
[
"Zheng",
"Liang",
""
],
[
"Yang",
"Yi",
""
],
[
"Huang",
"Yongfeng",
""
]
] | TITLE: Twitter100k: A Real-world Dataset for Weakly Supervised Cross-Media
Retrieval
ABSTRACT: This paper contributes a new large-scale dataset for weakly supervised
cross-media retrieval, named Twitter100k. Current datasets, such as Wikipedia,
NUS Wide and Flickr30k, have two major limitations. First, these datasets are
lacking in content diversity, i.e., only some pre-defined classes are covered.
Second, texts in these datasets are written in well-organized language, leading
to inconsistency with realistic applications. To overcome these drawbacks, the
proposed Twitter100k dataset is characterized by two aspects: 1) it has 100,000
image-text pairs randomly crawled from Twitter and thus has no constraint in
the image categories; 2) text in Twitter100k is written in informal language by
the users.
Since strongly supervised methods leverage the class labels that may be
missing in practice, this paper focuses on weakly supervised learning for
cross-media retrieval, in which only text-image pairs are exploited during
training. We extensively benchmark the performance of four subspace learning
methods and three variants of the Correspondence AutoEncoder, along with
various text features on Wikipedia, Flickr30k and Twitter100k. Novel insights
are provided. As a minor contribution, inspired by the characteristic of
Twitter100k, we propose an OCR-based cross-media retrieval method. In
experiment, we show that the proposed OCR-based method improves the baseline
performance.
| new_dataset | 0.969957 |
1610.02517 | Ali Bou Nassif | Mohammad Azzeh, Ali Bou Nassif | A Hybrid Model for Estimating Software Project Effort from Use Case
Points | null | null | 10.1016/j.asoc.2016.05.008 | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Early software effort estimation is a hallmark of successful software project
management. Building a reliable effort estimation model usually requires
historical data. Unfortunately, since the information available at early stages
of software development is scarce, it is recommended to use software size
metrics as key cost factor of effort estimation. Use Case Points (UCP) is a
prominent size measure designed mainly for object-oriented projects.
Nevertheless, there are no established models that can translate UCP into its
corresponding effort, therefore, most models use productivity as a second cost
driver. The productivity in those models is usually guessed by experts and does
not depend on historical data, which makes it subject to uncertainty. Thus,
these models were not well examined using a large number of historical data. In
this paper, we designed a hybrid model that consists of classification and
prediction stages using a support vector machine and radial basis neural
networks. The proposed model was constructed over a large number of
observations collected from industrial and student projects. The proposed model
was compared against previous UCP prediction models. The validation and
empirical results demonstrated that the proposed model significantly surpasses
these models on all datasets. The main conclusion is that the environmental
factors of UCP can be used to classify and estimate productivity.
| [
{
"version": "v1",
"created": "Sat, 8 Oct 2016 12:18:18 GMT"
}
] | 2017-03-20T00:00:00 | [
[
"Azzeh",
"Mohammad",
""
],
[
"Nassif",
"Ali Bou",
""
]
] | TITLE: A Hybrid Model for Estimating Software Project Effort from Use Case
Points
ABSTRACT: Early software effort estimation is a hallmark of successful software project
management. Building a reliable effort estimation model usually requires
historical data. Unfortunately, since the information available at early stages
of software development is scarce, it is recommended to use software size
metrics as key cost factor of effort estimation. Use Case Points (UCP) is a
prominent size measure designed mainly for object-oriented projects.
Nevertheless, there are no established models that can translate UCP into its
corresponding effort, therefore, most models use productivity as a second cost
driver. The productivity in those models is usually guessed by experts and does
not depend on historical data, which makes it subject to uncertainty. Thus,
these models were not well examined using a large number of historical data. In
this paper, we designed a hybrid model that consists of classification and
prediction stages using a support vector machine and radial basis neural
networks. The proposed model was constructed over a large number of
observations collected from industrial and student projects. The proposed model
was compared against previous UCP prediction models. The validation and
empirical results demonstrated that the proposed model significantly surpasses
these models on all datasets. The main conclusion is that the environmental
factors of UCP can be used to classify and estimate productivity.
| no_new_dataset | 0.948537 |
1611.06947 | Rongrong Tao | Rongrong Tao, Baojian Zhou, Feng Chen, Naifeng Liu, David Mares,
Patrick Butler, Naren Ramakrishnan | Can Self-Censorship in News Media be Detected Algorithmically? A Case
Study in Latin America | null | null | null | null | cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Censorship in social media has been well studied and provides insight into
how governments stifle freedom of expression online. Comparatively less (or no)
attention has been paid to detecting (self) censorship in traditional media
(e.g., news) using social media as a bellweather. We present a novel
unsupervised approach that views social media as a sensor to detect censorship
in news media wherein statistically significant differences between information
published in the news media and the correlated information published in social
media are automatically identified as candidate censored events. We develop a
hypothesis testing framework to identify and evaluate censored clusters of
keywords, and a new near-linear-time algorithm (called GraphDPD) to identify
the highest scoring clusters as indicators of censorship. We outline extensive
experiments on semi-synthetic data as well as real datasets (with Twitter and
local news media) from Mexico and Venezuela, highlighting the capability to
accurately detect real-world self censorship events.
| [
{
"version": "v1",
"created": "Mon, 21 Nov 2016 18:57:02 GMT"
},
{
"version": "v2",
"created": "Fri, 17 Mar 2017 04:28:53 GMT"
}
] | 2017-03-20T00:00:00 | [
[
"Tao",
"Rongrong",
""
],
[
"Zhou",
"Baojian",
""
],
[
"Chen",
"Feng",
""
],
[
"Liu",
"Naifeng",
""
],
[
"Mares",
"David",
""
],
[
"Butler",
"Patrick",
""
],
[
"Ramakrishnan",
"Naren",
""
]
] | TITLE: Can Self-Censorship in News Media be Detected Algorithmically? A Case
Study in Latin America
ABSTRACT: Censorship in social media has been well studied and provides insight into
how governments stifle freedom of expression online. Comparatively less (or no)
attention has been paid to detecting (self) censorship in traditional media
(e.g., news) using social media as a bellweather. We present a novel
unsupervised approach that views social media as a sensor to detect censorship
in news media wherein statistically significant differences between information
published in the news media and the correlated information published in social
media are automatically identified as candidate censored events. We develop a
hypothesis testing framework to identify and evaluate censored clusters of
keywords, and a new near-linear-time algorithm (called GraphDPD) to identify
the highest scoring clusters as indicators of censorship. We outline extensive
experiments on semi-synthetic data as well as real datasets (with Twitter and
local news media) from Mexico and Venezuela, highlighting the capability to
accurately detect real-world self censorship events.
| no_new_dataset | 0.948394 |
1701.04224 | Chunlin Tian | Chunlin Tian, Weijun Ji | Auxiliary Multimodal LSTM for Audio-visual Speech Recognition and
Lipreading | 8 pages, 4 figures | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The Aduio-visual Speech Recognition (AVSR) which employs both the video and
audio information to do Automatic Speech Recognition (ASR) is one of the
application of multimodal leaning making ASR system more robust and accuracy.
The traditional models usually treated AVSR as inference or projection but
strict prior limits its ability. As the revival of deep learning, Deep Neural
Networks (DNN) becomes an important toolkit in many traditional classification
tasks including ASR, image classification, natural language processing. Some
DNN models were used in AVSR like Multimodal Deep Autoencoders (MDAEs),
Multimodal Deep Belief Network (MDBN) and Multimodal Deep Boltzmann Machine
(MDBM) that actually work better than traditional methods. However, such DNN
models have several shortcomings: (1) They don't balance the modal fusion and
temporal fusion, or even haven't temporal fusion; (2)The architecture of these
models isn't end-to-end, the training and testing getting cumbersome. We
propose a DNN model, Auxiliary Multimodal LSTM (am-LSTM), to overcome such
weakness. The am-LSTM could be trained and tested once, moreover easy to train
and preventing overfitting automatically. The extensibility and flexibility are
also take into consideration. The experiments show that am-LSTM is much better
than traditional methods and other DNN models in three datasets.
| [
{
"version": "v1",
"created": "Mon, 16 Jan 2017 10:08:22 GMT"
},
{
"version": "v2",
"created": "Fri, 17 Mar 2017 14:57:06 GMT"
}
] | 2017-03-20T00:00:00 | [
[
"Tian",
"Chunlin",
""
],
[
"Ji",
"Weijun",
""
]
] | TITLE: Auxiliary Multimodal LSTM for Audio-visual Speech Recognition and
Lipreading
ABSTRACT: The Aduio-visual Speech Recognition (AVSR) which employs both the video and
audio information to do Automatic Speech Recognition (ASR) is one of the
application of multimodal leaning making ASR system more robust and accuracy.
The traditional models usually treated AVSR as inference or projection but
strict prior limits its ability. As the revival of deep learning, Deep Neural
Networks (DNN) becomes an important toolkit in many traditional classification
tasks including ASR, image classification, natural language processing. Some
DNN models were used in AVSR like Multimodal Deep Autoencoders (MDAEs),
Multimodal Deep Belief Network (MDBN) and Multimodal Deep Boltzmann Machine
(MDBM) that actually work better than traditional methods. However, such DNN
models have several shortcomings: (1) They don't balance the modal fusion and
temporal fusion, or even haven't temporal fusion; (2)The architecture of these
models isn't end-to-end, the training and testing getting cumbersome. We
propose a DNN model, Auxiliary Multimodal LSTM (am-LSTM), to overcome such
weakness. The am-LSTM could be trained and tested once, moreover easy to train
and preventing overfitting automatically. The extensibility and flexibility are
also take into consideration. The experiments show that am-LSTM is much better
than traditional methods and other DNN models in three datasets.
| no_new_dataset | 0.942029 |
1703.02769 | Kiran Garimella | Kiran Garimella, Ingmar Weber | A Long-Term Analysis of Polarization on Twitter | This is a preprint of a short paper accepted at ICWSM'17. Please cite
that version instead | null | null | null | cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Social media has played an important role in shaping political discourse over
the last decade. At the same time, it is often perceived to have increased
political polarization, thanks to the scale of discussions and their public
nature.
In this paper, we try to answer the question of whether political
polarization in the US on Twitter has increased over the last eight years. We
analyze a large longitudinal Twitter dataset of 679,000 users and look at signs
of polarization in their (i) network - how people follow political and media
accounts, (ii) tweeting behavior - whether they retweet content from both
sides, and (iii) content - how partisan the hashtags they use are. Our analysis
shows that online polarization has indeed increased over the past eight years
and that, depending on the measure, the relative change is 10%-20%. Our study
is one of very few with such a long-term perspective, encompassing two US
presidential elections and two mid-term elections, providing a rare
longitudinal analysis.
| [
{
"version": "v1",
"created": "Wed, 8 Mar 2017 10:12:45 GMT"
},
{
"version": "v2",
"created": "Fri, 17 Mar 2017 06:15:59 GMT"
}
] | 2017-03-20T00:00:00 | [
[
"Garimella",
"Kiran",
""
],
[
"Weber",
"Ingmar",
""
]
] | TITLE: A Long-Term Analysis of Polarization on Twitter
ABSTRACT: Social media has played an important role in shaping political discourse over
the last decade. At the same time, it is often perceived to have increased
political polarization, thanks to the scale of discussions and their public
nature.
In this paper, we try to answer the question of whether political
polarization in the US on Twitter has increased over the last eight years. We
analyze a large longitudinal Twitter dataset of 679,000 users and look at signs
of polarization in their (i) network - how people follow political and media
accounts, (ii) tweeting behavior - whether they retweet content from both
sides, and (iii) content - how partisan the hashtags they use are. Our analysis
shows that online polarization has indeed increased over the past eight years
and that, depending on the measure, the relative change is 10%-20%. Our study
is one of very few with such a long-term perspective, encompassing two US
presidential elections and two mid-term elections, providing a rare
longitudinal analysis.
| no_new_dataset | 0.911101 |
1703.04564 | Mohammad Azzeh | Mohammad Azzeh, Ali Bou Nassif | Analogy-based effort estimation: a new method to discover set of
analogies from dataset characteristics | null | null | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Analogy-based effort estimation (ABE) is one of the efficient methods for
software effort estimation because of its outstanding performance and
capability of handling noisy datasets. Conventional ABE models usually use the
same number of analogies for all projects in the datasets in order to make good
estimates. The authors' claim is that using same number of analogies may
produce overall best performance for the whole dataset but not necessarily best
performance for each individual project. Therefore there is a need to better
understand the dataset characteristics in order to discover the optimum set of
analogies for each project rather than using a static k nearest projects.
Method: We propose a new technique based on Bisecting k-medoids clustering
algorithm to come up with the best set of analogies for each individual project
before making the prediction. Results & Conclusions: With Bisecting k-medoids
it is possible to better understand the dataset characteristic, and
automatically find best set of analogies for each test project. Performance
figures of the proposed estimation method are promising and better than those
of other regular ABE models
| [
{
"version": "v1",
"created": "Sat, 11 Mar 2017 20:27:08 GMT"
}
] | 2017-03-20T00:00:00 | [
[
"Azzeh",
"Mohammad",
""
],
[
"Nassif",
"Ali Bou",
""
]
] | TITLE: Analogy-based effort estimation: a new method to discover set of
analogies from dataset characteristics
ABSTRACT: Analogy-based effort estimation (ABE) is one of the efficient methods for
software effort estimation because of its outstanding performance and
capability of handling noisy datasets. Conventional ABE models usually use the
same number of analogies for all projects in the datasets in order to make good
estimates. The authors' claim is that using same number of analogies may
produce overall best performance for the whole dataset but not necessarily best
performance for each individual project. Therefore there is a need to better
understand the dataset characteristics in order to discover the optimum set of
analogies for each project rather than using a static k nearest projects.
Method: We propose a new technique based on Bisecting k-medoids clustering
algorithm to come up with the best set of analogies for each individual project
before making the prediction. Results & Conclusions: With Bisecting k-medoids
it is possible to better understand the dataset characteristic, and
automatically find best set of analogies for each test project. Performance
figures of the proposed estimation method are promising and better than those
of other regular ABE models
| no_new_dataset | 0.950549 |
1703.04565 | Mohammad Azzeh | Mohammad Azzeh, Ali Bou Nassif | Fuzzy Model Tree For Early Effort Estimation | null | null | null | null | cs.SE cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Use Case Points (UCP) is a well-known method to estimate the project size,
based on Use Case diagram, at early phases of software development. Although
the Use Case diagram is widely accepted as a de-facto model for analyzing
object oriented software requirements over the world, UCP method did not take
sufficient amount of attention because, as yet, there is no consensus on how to
produce software effort from UCP. This paper aims to study the potential of
using Fuzzy Model Tree to derive effort estimates based on UCP size measure
using a dataset collected for that purpose. The proposed approach has been
validated against Treeboost model, Multiple Linear Regression and classical
effort estimation based on the UCP model. The obtained results are promising
and show better performance than those obtained by classical UCP, Multiple
Linear Regression and slightly better than those obtained by Tree boost model.
| [
{
"version": "v1",
"created": "Sat, 11 Mar 2017 20:24:06 GMT"
}
] | 2017-03-20T00:00:00 | [
[
"Azzeh",
"Mohammad",
""
],
[
"Nassif",
"Ali Bou",
""
]
] | TITLE: Fuzzy Model Tree For Early Effort Estimation
ABSTRACT: Use Case Points (UCP) is a well-known method to estimate the project size,
based on Use Case diagram, at early phases of software development. Although
the Use Case diagram is widely accepted as a de-facto model for analyzing
object oriented software requirements over the world, UCP method did not take
sufficient amount of attention because, as yet, there is no consensus on how to
produce software effort from UCP. This paper aims to study the potential of
using Fuzzy Model Tree to derive effort estimates based on UCP size measure
using a dataset collected for that purpose. The proposed approach has been
validated against Treeboost model, Multiple Linear Regression and classical
effort estimation based on the UCP model. The obtained results are promising
and show better performance than those obtained by classical UCP, Multiple
Linear Regression and slightly better than those obtained by Tree boost model.
| no_new_dataset | 0.899387 |
1703.04567 | Mohammad Azzeh | Mohammad Azzeh, Yousef Elsheikh | Learning best K analogies from data distribution for case-based software
effort estimation | arXiv admin note: substantial text overlap with arXiv: 1703.04564 | null | null | null | cs.SE cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Case-Based Reasoning (CBR) has been widely used to generate good software
effort estimates. The predictive performance of CBR is a dataset dependent and
subject to extremely large space of configuration possibilities. Regardless of
the type of adaptation technique, deciding on the optimal number of similar
cases to be used before applying CBR is a key challenge. In this paper we
propose a new technique based on Bisecting k-medoids clustering algorithm to
better understanding the structure of a dataset and discovering the the optimal
cases for each individual project by excluding irrelevant cases. Results
obtained showed that understanding of the data characteristic prior prediction
stage can help in automatically finding the best number of cases for each test
project. Performance figures of the proposed estimation method are better than
those of other regular K-based CBR methods.
| [
{
"version": "v1",
"created": "Sat, 11 Mar 2017 20:19:05 GMT"
}
] | 2017-03-20T00:00:00 | [
[
"Azzeh",
"Mohammad",
""
],
[
"Elsheikh",
"Yousef",
""
]
] | TITLE: Learning best K analogies from data distribution for case-based software
effort estimation
ABSTRACT: Case-Based Reasoning (CBR) has been widely used to generate good software
effort estimates. The predictive performance of CBR is a dataset dependent and
subject to extremely large space of configuration possibilities. Regardless of
the type of adaptation technique, deciding on the optimal number of similar
cases to be used before applying CBR is a key challenge. In this paper we
propose a new technique based on Bisecting k-medoids clustering algorithm to
better understanding the structure of a dataset and discovering the the optimal
cases for each individual project by excluding irrelevant cases. Results
obtained showed that understanding of the data characteristic prior prediction
stage can help in automatically finding the best number of cases for each test
project. Performance figures of the proposed estimation method are better than
those of other regular K-based CBR methods.
| no_new_dataset | 0.951006 |
1703.04568 | Mohammad Azzeh | Mohammad Azzeh, Ali Bou Nassif, Leandro L Minku | An empirical evaluation of ensemble adjustment methods for analogy-based
effort estimation | null | null | 10.1016/j.jss.2015.01.028 | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Objective: This paper investigates the potential of ensemble learning for
variants of adjustment methods used in analogy-based effort estimation. The
number k of analogies to be used is also investigated. Method We perform a
large scale comparison study where many ensembles constructed from n out of 40
possible valid variants of adjustment methods are applied to eight datasets.
The performance of each method was evaluated based on standardized accuracy and
effect size. Results: The results have been subjected to statistical
significance testing, and show reasonable significant improvements on the
predictive performance where ensemble methods are applied. Conclusion: Our
conclusions suggest that ensembles of adjustment methods can work well and
achieve good performance, even though they are not always superior to single
methods. We also recommend constructing ensembles from only linear adjustment
methods, as they have shown better performance and were frequently ranked
higher.
| [
{
"version": "v1",
"created": "Sat, 11 Mar 2017 20:16:37 GMT"
}
] | 2017-03-20T00:00:00 | [
[
"Azzeh",
"Mohammad",
""
],
[
"Nassif",
"Ali Bou",
""
],
[
"Minku",
"Leandro L",
""
]
] | TITLE: An empirical evaluation of ensemble adjustment methods for analogy-based
effort estimation
ABSTRACT: Objective: This paper investigates the potential of ensemble learning for
variants of adjustment methods used in analogy-based effort estimation. The
number k of analogies to be used is also investigated. Method We perform a
large scale comparison study where many ensembles constructed from n out of 40
possible valid variants of adjustment methods are applied to eight datasets.
The performance of each method was evaluated based on standardized accuracy and
effect size. Results: The results have been subjected to statistical
significance testing, and show reasonable significant improvements on the
predictive performance where ensemble methods are applied. Conclusion: Our
conclusions suggest that ensembles of adjustment methods can work well and
achieve good performance, even though they are not always superior to single
methods. We also recommend constructing ensembles from only linear adjustment
methods, as they have shown better performance and were frequently ranked
higher.
| no_new_dataset | 0.950088 |
1703.05778 | Ali Sharifara | Ali Sharifara, and Amir Ghaderi | Medical Image Watermarking using 2D-DWT with Enhanced security and
capacity | null | null | null | null | cs.MM cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Teleradiology enables medical images to be transferred over the computer
networks for many purposes including clinical interpretation, diagnosis,
archive, etc. In telemedicine, medical images can be manipulated while
transferring. In addition, medical information security requirements are
specified by the legislative rules, and concerned entities must adhere to them.
In this research, we propose a new scheme based on 2-dimensional Discrete
Wavelet Transform (2D DWT) to improve the robustness and authentication of
medical images. In addition, the current research improves security and
capacity of watermarking using encryption and compression in medical images.
The evaluation is performed on the personal dataset, which contains 194 CTI and
68 MRI cases.
| [
{
"version": "v1",
"created": "Thu, 16 Mar 2017 18:05:32 GMT"
}
] | 2017-03-20T00:00:00 | [
[
"Sharifara",
"Ali",
""
],
[
"Ghaderi",
"Amir",
""
]
] | TITLE: Medical Image Watermarking using 2D-DWT with Enhanced security and
capacity
ABSTRACT: Teleradiology enables medical images to be transferred over the computer
networks for many purposes including clinical interpretation, diagnosis,
archive, etc. In telemedicine, medical images can be manipulated while
transferring. In addition, medical information security requirements are
specified by the legislative rules, and concerned entities must adhere to them.
In this research, we propose a new scheme based on 2-dimensional Discrete
Wavelet Transform (2D DWT) to improve the robustness and authentication of
medical images. In addition, the current research improves security and
capacity of watermarking using encryption and compression in medical images.
The evaluation is performed on the personal dataset, which contains 194 CTI and
68 MRI cases.
| new_dataset | 0.956391 |
1703.05819 | Dmytro Karamshuk | Dmytro Karamshuk, Tetyana Lokot, Oleksandr Pryymak, Nishanth Sastry | Identifying Partisan Slant in News Articles and Twitter during Political
Crises | International Conference on Social Informatics (SocInfo 2016) | null | 10.1007/978-3-319-47880-7_16 | null | cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we are interested in understanding the interrelationships
between mainstream and social media in forming public opinion during mass
crises, specifically in regards to how events are framed in the mainstream news
and on social networks and to how the language used in those frames may allow
to infer political slant and partisanship. We study the lingual choices for
political agenda setting in mainstream and social media by analyzing a dataset
of more than 40M tweets and more than 4M news articles from the mass protests
in Ukraine during 2013-2014 - known as "Euromaidan" - and the post-Euromaidan
conflict between Russian, pro-Russian and Ukrainian forces in eastern Ukraine
and Crimea. We design a natural language processing algorithm to analyze at
scale the linguistic markers which point to a particular political leaning in
online media and show that political slant in news articles and Twitter posts
can be inferred with a high level of accuracy. These findings allow us to
better understand the dynamics of partisan opinion formation during mass crises
and the interplay between main- stream and social media in such circumstances.
| [
{
"version": "v1",
"created": "Thu, 16 Mar 2017 21:07:59 GMT"
}
] | 2017-03-20T00:00:00 | [
[
"Karamshuk",
"Dmytro",
""
],
[
"Lokot",
"Tetyana",
""
],
[
"Pryymak",
"Oleksandr",
""
],
[
"Sastry",
"Nishanth",
""
]
] | TITLE: Identifying Partisan Slant in News Articles and Twitter during Political
Crises
ABSTRACT: In this paper, we are interested in understanding the interrelationships
between mainstream and social media in forming public opinion during mass
crises, specifically in regards to how events are framed in the mainstream news
and on social networks and to how the language used in those frames may allow
to infer political slant and partisanship. We study the lingual choices for
political agenda setting in mainstream and social media by analyzing a dataset
of more than 40M tweets and more than 4M news articles from the mass protests
in Ukraine during 2013-2014 - known as "Euromaidan" - and the post-Euromaidan
conflict between Russian, pro-Russian and Ukrainian forces in eastern Ukraine
and Crimea. We design a natural language processing algorithm to analyze at
scale the linguistic markers which point to a particular political leaning in
online media and show that political slant in news articles and Twitter posts
can be inferred with a high level of accuracy. These findings allow us to
better understand the dynamics of partisan opinion formation during mass crises
and the interplay between main- stream and social media in such circumstances.
| no_new_dataset | 0.910067 |
1703.06003 | Huy Phan | Huy Q. Phan, Hongbo Fu, and Antoni B. Chan | Color Orchestra: Ordering Color Palettes for Interpolation and
Prediction | IEEE Transactions on Visualization and Computer Graphics | null | null | null | cs.CV cs.GR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Color theme or color palette can deeply influence the quality and the feeling
of a photograph or a graphical design. Although color palettes may come from
different sources such as online crowd-sourcing, photographs and graphical
designs, in this paper, we consider color palettes extracted from fine art
collections, which we believe to be an abundant source of stylistic and unique
color themes. We aim to capture color styles embedded in these collections by
means of statistical models and to build practical applications upon these
models. As artists often use their personal color themes in their paintings,
making these palettes appear frequently in the dataset, we employed density
estimation to capture the characteristics of palette data. Via density
estimation, we carried out various predictions and interpolations on palettes,
which led to promising applications such as photo-style exploration, real-time
color suggestion, and enriched photo recolorization. It was, however,
challenging to apply density estimation to palette data as palettes often come
as unordered sets of colors, which make it difficult to use conventional
metrics on them. To this end, we developed a divide-and-conquer sorting
algorithm to rearrange the colors in the palettes in a coherent order, which
allows meaningful interpolation between color palettes. To confirm the
performance of our model, we also conducted quantitative experiments on
datasets of digitized paintings collected from the Internet and received
favorable results.
| [
{
"version": "v1",
"created": "Fri, 17 Mar 2017 13:25:49 GMT"
}
] | 2017-03-20T00:00:00 | [
[
"Phan",
"Huy Q.",
""
],
[
"Fu",
"Hongbo",
""
],
[
"Chan",
"Antoni B.",
""
]
] | TITLE: Color Orchestra: Ordering Color Palettes for Interpolation and
Prediction
ABSTRACT: Color theme or color palette can deeply influence the quality and the feeling
of a photograph or a graphical design. Although color palettes may come from
different sources such as online crowd-sourcing, photographs and graphical
designs, in this paper, we consider color palettes extracted from fine art
collections, which we believe to be an abundant source of stylistic and unique
color themes. We aim to capture color styles embedded in these collections by
means of statistical models and to build practical applications upon these
models. As artists often use their personal color themes in their paintings,
making these palettes appear frequently in the dataset, we employed density
estimation to capture the characteristics of palette data. Via density
estimation, we carried out various predictions and interpolations on palettes,
which led to promising applications such as photo-style exploration, real-time
color suggestion, and enriched photo recolorization. It was, however,
challenging to apply density estimation to palette data as palettes often come
as unordered sets of colors, which make it difficult to use conventional
metrics on them. To this end, we developed a divide-and-conquer sorting
algorithm to rearrange the colors in the palettes in a coherent order, which
allows meaningful interpolation between color palettes. To confirm the
performance of our model, we also conducted quantitative experiments on
datasets of digitized paintings collected from the Internet and received
favorable results.
| no_new_dataset | 0.955858 |
1703.06063 | Mansaf Alam Dr | Samiya Khan, Mansaf Alam | Outcome-Based Quality Assessment Framework for Higher Education | null | null | null | null | cs.CY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This research paper proposes a quality framework for higher education that
evaluates the performance of institutions on the basis of performance of
outgoing students. Literature was surveyed to evaluate existing quality
frameworks and develop a framework that provides insights on an unexplored
dimension of quality. In order to implement and test the framework, cloud-based
big data technology, BigQuery, was used with R to perform analytics. It was
found that how the students fair after passing out of a course is the outcome
of educational process. This aspect can also be used as a quality metric for
performance evaluation and management of educational organizations. However, it
has not been taken into account in existing research. The lack of an integrated
data collection system and rich datasets for educational intelligence
applications, are some of the limitations that plague this area of research.
Educational organizations are responsible for the performance of their students
even after they complete their course. The inclusion of this dimension to
quality assessment shall allow evaluation of educational institutions on these
grounds. Assurance of this quality dimension shall boost enrolments in
postgraduate and research degrees. Moreover, educational institutions will be
motivated to groom students for placements or higher studies.
| [
{
"version": "v1",
"created": "Wed, 15 Mar 2017 09:59:23 GMT"
}
] | 2017-03-20T00:00:00 | [
[
"Khan",
"Samiya",
""
],
[
"Alam",
"Mansaf",
""
]
] | TITLE: Outcome-Based Quality Assessment Framework for Higher Education
ABSTRACT: This research paper proposes a quality framework for higher education that
evaluates the performance of institutions on the basis of performance of
outgoing students. Literature was surveyed to evaluate existing quality
frameworks and develop a framework that provides insights on an unexplored
dimension of quality. In order to implement and test the framework, cloud-based
big data technology, BigQuery, was used with R to perform analytics. It was
found that how the students fair after passing out of a course is the outcome
of educational process. This aspect can also be used as a quality metric for
performance evaluation and management of educational organizations. However, it
has not been taken into account in existing research. The lack of an integrated
data collection system and rich datasets for educational intelligence
applications, are some of the limitations that plague this area of research.
Educational organizations are responsible for the performance of their students
even after they complete their course. The inclusion of this dimension to
quality assessment shall allow evaluation of educational institutions on these
grounds. Assurance of this quality dimension shall boost enrolments in
postgraduate and research degrees. Moreover, educational institutions will be
motivated to groom students for placements or higher studies.
| no_new_dataset | 0.940572 |
1703.06108 | Nemanja Spasojevic | Prantik Bhattacharyya, Nemanja Spasojevic | Global Entity Ranking Across Multiple Languages | 2 Pages, 1 Figure, 2 Tables, WWW2017 Companion, WWW 2017 Companion | null | 10.1145/3041021.3054213 | null | cs.IR cs.CL cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present work on building a global long-tailed ranking of entities across
multiple languages using Wikipedia and Freebase knowledge bases. We identify
multiple features and build a model to rank entities using a ground-truth
dataset of more than 10 thousand labels. The final system ranks 27 million
entities with 75% precision and 48% F1 score. We provide performance evaluation
and empirical evidence of the quality of ranking across languages, and open the
final ranked lists for future research.
| [
{
"version": "v1",
"created": "Fri, 17 Mar 2017 17:16:02 GMT"
}
] | 2017-03-20T00:00:00 | [
[
"Bhattacharyya",
"Prantik",
""
],
[
"Spasojevic",
"Nemanja",
""
]
] | TITLE: Global Entity Ranking Across Multiple Languages
ABSTRACT: We present work on building a global long-tailed ranking of entities across
multiple languages using Wikipedia and Freebase knowledge bases. We identify
multiple features and build a model to rank entities using a ground-truth
dataset of more than 10 thousand labels. The final system ranks 27 million
entities with 75% precision and 48% F1 score. We provide performance evaluation
and empirical evidence of the quality of ranking across languages, and open the
final ranked lists for future research.
| no_new_dataset | 0.941547 |
1510.02969 | Pooya Khorrami | Pooya Khorrami, Tom Le Paine, Thomas S. Huang | Do Deep Neural Networks Learn Facial Action Units When Doing Expression
Recognition? | Accepted at ICCV 2015 CV4AC Workshop. Corrected numbers in Tables 2
and 3 | null | null | null | cs.CV cs.LG cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Despite being the appearance-based classifier of choice in recent years,
relatively few works have examined how much convolutional neural networks
(CNNs) can improve performance on accepted expression recognition benchmarks
and, more importantly, examine what it is they actually learn. In this work,
not only do we show that CNNs can achieve strong performance, but we also
introduce an approach to decipher which portions of the face influence the
CNN's predictions. First, we train a zero-bias CNN on facial expression data
and achieve, to our knowledge, state-of-the-art performance on two expression
recognition benchmarks: the extended Cohn-Kanade (CK+) dataset and the Toronto
Face Dataset (TFD). We then qualitatively analyze the network by visualizing
the spatial patterns that maximally excite different neurons in the
convolutional layers and show how they resemble Facial Action Units (FAUs).
Finally, we use the FAU labels provided in the CK+ dataset to verify that the
FAUs observed in our filter visualizations indeed align with the subject's
facial movements.
| [
{
"version": "v1",
"created": "Sat, 10 Oct 2015 18:53:21 GMT"
},
{
"version": "v2",
"created": "Fri, 28 Oct 2016 06:12:07 GMT"
},
{
"version": "v3",
"created": "Thu, 16 Mar 2017 03:07:21 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Khorrami",
"Pooya",
""
],
[
"Paine",
"Tom Le",
""
],
[
"Huang",
"Thomas S.",
""
]
] | TITLE: Do Deep Neural Networks Learn Facial Action Units When Doing Expression
Recognition?
ABSTRACT: Despite being the appearance-based classifier of choice in recent years,
relatively few works have examined how much convolutional neural networks
(CNNs) can improve performance on accepted expression recognition benchmarks
and, more importantly, examine what it is they actually learn. In this work,
not only do we show that CNNs can achieve strong performance, but we also
introduce an approach to decipher which portions of the face influence the
CNN's predictions. First, we train a zero-bias CNN on facial expression data
and achieve, to our knowledge, state-of-the-art performance on two expression
recognition benchmarks: the extended Cohn-Kanade (CK+) dataset and the Toronto
Face Dataset (TFD). We then qualitatively analyze the network by visualizing
the spatial patterns that maximally excite different neurons in the
convolutional layers and show how they resemble Facial Action Units (FAUs).
Finally, we use the FAU labels provided in the CK+ dataset to verify that the
FAUs observed in our filter visualizations indeed align with the subject's
facial movements.
| no_new_dataset | 0.939582 |
1604.07513 | Hirokatsu Kataoka | Teppei Suzuki, Soma Shirakabe, Yudai Miyashita, Akio Nakamura, Yutaka
Satoh, Hirokatsu Kataoka | Semantic Change Detection with Hypermaps | null | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Change detection is the study of detecting changes between two different
images of a scene taken at different times. By the detected change areas,
however, a human cannot understand how different the two images. Therefore, a
semantic understanding is required in the change detection research such as
disaster investigation. The paper proposes the concept of semantic change
detection, which involves intuitively inserting semantic meaning into detected
change areas. We mainly focus on the novel semantic segmentation in addition to
a conventional change detection approach. In order to solve this problem and
obtain a high-level of performance, we propose an improvement to the
hypercolumns representation, hereafter known as hypermaps, which effectively
uses convolutional maps obtained from convolutional neural networks (CNNs). We
also employ multi-scale feature representation captured by different image
patches. We applied our method to the TSUNAMI Panoramic Change Detection
dataset, and re-annotated the changed areas of the dataset via semantic
classes. The results show that our multi-scale hypermaps provided outstanding
performance on the re-annotated TSUNAMI dataset.
| [
{
"version": "v1",
"created": "Tue, 26 Apr 2016 04:31:31 GMT"
},
{
"version": "v2",
"created": "Thu, 16 Mar 2017 01:46:37 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Suzuki",
"Teppei",
""
],
[
"Shirakabe",
"Soma",
""
],
[
"Miyashita",
"Yudai",
""
],
[
"Nakamura",
"Akio",
""
],
[
"Satoh",
"Yutaka",
""
],
[
"Kataoka",
"Hirokatsu",
""
]
] | TITLE: Semantic Change Detection with Hypermaps
ABSTRACT: Change detection is the study of detecting changes between two different
images of a scene taken at different times. By the detected change areas,
however, a human cannot understand how different the two images. Therefore, a
semantic understanding is required in the change detection research such as
disaster investigation. The paper proposes the concept of semantic change
detection, which involves intuitively inserting semantic meaning into detected
change areas. We mainly focus on the novel semantic segmentation in addition to
a conventional change detection approach. In order to solve this problem and
obtain a high-level of performance, we propose an improvement to the
hypercolumns representation, hereafter known as hypermaps, which effectively
uses convolutional maps obtained from convolutional neural networks (CNNs). We
also employ multi-scale feature representation captured by different image
patches. We applied our method to the TSUNAMI Panoramic Change Detection
dataset, and re-annotated the changed areas of the dataset via semantic
classes. The results show that our multi-scale hypermaps provided outstanding
performance on the re-annotated TSUNAMI dataset.
| no_new_dataset | 0.951006 |
1607.03476 | Paul Henderson | Paul Henderson, Vittorio Ferrari | End-to-end training of object class detectors for mean average precision | This version has minor additions to results (ablation study) and
discussion | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a method for training CNN-based object class detectors directly
using mean average precision (mAP) as the training loss, in a truly end-to-end
fashion that includes non-maximum suppression (NMS) at training time. This
contrasts with the traditional approach of training a CNN for a window
classification loss, then applying NMS only at test time, when mAP is used as
the evaluation metric in place of classification accuracy. However, mAP
following NMS forms a piecewise-constant structured loss over thousands of
windows, with gradients that do not convey useful information for gradient
descent. Hence, we define new, general gradient-like quantities for piecewise
constant functions, which have wide applicability. We describe how to calculate
these efficiently for mAP following NMS, enabling to train a detector based on
Fast R-CNN directly for mAP. This model achieves equivalent performance to the
standard Fast R-CNN on the PASCAL VOC 2007 and 2012 datasets, while being
conceptually more appealing as the very same model and loss are used at both
training and test time.
| [
{
"version": "v1",
"created": "Tue, 12 Jul 2016 19:45:12 GMT"
},
{
"version": "v2",
"created": "Thu, 16 Mar 2017 13:55:07 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Henderson",
"Paul",
""
],
[
"Ferrari",
"Vittorio",
""
]
] | TITLE: End-to-end training of object class detectors for mean average precision
ABSTRACT: We present a method for training CNN-based object class detectors directly
using mean average precision (mAP) as the training loss, in a truly end-to-end
fashion that includes non-maximum suppression (NMS) at training time. This
contrasts with the traditional approach of training a CNN for a window
classification loss, then applying NMS only at test time, when mAP is used as
the evaluation metric in place of classification accuracy. However, mAP
following NMS forms a piecewise-constant structured loss over thousands of
windows, with gradients that do not convey useful information for gradient
descent. Hence, we define new, general gradient-like quantities for piecewise
constant functions, which have wide applicability. We describe how to calculate
these efficiently for mAP following NMS, enabling to train a detector based on
Fast R-CNN directly for mAP. This model achieves equivalent performance to the
standard Fast R-CNN on the PASCAL VOC 2007 and 2012 datasets, while being
conceptually more appealing as the very same model and loss are used at both
training and test time.
| no_new_dataset | 0.952042 |
1608.03542 | Daniel Hewlett | Daniel Hewlett, Alexandre Lacoste, Llion Jones, Illia Polosukhin,
Andrew Fandrianto, Jay Han, Matthew Kelcey, David Berthelot | WikiReading: A Novel Large-scale Language Understanding Task over
Wikipedia | null | Proceedings of the 54th Annual Meeting of the Association for
Computational Linguistics (Volume 1: Long Papers), 2016, pp. 1535-1545 | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present WikiReading, a large-scale natural language understanding task and
publicly-available dataset with 18 million instances. The task is to predict
textual values from the structured knowledge base Wikidata by reading the text
of the corresponding Wikipedia articles. The task contains a rich variety of
challenging classification and extraction sub-tasks, making it well-suited for
end-to-end models such as deep neural networks (DNNs). We compare various
state-of-the-art DNN-based architectures for document classification,
information extraction, and question answering. We find that models supporting
a rich answer space, such as word or character sequences, perform best. Our
best-performing model, a word-level sequence to sequence model with a mechanism
to copy out-of-vocabulary words, obtains an accuracy of 71.8%.
| [
{
"version": "v1",
"created": "Thu, 11 Aug 2016 17:34:12 GMT"
},
{
"version": "v2",
"created": "Wed, 15 Mar 2017 19:58:44 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Hewlett",
"Daniel",
""
],
[
"Lacoste",
"Alexandre",
""
],
[
"Jones",
"Llion",
""
],
[
"Polosukhin",
"Illia",
""
],
[
"Fandrianto",
"Andrew",
""
],
[
"Han",
"Jay",
""
],
[
"Kelcey",
"Matthew",
""
],
[
"Berthelot",
"David",
""
]
] | TITLE: WikiReading: A Novel Large-scale Language Understanding Task over
Wikipedia
ABSTRACT: We present WikiReading, a large-scale natural language understanding task and
publicly-available dataset with 18 million instances. The task is to predict
textual values from the structured knowledge base Wikidata by reading the text
of the corresponding Wikipedia articles. The task contains a rich variety of
challenging classification and extraction sub-tasks, making it well-suited for
end-to-end models such as deep neural networks (DNNs). We compare various
state-of-the-art DNN-based architectures for document classification,
information extraction, and question answering. We find that models supporting
a rich answer space, such as word or character sequences, perform best. Our
best-performing model, a word-level sequence to sequence model with a mechanism
to copy out-of-vocabulary words, obtains an accuracy of 71.8%.
| new_dataset | 0.960842 |
1612.07597 | Andreas Henelius | Andreas Henelius, Antti Ukkonen, Kai Puolam\"aki | Finding Statistically Significant Attribute Interactions | 9 pages, 4 tables, 1 figure | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In many data exploration tasks it is meaningful to identify groups of
attribute interactions that are specific to a variable of interest. For
instance, in a dataset where the attributes are medical markers and the
variable of interest (class variable) is binary indicating presence/absence of
disease, we would like to know which medical markers interact with respect to
the binary class label. These interactions are useful in several practical
applications, for example, to gain insight into the structure of the data, in
feature selection, and in data anonymisation. We present a novel method, based
on statistical significance testing, that can be used to verify if the data set
has been created by a given factorised class-conditional joint distribution,
where the distribution is parametrised by a partition of its attributes.
Furthermore, we provide a method, named ASTRID, for automatically finding a
partition of attributes describing the distribution that has generated the
data. State-of-the-art classifiers are utilised to capture the interactions
present in the data by systematically breaking attribute interactions and
observing the effect of this breaking on classifier performance. We empirically
demonstrate the utility of the proposed method with examples using real and
synthetic data.
| [
{
"version": "v1",
"created": "Thu, 22 Dec 2016 13:53:42 GMT"
},
{
"version": "v2",
"created": "Thu, 16 Mar 2017 12:21:36 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Henelius",
"Andreas",
""
],
[
"Ukkonen",
"Antti",
""
],
[
"Puolamäki",
"Kai",
""
]
] | TITLE: Finding Statistically Significant Attribute Interactions
ABSTRACT: In many data exploration tasks it is meaningful to identify groups of
attribute interactions that are specific to a variable of interest. For
instance, in a dataset where the attributes are medical markers and the
variable of interest (class variable) is binary indicating presence/absence of
disease, we would like to know which medical markers interact with respect to
the binary class label. These interactions are useful in several practical
applications, for example, to gain insight into the structure of the data, in
feature selection, and in data anonymisation. We present a novel method, based
on statistical significance testing, that can be used to verify if the data set
has been created by a given factorised class-conditional joint distribution,
where the distribution is parametrised by a partition of its attributes.
Furthermore, we provide a method, named ASTRID, for automatically finding a
partition of attributes describing the distribution that has generated the
data. State-of-the-art classifiers are utilised to capture the interactions
present in the data by systematically breaking attribute interactions and
observing the effect of this breaking on classifier performance. We empirically
demonstrate the utility of the proposed method with examples using real and
synthetic data.
| no_new_dataset | 0.945349 |
1702.02628 | Amin Ghafouri | Amin Ghafouri, Aron Laszka, Abhishek Dubey, and Xenofon Koutsoukos | Optimal Detection of Faulty Traffic Sensors Used in Route Planning | Proceedings of The 2nd Workshop on Science of Smart City Operations
and Platforms Engineering (SCOPE 2017), Pittsburgh, PA USA, April 2017, 6
pages | null | 10.1145/3063386.3063767 | null | cs.AI cs.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In a smart city, real-time traffic sensors may be deployed for various
applications, such as route planning. Unfortunately, sensors are prone to
failures, which result in erroneous traffic data. Erroneous data can adversely
affect applications such as route planning, and can cause increased travel
time. To minimize the impact of sensor failures, we must detect them promptly
and accurately. However, typical detection algorithms may lead to a large
number of false positives (i.e., false alarms) and false negatives (i.e.,
missed detections), which can result in suboptimal route planning. In this
paper, we devise an effective detector for identifying faulty traffic sensors
using a prediction model based on Gaussian Processes. Further, we present an
approach for computing the optimal parameters of the detector which minimize
losses due to false-positive and false-negative errors. We also characterize
critical sensors, whose failure can have high impact on the route planning
application. Finally, we implement our method and evaluate it numerically using
a real-world dataset and the route planning platform OpenTripPlanner.
| [
{
"version": "v1",
"created": "Wed, 8 Feb 2017 21:49:46 GMT"
},
{
"version": "v2",
"created": "Thu, 16 Mar 2017 16:37:47 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Ghafouri",
"Amin",
""
],
[
"Laszka",
"Aron",
""
],
[
"Dubey",
"Abhishek",
""
],
[
"Koutsoukos",
"Xenofon",
""
]
] | TITLE: Optimal Detection of Faulty Traffic Sensors Used in Route Planning
ABSTRACT: In a smart city, real-time traffic sensors may be deployed for various
applications, such as route planning. Unfortunately, sensors are prone to
failures, which result in erroneous traffic data. Erroneous data can adversely
affect applications such as route planning, and can cause increased travel
time. To minimize the impact of sensor failures, we must detect them promptly
and accurately. However, typical detection algorithms may lead to a large
number of false positives (i.e., false alarms) and false negatives (i.e.,
missed detections), which can result in suboptimal route planning. In this
paper, we devise an effective detector for identifying faulty traffic sensors
using a prediction model based on Gaussian Processes. Further, we present an
approach for computing the optimal parameters of the detector which minimize
losses due to false-positive and false-negative errors. We also characterize
critical sensors, whose failure can have high impact on the route planning
application. Finally, we implement our method and evaluate it numerically using
a real-world dataset and the route planning platform OpenTripPlanner.
| no_new_dataset | 0.951818 |
1703.04103 | Grigorios Kalliatakis M.A. | Grigorios Kalliatakis, Shoaib Ehsan, Maria Fasli, Ales Leonardis,
Juergen Gall and Klaus D. McDonald-Maier | Detection of Human Rights Violations in Images: Can Convolutional Neural
Networks help? | In Proceedings of the 12th International Conference on Computer
Vision Theory and Applications (VISAPP 2017), 8 pages | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | After setting the performance benchmarks for image, video, speech and audio
processing, deep convolutional networks have been core to the greatest advances
in image recognition tasks in recent times. This raises the question of whether
there are any benefit in targeting these remarkable deep architectures with the
unattempted task of recognising human rights violations through digital images.
Under this perspective, we introduce a new, well-sampled human rights-centric
dataset called Human Rights Understanding (HRUN). We conduct a rigorous
evaluation on a common ground by combining this dataset with different
state-of-the-art deep convolutional architectures in order to achieve
recognition of human rights violations. Experimental results on the HRUN
dataset have shown that the best performing CNN architectures can achieve up to
88.10\% mean average precision. Additionally, our experiments demonstrate that
increasing the size of the training samples is crucial for achieving an
improvement on mean average precision principally when utilising very deep
networks.
| [
{
"version": "v1",
"created": "Sun, 12 Mar 2017 11:39:41 GMT"
},
{
"version": "v2",
"created": "Thu, 16 Mar 2017 10:37:25 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Kalliatakis",
"Grigorios",
""
],
[
"Ehsan",
"Shoaib",
""
],
[
"Fasli",
"Maria",
""
],
[
"Leonardis",
"Ales",
""
],
[
"Gall",
"Juergen",
""
],
[
"McDonald-Maier",
"Klaus D.",
""
]
] | TITLE: Detection of Human Rights Violations in Images: Can Convolutional Neural
Networks help?
ABSTRACT: After setting the performance benchmarks for image, video, speech and audio
processing, deep convolutional networks have been core to the greatest advances
in image recognition tasks in recent times. This raises the question of whether
there are any benefit in targeting these remarkable deep architectures with the
unattempted task of recognising human rights violations through digital images.
Under this perspective, we introduce a new, well-sampled human rights-centric
dataset called Human Rights Understanding (HRUN). We conduct a rigorous
evaluation on a common ground by combining this dataset with different
state-of-the-art deep convolutional architectures in order to achieve
recognition of human rights violations. Experimental results on the HRUN
dataset have shown that the best performing CNN architectures can achieve up to
88.10\% mean average precision. Additionally, our experiments demonstrate that
increasing the size of the training samples is crucial for achieving an
improvement on mean average precision principally when utilising very deep
networks.
| new_dataset | 0.964888 |
1703.04215 | Jinliang Xu | Jinliang Xu, Shangguang Wang, Fangchun Yang, Jie Tang | Multiple User Context Inference by Fusing Data Sources | This paper has been withdrawn by the author due to a crucial sign
error in some equations and figures | null | null | null | cs.IR cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Inference of user context information, including user's gender, age, marital
status, location and so on, has been proven to be valuable for building context
aware recommender system. However, prevalent existing studies on user context
inference have two shortcommings: 1. focusing on only a single data source
(e.g. Internet browsing logs, or mobile call records), and 2. ignoring the
interdependence of multiple user contexts (e.g. interdependence between age and
marital status), which have led to poor inference performance. To solve this
problem, in this paper, we first exploit tensor outer product to fuse multiple
data sources in the feature space to obtain an extensional user feature
representation. Following this, by taking this extensional user feature
representation as input, we propose a multiple attribute probabilistic model
called MulAProM to infer user contexts that can take advantage of the
interdependence between them. Our study is based on large telecommunication
datasets from the local mobile operator of Shanghai, China, and consists of two
data sources, 4.6 million call detail records and 7.5 million data traffic
records of 8,000 mobile users, collected in the course of six months. The
experimental results show that our model can outperform other models in terms
of \emph{recall}, \emph{precision}, and the \emph{F1-measure}.
| [
{
"version": "v1",
"created": "Mon, 13 Mar 2017 01:23:17 GMT"
},
{
"version": "v2",
"created": "Thu, 16 Mar 2017 15:06:20 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Xu",
"Jinliang",
""
],
[
"Wang",
"Shangguang",
""
],
[
"Yang",
"Fangchun",
""
],
[
"Tang",
"Jie",
""
]
] | TITLE: Multiple User Context Inference by Fusing Data Sources
ABSTRACT: Inference of user context information, including user's gender, age, marital
status, location and so on, has been proven to be valuable for building context
aware recommender system. However, prevalent existing studies on user context
inference have two shortcommings: 1. focusing on only a single data source
(e.g. Internet browsing logs, or mobile call records), and 2. ignoring the
interdependence of multiple user contexts (e.g. interdependence between age and
marital status), which have led to poor inference performance. To solve this
problem, in this paper, we first exploit tensor outer product to fuse multiple
data sources in the feature space to obtain an extensional user feature
representation. Following this, by taking this extensional user feature
representation as input, we propose a multiple attribute probabilistic model
called MulAProM to infer user contexts that can take advantage of the
interdependence between them. Our study is based on large telecommunication
datasets from the local mobile operator of Shanghai, China, and consists of two
data sources, 4.6 million call detail records and 7.5 million data traffic
records of 8,000 mobile users, collected in the course of six months. The
experimental results show that our model can outperform other models in terms
of \emph{recall}, \emph{precision}, and the \emph{F1-measure}.
| no_new_dataset | 0.947527 |
1703.04216 | Jinliang Xu | Jinliang Xu, Shangguang Wang, Fangchun Yang, Rong N. Chang | Cognitive Inference of Demographic Data by User Ratings | This paper has been withdrawn by the author due to a crucial sign
error in some equations and figures | null | null | null | cs.IR cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Cognitive inference of user demographics, such as gender and age, plays an
important role in creating user profiles for adjusting marketing strategies and
generating personalized recommendations because user demographic data is
usually not available due to data privacy concerns. At present, users can
readily express feedback regarding products or services that they have
purchased. During this process, user demographics are concealed, but the data
has never yet been successfully utilized to contribute to the cognitive
inference of user demographics. In this paper, we investigate the inference
power of user ratings data, and propose a simple yet general cognitive
inference model, called rating to profile (R2P), to infer user demographics
from user provided ratings. In particular, the proposed R2P model can achieve
the following: 1. Correctly integrate user ratings into model training. 2.Infer
multiple demographic attributes of users simultaneously, capturing the
underlying relevance between different demographic attributes. 3. Train its two
components, i.e. feature extractor and classifier, in an integrated manner
under a supervised learning paradigm, which effectively helps to discover
useful hidden patterns from highly sparse ratings data. We introduce how to
incorporate user ratings data into the research field of cognitive inference of
user demographic data, and detail the model development and optimization
process for the proposed R2P. Extensive experiments are conducted on two
real-world ratings datasets against various compared state-of-the-art methods,
and the results from multiple aspects demonstrate that our proposed R2P model
can significantly improve on the cognitive inference performance of user
demographic data.
| [
{
"version": "v1",
"created": "Mon, 13 Mar 2017 01:23:31 GMT"
},
{
"version": "v2",
"created": "Thu, 16 Mar 2017 15:07:33 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Xu",
"Jinliang",
""
],
[
"Wang",
"Shangguang",
""
],
[
"Yang",
"Fangchun",
""
],
[
"Chang",
"Rong N.",
""
]
] | TITLE: Cognitive Inference of Demographic Data by User Ratings
ABSTRACT: Cognitive inference of user demographics, such as gender and age, plays an
important role in creating user profiles for adjusting marketing strategies and
generating personalized recommendations because user demographic data is
usually not available due to data privacy concerns. At present, users can
readily express feedback regarding products or services that they have
purchased. During this process, user demographics are concealed, but the data
has never yet been successfully utilized to contribute to the cognitive
inference of user demographics. In this paper, we investigate the inference
power of user ratings data, and propose a simple yet general cognitive
inference model, called rating to profile (R2P), to infer user demographics
from user provided ratings. In particular, the proposed R2P model can achieve
the following: 1. Correctly integrate user ratings into model training. 2.Infer
multiple demographic attributes of users simultaneously, capturing the
underlying relevance between different demographic attributes. 3. Train its two
components, i.e. feature extractor and classifier, in an integrated manner
under a supervised learning paradigm, which effectively helps to discover
useful hidden patterns from highly sparse ratings data. We introduce how to
incorporate user ratings data into the research field of cognitive inference of
user demographic data, and detail the model development and optimization
process for the proposed R2P. Extensive experiments are conducted on two
real-world ratings datasets against various compared state-of-the-art methods,
and the results from multiple aspects demonstrate that our proposed R2P model
can significantly improve on the cognitive inference performance of user
demographic data.
| no_new_dataset | 0.943712 |
1703.05400 | Shin-Ming Cheng | Shin-Ming Cheng and Pin-Yu Chen and Ching-Chao Lin and Hsu-Chun Hsiao | Traffic-aware Patching for Cyber Security in Mobile IoT | 8 pages, 6 figures, To appear in July 2017 IEEE Communications
Magazine, feature topic on "Traffic Measurements for Cyber Security" | null | null | null | cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The various types of communication technologies and mobility features in
Internet of Things (IoT) on the one hand enable fruitful and attractive
applications, but on the other hand facilitates malware propagation, thereby
raising new challenges on handling IoT-empowered malware for cyber security.
Comparing with the malware propagation control scheme in traditional wireless
networks where nodes can be directly repaired and secured, in IoT, compromised
end devices are difficult to be patched. Alternatively, blocking malware via
patching intermediate nodes turns out to be a more feasible and practical
solution. Specifically, patching intermediate nodes can effectively prevent the
proliferation of malware propagation by securing infrastructure links and
limiting malware propagation to local device-to-device dissemination. This
article proposes a novel traffic-aware patching scheme to select important
intermediate nodes to patch, which applies to the IoT system with limited
patching resources and response time constraint. Experiments on real-world
trace datasets in IoT networks are conducted to demonstrate the advantage of
the proposed traffic-aware patching scheme in alleviating malware propagation.
| [
{
"version": "v1",
"created": "Wed, 15 Mar 2017 21:59:05 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Cheng",
"Shin-Ming",
""
],
[
"Chen",
"Pin-Yu",
""
],
[
"Lin",
"Ching-Chao",
""
],
[
"Hsiao",
"Hsu-Chun",
""
]
] | TITLE: Traffic-aware Patching for Cyber Security in Mobile IoT
ABSTRACT: The various types of communication technologies and mobility features in
Internet of Things (IoT) on the one hand enable fruitful and attractive
applications, but on the other hand facilitates malware propagation, thereby
raising new challenges on handling IoT-empowered malware for cyber security.
Comparing with the malware propagation control scheme in traditional wireless
networks where nodes can be directly repaired and secured, in IoT, compromised
end devices are difficult to be patched. Alternatively, blocking malware via
patching intermediate nodes turns out to be a more feasible and practical
solution. Specifically, patching intermediate nodes can effectively prevent the
proliferation of malware propagation by securing infrastructure links and
limiting malware propagation to local device-to-device dissemination. This
article proposes a novel traffic-aware patching scheme to select important
intermediate nodes to patch, which applies to the IoT system with limited
patching resources and response time constraint. Experiments on real-world
trace datasets in IoT networks are conducted to demonstrate the advantage of
the proposed traffic-aware patching scheme in alleviating malware propagation.
| no_new_dataset | 0.945298 |
1703.05411 | Tien Thanh Nguyen | Tien Thanh Nguyen, Xuan Cuong Pham, Alan Wee-Chung Liew, Witold
Pedrycz | Aggregation of Classifiers: A Justifiable Information Granularity
Approach | 33 pages, 3 figures | null | null | null | cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this study, we introduce a new approach to combine multi-classifiers in an
ensemble system. Instead of using numeric membership values encountered in
fixed combining rules, we construct interval membership values associated with
each class prediction at the level of meta-data of observation by using
concepts of information granules. In the proposed method, uncertainty
(diversity) of findings produced by the base classifiers is quantified by
interval-based information granules. The discriminative decision model is
generated by considering both the bounds and the length of the obtained
intervals. We select ten and then fifteen learning algorithms to build a
heterogeneous ensemble system and then conducted the experiment on a number of
UCI datasets. The experimental results demonstrate that the proposed approach
performs better than the benchmark algorithms including six fixed combining
methods, one trainable combining method, AdaBoost, Bagging, and Random
Subspace.
| [
{
"version": "v1",
"created": "Wed, 15 Mar 2017 22:48:05 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Nguyen",
"Tien Thanh",
""
],
[
"Pham",
"Xuan Cuong",
""
],
[
"Liew",
"Alan Wee-Chung",
""
],
[
"Pedrycz",
"Witold",
""
]
] | TITLE: Aggregation of Classifiers: A Justifiable Information Granularity
Approach
ABSTRACT: In this study, we introduce a new approach to combine multi-classifiers in an
ensemble system. Instead of using numeric membership values encountered in
fixed combining rules, we construct interval membership values associated with
each class prediction at the level of meta-data of observation by using
concepts of information granules. In the proposed method, uncertainty
(diversity) of findings produced by the base classifiers is quantified by
interval-based information granules. The discriminative decision model is
generated by considering both the bounds and the length of the obtained
intervals. We select ten and then fifteen learning algorithms to build a
heterogeneous ensemble system and then conducted the experiment on a number of
UCI datasets. The experimental results demonstrate that the proposed approach
performs better than the benchmark algorithms including six fixed combining
methods, one trainable combining method, AdaBoost, Bagging, and Random
Subspace.
| no_new_dataset | 0.951459 |
1703.05422 | Travis Desell | Travis Desell | Large Scale Evolution of Convolutional Neural Networks Using Volunteer
Computing | 17 pages, 13 figures. Submitted to the 2017 Genetic and Evolutionary
Computation Conference (GECCO 2017) | null | null | null | cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work presents a new algorithm called evolutionary exploration of
augmenting convolutional topologies (EXACT), which is capable of evolving the
structure of convolutional neural networks (CNNs). EXACT is in part modeled
after the neuroevolution of augmenting topologies (NEAT) algorithm, with
notable exceptions to allow it to scale to large scale distributed computing
environments and evolve networks with convolutional filters. In addition to
multithreaded and MPI versions, EXACT has been implemented as part of a BOINC
volunteer computing project, allowing large scale evolution. During a period of
two months, over 4,500 volunteered computers on the Citizen Science Grid
trained over 120,000 CNNs and evolved networks reaching 98.32% test data
accuracy on the MNIST handwritten digits dataset. These results are even
stronger as the backpropagation strategy used to train the CNNs was fairly
rudimentary (ReLU units, L2 regularization and Nesterov momentum) and these
were initial test runs done without refinement of the backpropagation
hyperparameters. Further, the EXACT evolutionary strategy is independent of the
method used to train the CNNs, so they could be further improved by advanced
techniques like elastic distortions, pretraining and dropout. The evolved
networks are also quite interesting, showing "organic" structures and
significant differences from standard human designed architectures.
| [
{
"version": "v1",
"created": "Wed, 15 Mar 2017 23:17:24 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Desell",
"Travis",
""
]
] | TITLE: Large Scale Evolution of Convolutional Neural Networks Using Volunteer
Computing
ABSTRACT: This work presents a new algorithm called evolutionary exploration of
augmenting convolutional topologies (EXACT), which is capable of evolving the
structure of convolutional neural networks (CNNs). EXACT is in part modeled
after the neuroevolution of augmenting topologies (NEAT) algorithm, with
notable exceptions to allow it to scale to large scale distributed computing
environments and evolve networks with convolutional filters. In addition to
multithreaded and MPI versions, EXACT has been implemented as part of a BOINC
volunteer computing project, allowing large scale evolution. During a period of
two months, over 4,500 volunteered computers on the Citizen Science Grid
trained over 120,000 CNNs and evolved networks reaching 98.32% test data
accuracy on the MNIST handwritten digits dataset. These results are even
stronger as the backpropagation strategy used to train the CNNs was fairly
rudimentary (ReLU units, L2 regularization and Nesterov momentum) and these
were initial test runs done without refinement of the backpropagation
hyperparameters. Further, the EXACT evolutionary strategy is independent of the
method used to train the CNNs, so they could be further improved by advanced
techniques like elastic distortions, pretraining and dropout. The evolved
networks are also quite interesting, showing "organic" structures and
significant differences from standard human designed architectures.
| no_new_dataset | 0.949059 |
1703.05423 | Florian Strub | Florian Strub and Harm de Vries and Jeremie Mary and Bilal Piot and
Aaron Courville and Olivier Pietquin | End-to-end optimization of goal-driven and visually grounded dialogue
systems | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | End-to-end design of dialogue systems has recently become a popular research
topic thanks to powerful tools such as encoder-decoder architectures for
sequence-to-sequence learning. Yet, most current approaches cast human-machine
dialogue management as a supervised learning problem, aiming at predicting the
next utterance of a participant given the full history of the dialogue. This
vision is too simplistic to render the intrinsic planning problem inherent to
dialogue as well as its grounded nature, making the context of a dialogue
larger than the sole history. This is why only chit-chat and question answering
tasks have been addressed so far using end-to-end architectures. In this paper,
we introduce a Deep Reinforcement Learning method to optimize visually grounded
task-oriented dialogues, based on the policy gradient algorithm. This approach
is tested on a dataset of 120k dialogues collected through Mechanical Turk and
provides encouraging results at solving both the problem of generating natural
dialogues and the task of discovering a specific object in a complex picture.
| [
{
"version": "v1",
"created": "Wed, 15 Mar 2017 23:34:20 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Strub",
"Florian",
""
],
[
"de Vries",
"Harm",
""
],
[
"Mary",
"Jeremie",
""
],
[
"Piot",
"Bilal",
""
],
[
"Courville",
"Aaron",
""
],
[
"Pietquin",
"Olivier",
""
]
] | TITLE: End-to-end optimization of goal-driven and visually grounded dialogue
systems
ABSTRACT: End-to-end design of dialogue systems has recently become a popular research
topic thanks to powerful tools such as encoder-decoder architectures for
sequence-to-sequence learning. Yet, most current approaches cast human-machine
dialogue management as a supervised learning problem, aiming at predicting the
next utterance of a participant given the full history of the dialogue. This
vision is too simplistic to render the intrinsic planning problem inherent to
dialogue as well as its grounded nature, making the context of a dialogue
larger than the sole history. This is why only chit-chat and question answering
tasks have been addressed so far using end-to-end architectures. In this paper,
we introduce a Deep Reinforcement Learning method to optimize visually grounded
task-oriented dialogues, based on the policy gradient algorithm. This approach
is tested on a dataset of 120k dialogues collected through Mechanical Turk and
provides encouraging results at solving both the problem of generating natural
dialogues and the task of discovering a specific object in a complex picture.
| no_new_dataset | 0.937326 |
1703.05530 | Vincent Andrearczyk | Vincent Andrearczyk and Paul F. Whelan | Convolutional Neural Network on Three Orthogonal Planes for Dynamic
Texture Classification | 19 pages, 10 figures | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Dynamic Textures (DTs) are sequences of images of moving scenes that exhibit
certain stationarity properties in time such as smoke, vegetation and fire. The
analysis of DT is important for recognition, segmentation, synthesis or
retrieval for a range of applications including surveillance, medical imaging
and remote sensing. Deep learning methods have shown impressive results and are
now the new state of the art for a wide range of computer vision tasks
including image and video recognition and segmentation. In particular,
Convolutional Neural Networks (CNNs) have recently proven to be well suited for
texture analysis with a design similar to a filter bank approach. In this
paper, we develop a new approach to DT analysis based on a CNN method applied
on three orthogonal planes x y , xt and y t . We train CNNs on spatial frames
and temporal slices extracted from the DT sequences and combine their outputs
to obtain a competitive DT classifier. Our results on a wide range of commonly
used DT classification benchmark datasets prove the robustness of our approach.
Significant improvement of the state of the art is shown on the larger
datasets.
| [
{
"version": "v1",
"created": "Thu, 16 Mar 2017 09:30:07 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Andrearczyk",
"Vincent",
""
],
[
"Whelan",
"Paul F.",
""
]
] | TITLE: Convolutional Neural Network on Three Orthogonal Planes for Dynamic
Texture Classification
ABSTRACT: Dynamic Textures (DTs) are sequences of images of moving scenes that exhibit
certain stationarity properties in time such as smoke, vegetation and fire. The
analysis of DT is important for recognition, segmentation, synthesis or
retrieval for a range of applications including surveillance, medical imaging
and remote sensing. Deep learning methods have shown impressive results and are
now the new state of the art for a wide range of computer vision tasks
including image and video recognition and segmentation. In particular,
Convolutional Neural Networks (CNNs) have recently proven to be well suited for
texture analysis with a design similar to a filter bank approach. In this
paper, we develop a new approach to DT analysis based on a CNN method applied
on three orthogonal planes x y , xt and y t . We train CNNs on spatial frames
and temporal slices extracted from the DT sequences and combine their outputs
to obtain a competitive DT classifier. Our results on a wide range of commonly
used DT classification benchmark datasets prove the robustness of our approach.
Significant improvement of the state of the art is shown on the larger
datasets.
| no_new_dataset | 0.955236 |
1703.05584 | Mohammad Azzeh | Mohammad Azzeh | Software effort estimation based on optimized model tree | null | null | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Background: It is widely recognized that software effort estimation is a
regression problem. Model Tree (MT) is one of the Machine Learning based
regression techniques that is useful for software effort estimation, but as
other machine learning algorithms, the MT has a large space of configuration
and requires to carefully setting its parameters. The choice of such parameters
is a dataset dependent so no general guideline can govern this process which
forms the motivation of this work. Aims: This study investigates the effect of
using the most recent optimization algorithm called Bees algorithm to specify
the optimal choice of MT parameters that fit a dataset and therefore improve
prediction accuracy. Method: We used MT with optimal parameters identified by
the Bees algorithm to construct software effort estimation model. The model has
been validated over eight datasets come from two main sources: PROMISE and
ISBSG. Also we used 3-Fold cross validation to empirically assess the
prediction accuracies of different estimation models. As benchmark, results are
also compared to those obtained with Stepwise Regression Case-Based Reasoning
and Multi-Layer Perceptron. Results: The results obtained from combination of
MT and Bees algorithm are encouraging and outperforms other well-known
estimation methods applied on employed datasets. They are also interesting
enough to suggest the effectiveness of MT among the techniques that are
suitable for effort estimation. Conclusions: The use of the Bees algorithm
enabled us to automatically find optimal MT parameters required to construct
effort estimation models that fit each individual dataset. Also it provided a
significant improvement on prediction accuracy.
| [
{
"version": "v1",
"created": "Mon, 13 Mar 2017 18:23:55 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Azzeh",
"Mohammad",
""
]
] | TITLE: Software effort estimation based on optimized model tree
ABSTRACT: Background: It is widely recognized that software effort estimation is a
regression problem. Model Tree (MT) is one of the Machine Learning based
regression techniques that is useful for software effort estimation, but as
other machine learning algorithms, the MT has a large space of configuration
and requires to carefully setting its parameters. The choice of such parameters
is a dataset dependent so no general guideline can govern this process which
forms the motivation of this work. Aims: This study investigates the effect of
using the most recent optimization algorithm called Bees algorithm to specify
the optimal choice of MT parameters that fit a dataset and therefore improve
prediction accuracy. Method: We used MT with optimal parameters identified by
the Bees algorithm to construct software effort estimation model. The model has
been validated over eight datasets come from two main sources: PROMISE and
ISBSG. Also we used 3-Fold cross validation to empirically assess the
prediction accuracies of different estimation models. As benchmark, results are
also compared to those obtained with Stepwise Regression Case-Based Reasoning
and Multi-Layer Perceptron. Results: The results obtained from combination of
MT and Bees algorithm are encouraging and outperforms other well-known
estimation methods applied on employed datasets. They are also interesting
enough to suggest the effectiveness of MT among the techniques that are
suitable for effort estimation. Conclusions: The use of the Bees algorithm
enabled us to automatically find optimal MT parameters required to construct
effort estimation models that fit each individual dataset. Also it provided a
significant improvement on prediction accuracy.
| no_new_dataset | 0.9463 |
1703.05605 | Li Liu | Li Liu, Fumin Shen, Yuming Shen, Xianglong Liu, and Ling Shao | Deep Sketch Hashing: Fast Free-hand Sketch-Based Image Retrieval | This paper will appear as a spotlight paper in CVPR2017 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Free-hand sketch-based image retrieval (SBIR) is a specific cross-view
retrieval task, in which queries are abstract and ambiguous sketches while the
retrieval database is formed with natural images. Work in this area mainly
focuses on extracting representative and shared features for sketches and
natural images. However, these can neither cope well with the geometric
distortion between sketches and images nor be feasible for large-scale SBIR due
to the heavy continuous-valued distance computation. In this paper, we speed up
SBIR by introducing a novel binary coding method, named \textbf{Deep Sketch
Hashing} (DSH), where a semi-heterogeneous deep architecture is proposed and
incorporated into an end-to-end binary coding framework. Specifically, three
convolutional neural networks are utilized to encode free-hand sketches,
natural images and, especially, the auxiliary sketch-tokens which are adopted
as bridges to mitigate the sketch-image geometric distortion. The learned DSH
codes can effectively capture the cross-view similarities as well as the
intrinsic semantic correlations between different categories. To the best of
our knowledge, DSH is the first hashing work specifically designed for
category-level SBIR with an end-to-end deep architecture. The proposed DSH is
comprehensively evaluated on two large-scale datasets of TU-Berlin Extension
and Sketchy, and the experiments consistently show DSH's superior SBIR
accuracies over several state-of-the-art methods, while achieving significantly
reduced retrieval time and memory footprint.
| [
{
"version": "v1",
"created": "Thu, 16 Mar 2017 13:18:36 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Liu",
"Li",
""
],
[
"Shen",
"Fumin",
""
],
[
"Shen",
"Yuming",
""
],
[
"Liu",
"Xianglong",
""
],
[
"Shao",
"Ling",
""
]
] | TITLE: Deep Sketch Hashing: Fast Free-hand Sketch-Based Image Retrieval
ABSTRACT: Free-hand sketch-based image retrieval (SBIR) is a specific cross-view
retrieval task, in which queries are abstract and ambiguous sketches while the
retrieval database is formed with natural images. Work in this area mainly
focuses on extracting representative and shared features for sketches and
natural images. However, these can neither cope well with the geometric
distortion between sketches and images nor be feasible for large-scale SBIR due
to the heavy continuous-valued distance computation. In this paper, we speed up
SBIR by introducing a novel binary coding method, named \textbf{Deep Sketch
Hashing} (DSH), where a semi-heterogeneous deep architecture is proposed and
incorporated into an end-to-end binary coding framework. Specifically, three
convolutional neural networks are utilized to encode free-hand sketches,
natural images and, especially, the auxiliary sketch-tokens which are adopted
as bridges to mitigate the sketch-image geometric distortion. The learned DSH
codes can effectively capture the cross-view similarities as well as the
intrinsic semantic correlations between different categories. To the best of
our knowledge, DSH is the first hashing work specifically designed for
category-level SBIR with an end-to-end deep architecture. The proposed DSH is
comprehensively evaluated on two large-scale datasets of TU-Berlin Extension
and Sketchy, and the experiments consistently show DSH's superior SBIR
accuracies over several state-of-the-art methods, while achieving significantly
reduced retrieval time and memory footprint.
| no_new_dataset | 0.946941 |
1703.05724 | Sailesh Conjeti | Sailesh Conjeti, Magdalini Paschali, Amin Katouzian and Nassir Navab | Learning Robust Hash Codes for Multiple Instance Image Retrieval | 10 pages, 7 figures, under review at MICCAI 2017 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, for the first time, we introduce a multiple instance (MI) deep
hashing technique for learning discriminative hash codes with weak bag-level
supervision suited for large-scale retrieval. We learn such hash codes by
aggregating deeply learnt hierarchical representations across bag members
through a dedicated MI pool layer. For better trainability and retrieval
quality, we propose a two-pronged approach that includes robust optimization
and training with an auxiliary single instance hashing arm which is
down-regulated gradually. We pose retrieval for tumor assessment as an MI
problem because tumors often coexist with benign masses and could exhibit
complementary signatures when scanned from different anatomical views.
Experimental validations on benchmark mammography and histology datasets
demonstrate improved retrieval performance over the state-of-the-art methods.
| [
{
"version": "v1",
"created": "Thu, 16 Mar 2017 17:07:26 GMT"
}
] | 2017-03-17T00:00:00 | [
[
"Conjeti",
"Sailesh",
""
],
[
"Paschali",
"Magdalini",
""
],
[
"Katouzian",
"Amin",
""
],
[
"Navab",
"Nassir",
""
]
] | TITLE: Learning Robust Hash Codes for Multiple Instance Image Retrieval
ABSTRACT: In this paper, for the first time, we introduce a multiple instance (MI) deep
hashing technique for learning discriminative hash codes with weak bag-level
supervision suited for large-scale retrieval. We learn such hash codes by
aggregating deeply learnt hierarchical representations across bag members
through a dedicated MI pool layer. For better trainability and retrieval
quality, we propose a two-pronged approach that includes robust optimization
and training with an auxiliary single instance hashing arm which is
down-regulated gradually. We pose retrieval for tumor assessment as an MI
problem because tumors often coexist with benign masses and could exhibit
complementary signatures when scanned from different anatomical views.
Experimental validations on benchmark mammography and histology datasets
demonstrate improved retrieval performance over the state-of-the-art methods.
| no_new_dataset | 0.945197 |
1312.2923 | Adam Sykulski Dr | Adam M. Sykulski, Sofia C. Olhede, Jonathan M. Lilly, Eric Danioux | Lagrangian Time Series Models for Ocean Surface Drifter Trajectories | 21 pages, 10 figures | Journal of the Royal Statistical Society (Series C, Applied
Statistics), 65(1), 29-50, 2016 | 10.1111/rssc.12112 | null | stat.AP physics.ao-ph physics.flu-dyn stat.ME | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper proposes stochastic models for the analysis of ocean surface
trajectories obtained from freely-drifting satellite-tracked instruments. The
proposed time series models are used to summarise large multivariate datasets
and infer important physical parameters of inertial oscillations and other
ocean processes. Nonstationary time series methods are employed to account for
the spatiotemporal variability of each trajectory. Because the datasets are
large, we construct computationally efficient methods through the use of
frequency-domain modelling and estimation, with the data expressed as
complex-valued time series. We detail how practical issues related to sampling
and model misspecification may be addressed using semi-parametric techniques
for time series, and we demonstrate the effectiveness of our stochastic models
through application to both real-world data and to numerical model output.
| [
{
"version": "v1",
"created": "Tue, 10 Dec 2013 19:34:43 GMT"
},
{
"version": "v2",
"created": "Wed, 18 Mar 2015 22:44:56 GMT"
},
{
"version": "v3",
"created": "Wed, 22 Apr 2015 00:05:09 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Sykulski",
"Adam M.",
""
],
[
"Olhede",
"Sofia C.",
""
],
[
"Lilly",
"Jonathan M.",
""
],
[
"Danioux",
"Eric",
""
]
] | TITLE: Lagrangian Time Series Models for Ocean Surface Drifter Trajectories
ABSTRACT: This paper proposes stochastic models for the analysis of ocean surface
trajectories obtained from freely-drifting satellite-tracked instruments. The
proposed time series models are used to summarise large multivariate datasets
and infer important physical parameters of inertial oscillations and other
ocean processes. Nonstationary time series methods are employed to account for
the spatiotemporal variability of each trajectory. Because the datasets are
large, we construct computationally efficient methods through the use of
frequency-domain modelling and estimation, with the data expressed as
complex-valued time series. We detail how practical issues related to sampling
and model misspecification may be addressed using semi-parametric techniques
for time series, and we demonstrate the effectiveness of our stochastic models
through application to both real-world data and to numerical model output.
| no_new_dataset | 0.949059 |
1602.06662 | Mikael Henaff | Mikael Henaff, Arthur Szlam, Yann LeCun | Recurrent Orthogonal Networks and Long-Memory Tasks | null | null | null | null | cs.NE cs.AI cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Although RNNs have been shown to be powerful tools for processing sequential
data, finding architectures or optimization strategies that allow them to model
very long term dependencies is still an active area of research. In this work,
we carefully analyze two synthetic datasets originally outlined in (Hochreiter
and Schmidhuber, 1997) which are used to evaluate the ability of RNNs to store
information over many time steps. We explicitly construct RNN solutions to
these problems, and using these constructions, illuminate both the problems
themselves and the way in which RNNs store different types of information in
their hidden states. These constructions furthermore explain the success of
recent methods that specify unitary initializations or constraints on the
transition matrices.
| [
{
"version": "v1",
"created": "Mon, 22 Feb 2016 06:51:25 GMT"
},
{
"version": "v2",
"created": "Wed, 15 Mar 2017 17:45:08 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Henaff",
"Mikael",
""
],
[
"Szlam",
"Arthur",
""
],
[
"LeCun",
"Yann",
""
]
] | TITLE: Recurrent Orthogonal Networks and Long-Memory Tasks
ABSTRACT: Although RNNs have been shown to be powerful tools for processing sequential
data, finding architectures or optimization strategies that allow them to model
very long term dependencies is still an active area of research. In this work,
we carefully analyze two synthetic datasets originally outlined in (Hochreiter
and Schmidhuber, 1997) which are used to evaluate the ability of RNNs to store
information over many time steps. We explicitly construct RNN solutions to
these problems, and using these constructions, illuminate both the problems
themselves and the way in which RNNs store different types of information in
their hidden states. These constructions furthermore explain the success of
recent methods that specify unitary initializations or constraints on the
transition matrices.
| no_new_dataset | 0.94743 |
1610.02242 | Samuli Laine | Samuli Laine, Timo Aila | Temporal Ensembling for Semi-Supervised Learning | Final ICLR 2017 version. Includes new results for CIFAR-100 with
additional unlabeled data from Tiny Images dataset | null | null | null | cs.NE cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we present a simple and efficient method for training deep
neural networks in a semi-supervised setting where only a small portion of
training data is labeled. We introduce self-ensembling, where we form a
consensus prediction of the unknown labels using the outputs of the
network-in-training on different epochs, and most importantly, under different
regularization and input augmentation conditions. This ensemble prediction can
be expected to be a better predictor for the unknown labels than the output of
the network at the most recent training epoch, and can thus be used as a target
for training. Using our method, we set new records for two standard
semi-supervised learning benchmarks, reducing the (non-augmented)
classification error rate from 18.44% to 7.05% in SVHN with 500 labels and from
18.63% to 16.55% in CIFAR-10 with 4000 labels, and further to 5.12% and 12.16%
by enabling the standard augmentations. We additionally obtain a clear
improvement in CIFAR-100 classification accuracy by using random images from
the Tiny Images dataset as unlabeled extra inputs during training. Finally, we
demonstrate good tolerance to incorrect labels.
| [
{
"version": "v1",
"created": "Fri, 7 Oct 2016 12:15:42 GMT"
},
{
"version": "v2",
"created": "Mon, 7 Nov 2016 13:27:40 GMT"
},
{
"version": "v3",
"created": "Wed, 15 Mar 2017 14:22:41 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Laine",
"Samuli",
""
],
[
"Aila",
"Timo",
""
]
] | TITLE: Temporal Ensembling for Semi-Supervised Learning
ABSTRACT: In this paper, we present a simple and efficient method for training deep
neural networks in a semi-supervised setting where only a small portion of
training data is labeled. We introduce self-ensembling, where we form a
consensus prediction of the unknown labels using the outputs of the
network-in-training on different epochs, and most importantly, under different
regularization and input augmentation conditions. This ensemble prediction can
be expected to be a better predictor for the unknown labels than the output of
the network at the most recent training epoch, and can thus be used as a target
for training. Using our method, we set new records for two standard
semi-supervised learning benchmarks, reducing the (non-augmented)
classification error rate from 18.44% to 7.05% in SVHN with 500 labels and from
18.63% to 16.55% in CIFAR-10 with 4000 labels, and further to 5.12% and 12.16%
by enabling the standard augmentations. We additionally obtain a clear
improvement in CIFAR-100 classification accuracy by using random images from
the Tiny Images dataset as unlabeled extra inputs during training. Finally, we
demonstrate good tolerance to incorrect labels.
| no_new_dataset | 0.948155 |
1701.06454 | Domagoj Vrgo\v{c} | Jorge Baier, Dietrich Daroch, Juan Reutter, Domagoj Vrgo\v{c} | Evaluating navigational RDF queries over the Web | null | null | null | null | cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Semantic Web, and its underlying data format RDF, lend themselves naturally
to navigational querying due to their graph-like structure. This is
particularly evident when considering RDF data on the Web, where various
separately published datasets reference each other and form a giant graph known
as the Web of Linked Data. And while navigational queries over singular RDF
datasets are supported through SPARQL property paths, not much is known about
evaluating them over Linked Data. In this paper we propose a method for
evaluating property path queries over the Web based on the classical AI search
algorithm A*, show its optimality in the open world setting of the Web, and
test it using real world queries which access a variety of RDF datasets
available online.
| [
{
"version": "v1",
"created": "Mon, 23 Jan 2017 15:31:17 GMT"
},
{
"version": "v2",
"created": "Tue, 14 Mar 2017 20:24:41 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Baier",
"Jorge",
""
],
[
"Daroch",
"Dietrich",
""
],
[
"Reutter",
"Juan",
""
],
[
"Vrgoč",
"Domagoj",
""
]
] | TITLE: Evaluating navigational RDF queries over the Web
ABSTRACT: Semantic Web, and its underlying data format RDF, lend themselves naturally
to navigational querying due to their graph-like structure. This is
particularly evident when considering RDF data on the Web, where various
separately published datasets reference each other and form a giant graph known
as the Web of Linked Data. And while navigational queries over singular RDF
datasets are supported through SPARQL property paths, not much is known about
evaluating them over Linked Data. In this paper we propose a method for
evaluating property path queries over the Web based on the classical AI search
algorithm A*, show its optimality in the open world setting of the Web, and
test it using real world queries which access a variety of RDF datasets
available online.
| no_new_dataset | 0.948537 |
1702.06740 | Qiaolin Xia | Qiaolin Xia, Baobao Chang, Zhifang Sui | Improving Chinese SRL with Heterogeneous Annotations | This paper has been withdrawn by the author due to a crucial error in
equation 10 | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Previous studies on Chinese semantic role labeling (SRL) have concentrated on
single semantically annotated corpus. But the training data of single corpus is
often limited. Meanwhile, there usually exists other semantically annotated
corpora for Chinese SRL scattered across different annotation frameworks. Data
sparsity remains a bottleneck. This situation calls for larger training
datasets, or effective approaches which can take advantage of highly
heterogeneous data. In these papers, we focus mainly on the latter, that is, to
improve Chinese SRL by using heterogeneous corpora together. We propose a novel
progressive learning model which augments the Progressive Neural Network with
Gated Recurrent Adapters. The model can accommodate heterogeneous inputs and
effectively transfer knowledge between them. We also release a new corpus,
Chinese SemBank, for Chinese SRL. Experiments on CPB 1.0 show that ours model
outperforms state-of-the-art methods.
| [
{
"version": "v1",
"created": "Wed, 22 Feb 2017 10:34:47 GMT"
},
{
"version": "v2",
"created": "Mon, 13 Mar 2017 06:46:12 GMT"
},
{
"version": "v3",
"created": "Tue, 14 Mar 2017 13:05:23 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Xia",
"Qiaolin",
""
],
[
"Chang",
"Baobao",
""
],
[
"Sui",
"Zhifang",
""
]
] | TITLE: Improving Chinese SRL with Heterogeneous Annotations
ABSTRACT: Previous studies on Chinese semantic role labeling (SRL) have concentrated on
single semantically annotated corpus. But the training data of single corpus is
often limited. Meanwhile, there usually exists other semantically annotated
corpora for Chinese SRL scattered across different annotation frameworks. Data
sparsity remains a bottleneck. This situation calls for larger training
datasets, or effective approaches which can take advantage of highly
heterogeneous data. In these papers, we focus mainly on the latter, that is, to
improve Chinese SRL by using heterogeneous corpora together. We propose a novel
progressive learning model which augments the Progressive Neural Network with
Gated Recurrent Adapters. The model can accommodate heterogeneous inputs and
effectively transfer knowledge between them. We also release a new corpus,
Chinese SemBank, for Chinese SRL. Experiments on CPB 1.0 show that ours model
outperforms state-of-the-art methods.
| no_new_dataset | 0.528651 |
1702.07021 | Trang Pham | Trang Pham, Truyen Tran, Svetha Venkatesh | One Size Fits Many: Column Bundle for Multi-X Learning | null | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Much recent machine learning research has been directed towards leveraging
shared statistics among labels, instances and data views, commonly referred to
as multi-label, multi-instance and multi-view learning. The underlying premises
are that there exist correlations among input parts and among output targets,
and the predictive performance would increase when the correlations are
incorporated. In this paper, we propose Column Bundle (CLB), a novel deep
neural network for capturing the shared statistics in data. CLB is generic that
the same architecture can be applied for various types of shared statistics by
changing only input and output handling. CLB is capable of scaling to thousands
of input parts and output labels by avoiding explicit modeling of pairwise
relations. We evaluate CLB on different types of data: (a) multi-label, (b)
multi-view, (c) multi-view/multi-label and (d) multi-instance. CLB demonstrates
a comparable and competitive performance in all datasets against
state-of-the-art methods designed specifically for each type.
| [
{
"version": "v1",
"created": "Wed, 22 Feb 2017 21:54:12 GMT"
},
{
"version": "v2",
"created": "Tue, 14 Mar 2017 00:44:14 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Pham",
"Trang",
""
],
[
"Tran",
"Truyen",
""
],
[
"Venkatesh",
"Svetha",
""
]
] | TITLE: One Size Fits Many: Column Bundle for Multi-X Learning
ABSTRACT: Much recent machine learning research has been directed towards leveraging
shared statistics among labels, instances and data views, commonly referred to
as multi-label, multi-instance and multi-view learning. The underlying premises
are that there exist correlations among input parts and among output targets,
and the predictive performance would increase when the correlations are
incorporated. In this paper, we propose Column Bundle (CLB), a novel deep
neural network for capturing the shared statistics in data. CLB is generic that
the same architecture can be applied for various types of shared statistics by
changing only input and output handling. CLB is capable of scaling to thousands
of input parts and output labels by avoiding explicit modeling of pairwise
relations. We evaluate CLB on different types of data: (a) multi-label, (b)
multi-view, (c) multi-view/multi-label and (d) multi-instance. CLB demonstrates
a comparable and competitive performance in all datasets against
state-of-the-art methods designed specifically for each type.
| no_new_dataset | 0.94887 |
1702.07025 | Cristina Vasconcelos | Cristina Nader Vasconcelos, B\'arbara Nader Vasconcelos | Convolutional Neural Network Committees for Melanoma Classification with
Classical And Expert Knowledge Based Image Transforms Data Augmentation | null | null | null | null | cs.CV | http://creativecommons.org/publicdomain/zero/1.0/ | Skin cancer is a major public health problem, as is the most common type of
cancer and represents more than half of cancer diagnoses worldwide. Early
detection influences the outcome of the disease and motivates our work. We
investigate the composition of CNN committees and data augmentation for the the
ISBI 2017 Melanoma Classification Challenge (named Skin Lesion Analysis towards
Melanoma Detection) facing the peculiarities of dealing with such a small,
unbalanced, biological database. For that, we explore committees of
Convolutional Neural Networks trained over the ISBI challenge training dataset
artificially augmented by both classical image processing transforms and image
warping guided by specialist knowledge about the lesion axis and improve the
final classifier invariance to common melanoma variations.
| [
{
"version": "v1",
"created": "Wed, 22 Feb 2017 22:17:13 GMT"
},
{
"version": "v2",
"created": "Wed, 15 Mar 2017 11:50:58 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Vasconcelos",
"Cristina Nader",
""
],
[
"Vasconcelos",
"Bárbara Nader",
""
]
] | TITLE: Convolutional Neural Network Committees for Melanoma Classification with
Classical And Expert Knowledge Based Image Transforms Data Augmentation
ABSTRACT: Skin cancer is a major public health problem, as is the most common type of
cancer and represents more than half of cancer diagnoses worldwide. Early
detection influences the outcome of the disease and motivates our work. We
investigate the composition of CNN committees and data augmentation for the the
ISBI 2017 Melanoma Classification Challenge (named Skin Lesion Analysis towards
Melanoma Detection) facing the peculiarities of dealing with such a small,
unbalanced, biological database. For that, we explore committees of
Convolutional Neural Networks trained over the ISBI challenge training dataset
artificially augmented by both classical image processing transforms and image
warping guided by specialist knowledge about the lesion axis and improve the
final classifier invariance to common melanoma variations.
| no_new_dataset | 0.950457 |
1703.02180 | Yunpeng Chen | Chen Yunpeng, Jin Xiaojie, Kang Bingyi, Feng Jiashi, Yan Shuicheng | Sharing Residual Units Through Collective Tensor Factorization in Deep
Neural Networks | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Residual units are wildly used for alleviating optimization difficulties when
building deep neural networks. However, the performance gain does not well
compensate the model size increase, indicating low parameter efficiency in
these residual units. In this work, we first revisit the residual function in
several variations of residual units and demonstrate that these residual
functions can actually be explained with a unified framework based on
generalized block term decomposition. Then, based on the new explanation, we
propose a new architecture, Collective Residual Unit (CRU), which enhances the
parameter efficiency of deep neural networks through collective tensor
factorization. CRU enables knowledge sharing across different residual units
using shared factors. Experimental results show that our proposed CRU Network
demonstrates outstanding parameter efficiency, achieving comparable
classification performance to ResNet-200 with the model size of ResNet-50. By
building a deeper network using CRU, we can achieve state-of-the-art single
model classification accuracy on ImageNet-1k and Places365-Standard benchmark
datasets. (Code and trained models are available on GitHub)
| [
{
"version": "v1",
"created": "Tue, 7 Mar 2017 02:20:57 GMT"
},
{
"version": "v2",
"created": "Wed, 15 Mar 2017 15:00:26 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Yunpeng",
"Chen",
""
],
[
"Xiaojie",
"Jin",
""
],
[
"Bingyi",
"Kang",
""
],
[
"Jiashi",
"Feng",
""
],
[
"Shuicheng",
"Yan",
""
]
] | TITLE: Sharing Residual Units Through Collective Tensor Factorization in Deep
Neural Networks
ABSTRACT: Residual units are wildly used for alleviating optimization difficulties when
building deep neural networks. However, the performance gain does not well
compensate the model size increase, indicating low parameter efficiency in
these residual units. In this work, we first revisit the residual function in
several variations of residual units and demonstrate that these residual
functions can actually be explained with a unified framework based on
generalized block term decomposition. Then, based on the new explanation, we
propose a new architecture, Collective Residual Unit (CRU), which enhances the
parameter efficiency of deep neural networks through collective tensor
factorization. CRU enables knowledge sharing across different residual units
using shared factors. Experimental results show that our proposed CRU Network
demonstrates outstanding parameter efficiency, achieving comparable
classification performance to ResNet-200 with the model size of ResNet-50. By
building a deeper network using CRU, we can achieve state-of-the-art single
model classification accuracy on ImageNet-1k and Places365-Standard benchmark
datasets. (Code and trained models are available on GitHub)
| no_new_dataset | 0.947769 |
1703.03372 | Dhanesh Ramachandram | Dhanesh Ramachandram and Terrance DeVries | LesionSeg: Semantic segmentation of skin lesions using Deep
Convolutional Neural Network | null | null | null | null | cs.CV cs.AI cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a method for skin lesion segmentation for the ISIC 2017 Skin
Lesion Segmentation Challenge. Our approach is based on a Fully Convolutional
Network architecture which is trained end to end, from scratch, on a limited
dataset. Our semantic segmentation architecture utilizes several recent
innovations in particularly in the combined use of (i) use of atrous
convolutions to increase the effective field of view of the network's receptive
field without increasing the number of parameters, (ii) the use of
network-in-network $1\times1$ convolution layers to add capacity to the network
and (iii) state-of-art super-resolution upsampling of predictions using
subpixel CNN layers. We reported a mean IOU score of 0.642 on the validation
set provided by the organisers.
| [
{
"version": "v1",
"created": "Thu, 9 Mar 2017 17:52:28 GMT"
},
{
"version": "v2",
"created": "Fri, 10 Mar 2017 19:56:40 GMT"
},
{
"version": "v3",
"created": "Wed, 15 Mar 2017 01:37:18 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Ramachandram",
"Dhanesh",
""
],
[
"DeVries",
"Terrance",
""
]
] | TITLE: LesionSeg: Semantic segmentation of skin lesions using Deep
Convolutional Neural Network
ABSTRACT: We present a method for skin lesion segmentation for the ISIC 2017 Skin
Lesion Segmentation Challenge. Our approach is based on a Fully Convolutional
Network architecture which is trained end to end, from scratch, on a limited
dataset. Our semantic segmentation architecture utilizes several recent
innovations in particularly in the combined use of (i) use of atrous
convolutions to increase the effective field of view of the network's receptive
field without increasing the number of parameters, (ii) the use of
network-in-network $1\times1$ convolution layers to add capacity to the network
and (iii) state-of-art super-resolution upsampling of predictions using
subpixel CNN layers. We reported a mean IOU score of 0.642 on the validation
set provided by the organisers.
| no_new_dataset | 0.951278 |
1703.03937 | Xavier Alameda-Pineda | Xavier Alameda-Pineda and Andrea Pilzer and Dan Xu and Nicu Sebe and
Elisa Ricci | Viraliency: Pooling Local Virality | Accepted at IEEE CVPR 2017 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In our overly-connected world, the automatic recognition of virality - the
quality of an image or video to be rapidly and widely spread in social networks
- is of crucial importance, and has recently awaken the interest of the
computer vision community. Concurrently, recent progress in deep learning
architectures showed that global pooling strategies allow the extraction of
activation maps, which highlight the parts of the image most likely to contain
instances of a certain class. We extend this concept by introducing a pooling
layer that learns the size of the support area to be averaged: the learned
top-N average (LENA) pooling. We hypothesize that the latent concepts (feature
maps) describing virality may require such a rich pooling strategy. We assess
the effectiveness of the LENA layer by appending it on top of a convolutional
siamese architecture and evaluate its performance on the task of predicting and
localizing virality. We report experiments on two publicly available datasets
annotated for virality and show that our method outperforms state-of-the-art
approaches.
| [
{
"version": "v1",
"created": "Sat, 11 Mar 2017 10:01:11 GMT"
},
{
"version": "v2",
"created": "Wed, 15 Mar 2017 07:36:58 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Alameda-Pineda",
"Xavier",
""
],
[
"Pilzer",
"Andrea",
""
],
[
"Xu",
"Dan",
""
],
[
"Sebe",
"Nicu",
""
],
[
"Ricci",
"Elisa",
""
]
] | TITLE: Viraliency: Pooling Local Virality
ABSTRACT: In our overly-connected world, the automatic recognition of virality - the
quality of an image or video to be rapidly and widely spread in social networks
- is of crucial importance, and has recently awaken the interest of the
computer vision community. Concurrently, recent progress in deep learning
architectures showed that global pooling strategies allow the extraction of
activation maps, which highlight the parts of the image most likely to contain
instances of a certain class. We extend this concept by introducing a pooling
layer that learns the size of the support area to be averaged: the learned
top-N average (LENA) pooling. We hypothesize that the latent concepts (feature
maps) describing virality may require such a rich pooling strategy. We assess
the effectiveness of the LENA layer by appending it on top of a convolutional
siamese architecture and evaluate its performance on the task of predicting and
localizing virality. We report experiments on two publicly available datasets
annotated for virality and show that our method outperforms state-of-the-art
approaches.
| no_new_dataset | 0.946794 |
1703.04636 | Giovanni Poggi | Luca D'Amiano, Davide Cozzolino, Giovanni Poggi, Luisa Verdoliva | A PatchMatch-based Dense-field Algorithm for Video Copy-Move Detection
and Localization | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose a new algorithm for the reliable detection and localization of
video copy-move forgeries. Discovering well crafted video copy-moves may be
very difficult, especially when some uniform background is copied to occlude
foreground objects. To reliably detect both additive and occlusive copy-moves
we use a dense-field approach, with invariant features that guarantee
robustness to several post-processing operations. To limit complexity, a
suitable video-oriented version of PatchMatch is used, with a multiresolution
search strategy, and a focus on volumes of interest. Performance assessment
relies on a new dataset, designed ad hoc, with realistic copy-moves and a wide
variety of challenging situations. Experimental results show the proposed
method to detect and localize video copy-moves with good accuracy even in
adverse conditions.
| [
{
"version": "v1",
"created": "Tue, 14 Mar 2017 18:08:49 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"D'Amiano",
"Luca",
""
],
[
"Cozzolino",
"Davide",
""
],
[
"Poggi",
"Giovanni",
""
],
[
"Verdoliva",
"Luisa",
""
]
] | TITLE: A PatchMatch-based Dense-field Algorithm for Video Copy-Move Detection
and Localization
ABSTRACT: We propose a new algorithm for the reliable detection and localization of
video copy-move forgeries. Discovering well crafted video copy-moves may be
very difficult, especially when some uniform background is copied to occlude
foreground objects. To reliably detect both additive and occlusive copy-moves
we use a dense-field approach, with invariant features that guarantee
robustness to several post-processing operations. To limit complexity, a
suitable video-oriented version of PatchMatch is used, with a multiresolution
search strategy, and a focus on volumes of interest. Performance assessment
relies on a new dataset, designed ad hoc, with realistic copy-moves and a wide
variety of challenging situations. Experimental results show the proposed
method to detect and localize video copy-moves with good accuracy even in
adverse conditions.
| new_dataset | 0.953057 |
1703.04664 | Anshumali Shrivastava | Anshumali Shrivastava | Optimal Densification for Fast and Accurate Minwise Hashing | Fast Minwise Hashing | null | null | null | cs.DS cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Minwise hashing is a fundamental and one of the most successful hashing
algorithm in the literature. Recent advances based on the idea of
densification~\cite{Proc:OneHashLSH_ICML14,Proc:Shrivastava_UAI14} have shown
that it is possible to compute $k$ minwise hashes, of a vector with $d$
nonzeros, in mere $(d + k)$ computations, a significant improvement over the
classical $O(dk)$. These advances have led to an algorithmic improvement in the
query complexity of traditional indexing algorithms based on minwise hashing.
Unfortunately, the variance of the current densification techniques is
unnecessarily high, which leads to significantly poor accuracy compared to
vanilla minwise hashing, especially when the data is sparse. In this paper, we
provide a novel densification scheme which relies on carefully tailored
2-universal hashes. We show that the proposed scheme is variance-optimal, and
without losing the runtime efficiency, it is significantly more accurate than
existing densification techniques. As a result, we obtain a significantly
efficient hashing scheme which has the same variance and collision probability
as minwise hashing. Experimental evaluations on real sparse and
high-dimensional datasets validate our claims. We believe that given the
significant advantages, our method will replace minwise hashing implementations
in practice.
| [
{
"version": "v1",
"created": "Tue, 14 Mar 2017 18:49:57 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Shrivastava",
"Anshumali",
""
]
] | TITLE: Optimal Densification for Fast and Accurate Minwise Hashing
ABSTRACT: Minwise hashing is a fundamental and one of the most successful hashing
algorithm in the literature. Recent advances based on the idea of
densification~\cite{Proc:OneHashLSH_ICML14,Proc:Shrivastava_UAI14} have shown
that it is possible to compute $k$ minwise hashes, of a vector with $d$
nonzeros, in mere $(d + k)$ computations, a significant improvement over the
classical $O(dk)$. These advances have led to an algorithmic improvement in the
query complexity of traditional indexing algorithms based on minwise hashing.
Unfortunately, the variance of the current densification techniques is
unnecessarily high, which leads to significantly poor accuracy compared to
vanilla minwise hashing, especially when the data is sparse. In this paper, we
provide a novel densification scheme which relies on carefully tailored
2-universal hashes. We show that the proposed scheme is variance-optimal, and
without losing the runtime efficiency, it is significantly more accurate than
existing densification techniques. As a result, we obtain a significantly
efficient hashing scheme which has the same variance and collision probability
as minwise hashing. Experimental evaluations on real sparse and
high-dimensional datasets validate our claims. We believe that given the
significant advantages, our method will replace minwise hashing implementations
in practice.
| no_new_dataset | 0.9434 |
1703.04665 | Alexander Broad S | Alexander Broad and Brenna Argall | Geometry-Based Region Proposals for Real-Time Robot Detection of
Tabletop Objects | Update based on work presented at RSS 2016 Deep Learning Workshop | null | null | null | cs.RO cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a novel object detection pipeline for localization and recognition
in three dimensional environments. Our approach makes use of an RGB-D sensor
and combines state-of-the-art techniques from the robotics and computer vision
communities to create a robust, real-time detection system. We focus
specifically on solving the object detection problem for tabletop scenes, a
common environment for assistive manipulators. Our detection pipeline locates
objects in a point cloud representation of the scene. These clusters are
subsequently used to compute a bounding box around each object in the RGB
space. Each defined patch is then fed into a Convolutional Neural Network (CNN)
for object recognition. We also demonstrate that our region proposal method can
be used to develop novel datasets that are both large and diverse enough to
train deep learning models, and easy enough to collect that end-users can
develop their own datasets. Lastly, we validate the resulting system through an
extensive analysis of the accuracy and run-time of the full pipeline.
| [
{
"version": "v1",
"created": "Tue, 14 Mar 2017 18:51:18 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Broad",
"Alexander",
""
],
[
"Argall",
"Brenna",
""
]
] | TITLE: Geometry-Based Region Proposals for Real-Time Robot Detection of
Tabletop Objects
ABSTRACT: We present a novel object detection pipeline for localization and recognition
in three dimensional environments. Our approach makes use of an RGB-D sensor
and combines state-of-the-art techniques from the robotics and computer vision
communities to create a robust, real-time detection system. We focus
specifically on solving the object detection problem for tabletop scenes, a
common environment for assistive manipulators. Our detection pipeline locates
objects in a point cloud representation of the scene. These clusters are
subsequently used to compute a bounding box around each object in the RGB
space. Each defined patch is then fed into a Convolutional Neural Network (CNN)
for object recognition. We also demonstrate that our region proposal method can
be used to develop novel datasets that are both large and diverse enough to
train deep learning models, and easy enough to collect that end-users can
develop their own datasets. Lastly, we validate the resulting system through an
extensive analysis of the accuracy and run-time of the full pipeline.
| no_new_dataset | 0.776114 |
1703.04670 | Georgios Pavlakos | Georgios Pavlakos, Xiaowei Zhou, Aaron Chan, Konstantinos G. Derpanis,
Kostas Daniilidis | 6-DoF Object Pose from Semantic Keypoints | IEEE International Conference on Robotics and Automation (ICRA), 2017 | null | null | null | cs.CV cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents a novel approach to estimating the continuous six degree
of freedom (6-DoF) pose (3D translation and rotation) of an object from a
single RGB image. The approach combines semantic keypoints predicted by a
convolutional network (convnet) with a deformable shape model. Unlike prior
work, we are agnostic to whether the object is textured or textureless, as the
convnet learns the optimal representation from the available training image
data. Furthermore, the approach can be applied to instance- and class-based
pose recovery. Empirically, we show that the proposed approach can accurately
recover the 6-DoF object pose for both instance- and class-based scenarios with
a cluttered background. For class-based object pose estimation,
state-of-the-art accuracy is shown on the large-scale PASCAL3D+ dataset.
| [
{
"version": "v1",
"created": "Tue, 14 Mar 2017 18:58:46 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Pavlakos",
"Georgios",
""
],
[
"Zhou",
"Xiaowei",
""
],
[
"Chan",
"Aaron",
""
],
[
"Derpanis",
"Konstantinos G.",
""
],
[
"Daniilidis",
"Kostas",
""
]
] | TITLE: 6-DoF Object Pose from Semantic Keypoints
ABSTRACT: This paper presents a novel approach to estimating the continuous six degree
of freedom (6-DoF) pose (3D translation and rotation) of an object from a
single RGB image. The approach combines semantic keypoints predicted by a
convolutional network (convnet) with a deformable shape model. Unlike prior
work, we are agnostic to whether the object is textured or textureless, as the
convnet learns the optimal representation from the available training image
data. Furthermore, the approach can be applied to instance- and class-based
pose recovery. Empirically, we show that the proposed approach can accurately
recover the 6-DoF object pose for both instance- and class-based scenarios with
a cluttered background. For class-based object pose estimation,
state-of-the-art accuracy is shown on the large-scale PASCAL3D+ dataset.
| no_new_dataset | 0.947575 |
1703.04697 | Evgenii Chzhen | Evgenii Chzhen, Christophe Denis, Mohamed Hebiri, Joseph Salmon | On the benefits of output sparsity for multi-label classification | null | null | null | null | math.ST cs.LG stat.ML stat.TH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The multi-label classification framework, where each observation can be
associated with a set of labels, has generated a tremendous amount of attention
over recent years. The modern multi-label problems are typically large-scale in
terms of number of observations, features and labels, and the amount of labels
can even be comparable with the amount of observations. In this context,
different remedies have been proposed to overcome the curse of dimensionality.
In this work, we aim at exploiting the output sparsity by introducing a new
loss, called the sparse weighted Hamming loss. This proposed loss can be seen
as a weighted version of classical ones, where active and inactive labels are
weighted separately. Leveraging the influence of sparsity in the loss function,
we provide improved generalization bounds for the empirical risk minimizer, a
suitable property for large-scale problems. For this new loss, we derive rates
of convergence linear in the underlying output-sparsity rather than linear in
the number of labels. In practice, minimizing the associated risk can be
performed efficiently by using convex surrogates and modern convex optimization
algorithms. We provide experiments on various real-world datasets demonstrating
the pertinence of our approach when compared to non-weighted techniques.
| [
{
"version": "v1",
"created": "Tue, 14 Mar 2017 20:19:08 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Chzhen",
"Evgenii",
""
],
[
"Denis",
"Christophe",
""
],
[
"Hebiri",
"Mohamed",
""
],
[
"Salmon",
"Joseph",
""
]
] | TITLE: On the benefits of output sparsity for multi-label classification
ABSTRACT: The multi-label classification framework, where each observation can be
associated with a set of labels, has generated a tremendous amount of attention
over recent years. The modern multi-label problems are typically large-scale in
terms of number of observations, features and labels, and the amount of labels
can even be comparable with the amount of observations. In this context,
different remedies have been proposed to overcome the curse of dimensionality.
In this work, we aim at exploiting the output sparsity by introducing a new
loss, called the sparse weighted Hamming loss. This proposed loss can be seen
as a weighted version of classical ones, where active and inactive labels are
weighted separately. Leveraging the influence of sparsity in the loss function,
we provide improved generalization bounds for the empirical risk minimizer, a
suitable property for large-scale problems. For this new loss, we derive rates
of convergence linear in the underlying output-sparsity rather than linear in
the number of labels. In practice, minimizing the associated risk can be
performed efficiently by using convex surrogates and modern convex optimization
algorithms. We provide experiments on various real-world datasets demonstrating
the pertinence of our approach when compared to non-weighted techniques.
| no_new_dataset | 0.943608 |
1703.04819 | Sandra Avila | Afonso Menegola, Julia Tavares, Michel Fornaciali, Lin Tzy Li, Sandra
Avila, Eduardo Valle | RECOD Titans at ISIC Challenge 2017 | 5 pages | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This extended abstract describes the participation of RECOD Titans in parts 1
and 3 of the ISIC Challenge 2017 "Skin Lesion Analysis Towards Melanoma
Detection" (ISBI 2017). Although our team has a long experience with melanoma
classification, the ISIC Challenge 2017 was the very first time we worked on
skin-lesion segmentation. For part 1 (segmentation), our final submission used
four of our models: two trained with all 2000 samples, without a validation
split, for 250 and for 500 epochs respectively; and other two trained and
validated with two different 1600/400 splits, for 220 epochs. Those four
models, individually, achieved between 0.780 and 0.783 official validation
scores. Our final submission averaged the output of those four models achieved
a score of 0.793. For part 3 (classification), the submitted test run as well
as our last official validation run were the result from a meta-model that
assembled seven base deep-learning models: three based on Inception-V4 trained
on our largest dataset; three based on Inception trained on our smallest
dataset; and one based on ResNet-101 trained on our smaller dataset. The
results of those component models were stacked in a meta-learning layer based
on an SVM trained on the validation set of our largest dataset.
| [
{
"version": "v1",
"created": "Tue, 14 Mar 2017 23:11:04 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Menegola",
"Afonso",
""
],
[
"Tavares",
"Julia",
""
],
[
"Fornaciali",
"Michel",
""
],
[
"Li",
"Lin Tzy",
""
],
[
"Avila",
"Sandra",
""
],
[
"Valle",
"Eduardo",
""
]
] | TITLE: RECOD Titans at ISIC Challenge 2017
ABSTRACT: This extended abstract describes the participation of RECOD Titans in parts 1
and 3 of the ISIC Challenge 2017 "Skin Lesion Analysis Towards Melanoma
Detection" (ISBI 2017). Although our team has a long experience with melanoma
classification, the ISIC Challenge 2017 was the very first time we worked on
skin-lesion segmentation. For part 1 (segmentation), our final submission used
four of our models: two trained with all 2000 samples, without a validation
split, for 250 and for 500 epochs respectively; and other two trained and
validated with two different 1600/400 splits, for 220 epochs. Those four
models, individually, achieved between 0.780 and 0.783 official validation
scores. Our final submission averaged the output of those four models achieved
a score of 0.793. For part 3 (classification), the submitted test run as well
as our last official validation run were the result from a meta-model that
assembled seven base deep-learning models: three based on Inception-V4 trained
on our largest dataset; three based on Inception trained on our smallest
dataset; and one based on ResNet-101 trained on our smaller dataset. The
results of those component models were stacked in a meta-learning layer based
on an SVM trained on the validation set of our largest dataset.
| no_new_dataset | 0.942876 |
1703.04824 | Jan Haji\v{c} Jr | Jan Haji\v{c} jr., Pavel Pecina | In Search of a Dataset for Handwritten Optical Music Recognition:
Introducing MUSCIMA++ | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-sa/4.0/ | Optical Music Recognition (OMR) has long been without an adequate dataset and
ground truth for evaluating OMR systems, which has been a major problem for
establishing a state of the art in the field. Furthermore, machine learning
methods require training data. We analyze how the OMR processing pipeline can
be expressed in terms of gradually more complex ground truth, and based on this
analysis, we design the MUSCIMA++ dataset of handwritten music notation that
addresses musical symbol recognition and notation reconstruction. The MUSCIMA++
dataset version 0.9 consists of 140 pages of handwritten music, with 91255
manually annotated notation symbols and 82261 explicitly marked relationships
between symbol pairs. The dataset allows training and evaluating models for
symbol classification, symbol localization, and notation graph assembly, both
in isolation and jointly. Open-source tools are provided for manipulating the
dataset, visualizing the data and further annotation, and the dataset itself is
made available under an open license.
| [
{
"version": "v1",
"created": "Tue, 14 Mar 2017 23:21:26 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Hajič",
"Jan",
"jr."
],
[
"Pecina",
"Pavel",
""
]
] | TITLE: In Search of a Dataset for Handwritten Optical Music Recognition:
Introducing MUSCIMA++
ABSTRACT: Optical Music Recognition (OMR) has long been without an adequate dataset and
ground truth for evaluating OMR systems, which has been a major problem for
establishing a state of the art in the field. Furthermore, machine learning
methods require training data. We analyze how the OMR processing pipeline can
be expressed in terms of gradually more complex ground truth, and based on this
analysis, we design the MUSCIMA++ dataset of handwritten music notation that
addresses musical symbol recognition and notation reconstruction. The MUSCIMA++
dataset version 0.9 consists of 140 pages of handwritten music, with 91255
manually annotated notation symbols and 82261 explicitly marked relationships
between symbol pairs. The dataset allows training and evaluating models for
symbol classification, symbol localization, and notation graph assembly, both
in isolation and jointly. Open-source tools are provided for manipulating the
dataset, visualizing the data and further annotation, and the dataset itself is
made available under an open license.
| new_dataset | 0.965348 |
1703.04835 | Wei-An Lin | Wei-An Lin and Jun-Cheng Chen and Rama Chellappa | A Proximity-Aware Hierarchical Clustering of Faces | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we propose an unsupervised face clustering algorithm called
"Proximity-Aware Hierarchical Clustering" (PAHC) that exploits the local
structure of deep representations. In the proposed method, a similarity measure
between deep features is computed by evaluating linear SVM margins. SVMs are
trained using nearest neighbors of sample data, and thus do not require any
external training data. Clusters are then formed by thresholding the similarity
scores. We evaluate the clustering performance using three challenging
unconstrained face datasets, including Celebrity in Frontal-Profile (CFP),
IARPA JANUS Benchmark A (IJB-A), and JANUS Challenge Set 3 (JANUS CS3)
datasets. Experimental results demonstrate that the proposed approach can
achieve significant improvements over state-of-the-art methods. Moreover, we
also show that the proposed clustering algorithm can be applied to curate a set
of large-scale and noisy training dataset while maintaining sufficient amount
of images and their variations due to nuisance factors. The face verification
performance on JANUS CS3 improves significantly by finetuning a DCNN model with
the curated MS-Celeb-1M dataset which contains over three million face images.
| [
{
"version": "v1",
"created": "Tue, 14 Mar 2017 23:41:45 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Lin",
"Wei-An",
""
],
[
"Chen",
"Jun-Cheng",
""
],
[
"Chellappa",
"Rama",
""
]
] | TITLE: A Proximity-Aware Hierarchical Clustering of Faces
ABSTRACT: In this paper, we propose an unsupervised face clustering algorithm called
"Proximity-Aware Hierarchical Clustering" (PAHC) that exploits the local
structure of deep representations. In the proposed method, a similarity measure
between deep features is computed by evaluating linear SVM margins. SVMs are
trained using nearest neighbors of sample data, and thus do not require any
external training data. Clusters are then formed by thresholding the similarity
scores. We evaluate the clustering performance using three challenging
unconstrained face datasets, including Celebrity in Frontal-Profile (CFP),
IARPA JANUS Benchmark A (IJB-A), and JANUS Challenge Set 3 (JANUS CS3)
datasets. Experimental results demonstrate that the proposed approach can
achieve significant improvements over state-of-the-art methods. Moreover, we
also show that the proposed clustering algorithm can be applied to curate a set
of large-scale and noisy training dataset while maintaining sufficient amount
of images and their variations due to nuisance factors. The face verification
performance on JANUS CS3 improves significantly by finetuning a DCNN model with
the curated MS-Celeb-1M dataset which contains over three million face images.
| no_new_dataset | 0.92111 |
1703.04853 | Homa Foroughi | Homa Foroughi, Moein Shakeri, Nilanjan Ray, Hong Zhang | Face Recognition using Multi-Modal Low-Rank Dictionary Learning | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Face recognition has been widely studied due to its importance in different
applications; however, most of the proposed methods fail when face images are
occluded or captured under illumination and pose variations. Recently several
low-rank dictionary learning methods have been proposed and achieved promising
results for noisy observations. While these methods are mostly developed for
single-modality scenarios, recent studies demonstrated the advantages of
feature fusion from multiple inputs. We propose a multi-modal structured
low-rank dictionary learning method for robust face recognition, using raw
pixels of face images and their illumination invariant representation. The
proposed method learns robust and discriminative representations from
contaminated face images, even if there are few training samples with large
intra-class variations. Extensive experiments on different datasets validate
the superior performance and robustness of our method to severe illumination
variations and occlusion.
| [
{
"version": "v1",
"created": "Wed, 15 Mar 2017 00:38:01 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Foroughi",
"Homa",
""
],
[
"Shakeri",
"Moein",
""
],
[
"Ray",
"Nilanjan",
""
],
[
"Zhang",
"Hong",
""
]
] | TITLE: Face Recognition using Multi-Modal Low-Rank Dictionary Learning
ABSTRACT: Face recognition has been widely studied due to its importance in different
applications; however, most of the proposed methods fail when face images are
occluded or captured under illumination and pose variations. Recently several
low-rank dictionary learning methods have been proposed and achieved promising
results for noisy observations. While these methods are mostly developed for
single-modality scenarios, recent studies demonstrated the advantages of
feature fusion from multiple inputs. We propose a multi-modal structured
low-rank dictionary learning method for robust face recognition, using raw
pixels of face images and their illumination invariant representation. The
proposed method learns robust and discriminative representations from
contaminated face images, even if there are few training samples with large
intra-class variations. Extensive experiments on different datasets validate
the superior performance and robustness of our method to severe illumination
variations and occlusion.
| no_new_dataset | 0.947478 |
1703.04873 | Wei-Han Lee | Wei-Han Lee, Changchang Liu, Shouling Ji, Prateek Mittal, Ruby Lee | Quantification of De-anonymization Risks in Social Networks | Published in International Conference on Information Systems Security
and Privacy, 2017 | null | null | null | cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The risks of publishing privacy-sensitive data have received considerable
attention recently. Several de-anonymization attacks have been proposed to
re-identify individuals even if data anonymization techniques were applied.
However, there is no theoretical quantification for relating the data utility
that is preserved by the anonymization techniques and the data vulnerability
against de-anonymization attacks.
In this paper, we theoretically analyze the de-anonymization attacks and
provide conditions on the utility of the anonymized data (denoted by anonymized
utility) to achieve successful de-anonymization. To the best of our knowledge,
this is the first work on quantifying the relationships between anonymized
utility and de-anonymization capability. Unlike previous work, our
quantification analysis requires no assumptions about the graph model, thus
providing a general theoretical guide for developing practical
de-anonymization/anonymization techniques.
Furthermore, we evaluate state-of-the-art de-anonymization attacks on a
real-world Facebook dataset to show the limitations of previous work. By
comparing these experimental results and the theoretically achievable
de-anonymization capability derived in our analysis, we further demonstrate the
ineffectiveness of previous de-anonymization attacks and the potential of more
powerful de-anonymization attacks in the future.
| [
{
"version": "v1",
"created": "Wed, 15 Mar 2017 01:35:48 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Lee",
"Wei-Han",
""
],
[
"Liu",
"Changchang",
""
],
[
"Ji",
"Shouling",
""
],
[
"Mittal",
"Prateek",
""
],
[
"Lee",
"Ruby",
""
]
] | TITLE: Quantification of De-anonymization Risks in Social Networks
ABSTRACT: The risks of publishing privacy-sensitive data have received considerable
attention recently. Several de-anonymization attacks have been proposed to
re-identify individuals even if data anonymization techniques were applied.
However, there is no theoretical quantification for relating the data utility
that is preserved by the anonymization techniques and the data vulnerability
against de-anonymization attacks.
In this paper, we theoretically analyze the de-anonymization attacks and
provide conditions on the utility of the anonymized data (denoted by anonymized
utility) to achieve successful de-anonymization. To the best of our knowledge,
this is the first work on quantifying the relationships between anonymized
utility and de-anonymization capability. Unlike previous work, our
quantification analysis requires no assumptions about the graph model, thus
providing a general theoretical guide for developing practical
de-anonymization/anonymization techniques.
Furthermore, we evaluate state-of-the-art de-anonymization attacks on a
real-world Facebook dataset to show the limitations of previous work. By
comparing these experimental results and the theoretically achievable
de-anonymization capability derived in our analysis, we further demonstrate the
ineffectiveness of previous de-anonymization attacks and the potential of more
powerful de-anonymization attacks in the future.
| no_new_dataset | 0.946051 |
1703.04967 | Holger Roth | Mohammad Eshghi, Holger R. Roth, Masahiro Oda, Min Suk Chung, Kensaku
Mori | Comparison of the Deep-Learning-Based Automated Segmentation Methods for
the Head Sectioned Images of the Virtual Korean Human Project | Accepted for presentation at the 15th IAPR Conference on Machine
Vision Applications (MVA2017), Nagoya, Japan | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents an end-to-end pixelwise fully automated segmentation of
the head sectioned images of the Visible Korean Human (VKH) project based on
Deep Convolutional Neural Networks (DCNNs). By converting classification
networks into Fully Convolutional Networks (FCNs), a coarse prediction map,
with smaller size than the original input image, can be created for
segmentation purposes. To refine this map and to obtain a dense pixel-wise
output, standard FCNs use deconvolution layers to upsample the coarse map.
However, upsampling based on deconvolution increases the number of network
parameters and causes loss of detail because of interpolation. On the other
hand, dilated convolution is a new technique introduced recently that attempts
to capture multi-scale contextual information without increasing the network
parameters while keeping the resolution of the prediction maps high. We used
both a standard FCN and a dilated convolution based FCN for semantic
segmentation of the head sectioned images of the VKH dataset. Quantitative
results showed approximately 20% improvement in the segmentation accuracy when
using FCNs with dilated convolutions.
| [
{
"version": "v1",
"created": "Wed, 15 Mar 2017 06:49:01 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Eshghi",
"Mohammad",
""
],
[
"Roth",
"Holger R.",
""
],
[
"Oda",
"Masahiro",
""
],
[
"Chung",
"Min Suk",
""
],
[
"Mori",
"Kensaku",
""
]
] | TITLE: Comparison of the Deep-Learning-Based Automated Segmentation Methods for
the Head Sectioned Images of the Virtual Korean Human Project
ABSTRACT: This paper presents an end-to-end pixelwise fully automated segmentation of
the head sectioned images of the Visible Korean Human (VKH) project based on
Deep Convolutional Neural Networks (DCNNs). By converting classification
networks into Fully Convolutional Networks (FCNs), a coarse prediction map,
with smaller size than the original input image, can be created for
segmentation purposes. To refine this map and to obtain a dense pixel-wise
output, standard FCNs use deconvolution layers to upsample the coarse map.
However, upsampling based on deconvolution increases the number of network
parameters and causes loss of detail because of interpolation. On the other
hand, dilated convolution is a new technique introduced recently that attempts
to capture multi-scale contextual information without increasing the network
parameters while keeping the resolution of the prediction maps high. We used
both a standard FCN and a dilated convolution based FCN for semantic
segmentation of the head sectioned images of the VKH dataset. Quantitative
results showed approximately 20% improvement in the segmentation accuracy when
using FCNs with dilated convolutions.
| no_new_dataset | 0.94887 |
1703.04980 | Veronika Cheplygina | Veronika Cheplygina and Lauge S{\o}rensen and David M. J. Tax and
Jesper Holst Pedersen and Marco Loog and Marleen de Bruijne | Classification of COPD with Multiple Instance Learning | Published at International Conference on Pattern Recognition (ICPR)
2014 | null | 10.1109/ICPR.2014.268 | null | cs.CV stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Chronic obstructive pulmonary disease (COPD) is a lung disease where early
detection benefits the survival rate. COPD can be quantified by classifying
patches of computed tomography images, and combining patch labels into an
overall diagnosis for the image. As labeled patches are often not available,
image labels are propagated to the patches, incorrectly labeling healthy
patches in COPD patients as being affected by the disease. We approach
quantification of COPD from lung images as a multiple instance learning (MIL)
problem, which is more suitable for such weakly labeled data. We investigate
various MIL assumptions in the context of COPD and show that although a concept
region with COPD-related disease patterns is present, considering the whole
distribution of lung tissue patches improves the performance. The best method
is based on averaging instances and obtains an AUC of 0.742, which is higher
than the previously reported best of 0.713 on the same dataset. Using the full
training set further increases performance to 0.776, which is significantly
higher (DeLong test) than previous results.
| [
{
"version": "v1",
"created": "Wed, 15 Mar 2017 07:41:49 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Cheplygina",
"Veronika",
""
],
[
"Sørensen",
"Lauge",
""
],
[
"Tax",
"David M. J.",
""
],
[
"Pedersen",
"Jesper Holst",
""
],
[
"Loog",
"Marco",
""
],
[
"de Bruijne",
"Marleen",
""
]
] | TITLE: Classification of COPD with Multiple Instance Learning
ABSTRACT: Chronic obstructive pulmonary disease (COPD) is a lung disease where early
detection benefits the survival rate. COPD can be quantified by classifying
patches of computed tomography images, and combining patch labels into an
overall diagnosis for the image. As labeled patches are often not available,
image labels are propagated to the patches, incorrectly labeling healthy
patches in COPD patients as being affected by the disease. We approach
quantification of COPD from lung images as a multiple instance learning (MIL)
problem, which is more suitable for such weakly labeled data. We investigate
various MIL assumptions in the context of COPD and show that although a concept
region with COPD-related disease patterns is present, considering the whole
distribution of lung tissue patches improves the performance. The best method
is based on averaging instances and obtains an AUC of 0.742, which is higher
than the previously reported best of 0.713 on the same dataset. Using the full
training set further increases performance to 0.776, which is significantly
higher (DeLong test) than previous results.
| no_new_dataset | 0.95253 |
1703.04981 | Veronika Cheplygina | Veronika Cheplygina and Annegreet van Opbroek and M. Arfan Ikram and
Meike W. Vernooij and Marleen de Bruijne | Transfer Learning by Asymmetric Image Weighting for Segmentation across
Scanners | null | null | null | null | cs.CV stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Supervised learning has been very successful for automatic segmentation of
images from a single scanner. However, several papers report deteriorated
performances when using classifiers trained on images from one scanner to
segment images from other scanners. We propose a transfer learning classifier
that adapts to differences between training and test images. This method uses a
weighted ensemble of classifiers trained on individual images. The weight of
each classifier is determined by the similarity between its training image and
the test image.
We examine three unsupervised similarity measures, which can be used in
scenarios where no labeled data from a newly introduced scanner or scanning
protocol is available. The measures are based on a divergence, a bag distance,
and on estimating the labels with a clustering procedure. These measures are
asymmetric. We study whether the asymmetry can improve classification. Out of
the three similarity measures, the bag similarity measure is the most robust
across different studies and achieves excellent results on four brain tissue
segmentation datasets and three white matter lesion segmentation datasets,
acquired at different centers and with different scanners and scanning
protocols. We show that the asymmetry can indeed be informative, and that
computing the similarity from the test image to the training images is more
appropriate than the opposite direction.
| [
{
"version": "v1",
"created": "Wed, 15 Mar 2017 07:43:10 GMT"
}
] | 2017-03-16T00:00:00 | [
[
"Cheplygina",
"Veronika",
""
],
[
"van Opbroek",
"Annegreet",
""
],
[
"Ikram",
"M. Arfan",
""
],
[
"Vernooij",
"Meike W.",
""
],
[
"de Bruijne",
"Marleen",
""
]
] | TITLE: Transfer Learning by Asymmetric Image Weighting for Segmentation across
Scanners
ABSTRACT: Supervised learning has been very successful for automatic segmentation of
images from a single scanner. However, several papers report deteriorated
performances when using classifiers trained on images from one scanner to
segment images from other scanners. We propose a transfer learning classifier
that adapts to differences between training and test images. This method uses a
weighted ensemble of classifiers trained on individual images. The weight of
each classifier is determined by the similarity between its training image and
the test image.
We examine three unsupervised similarity measures, which can be used in
scenarios where no labeled data from a newly introduced scanner or scanning
protocol is available. The measures are based on a divergence, a bag distance,
and on estimating the labels with a clustering procedure. These measures are
asymmetric. We study whether the asymmetry can improve classification. Out of
the three similarity measures, the bag similarity measure is the most robust
across different studies and achieves excellent results on four brain tissue
segmentation datasets and three white matter lesion segmentation datasets,
acquired at different centers and with different scanners and scanning
protocols. We show that the asymmetry can indeed be informative, and that
computing the similarity from the test image to the training images is more
appropriate than the opposite direction.
| no_new_dataset | 0.949669 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.