Sorry, you need to enable JavaScript to visit this website.

Learning theory and algorithms (MLR-LEAR)

ONLINE CONVOLUTIONAL DICTIONARY LEARNING


While a number of different algorithms have recently been proposed for convolutional dictionary learning, this remains an expensive problem. The single biggest impediment to learning from large training sets is the memory requirements, which grow at least linearly with the size of the training set since all existing methods are batch algorithms. The work reported here addresses this limitation by extending online dictionary learning ideas to the convolutional context.

Paper Details

Authors:
Jialin Liu, Cristina Garcia-Cardona, Brendt Wohlberg, Wotao Yin
Submitted On:
19 September 2017 - 9:30pm
Short Link:
Type:
Event:
Presenter's Name:
Paper Code:
Document Year:
Cite

Document Files

slides_online_cdl

(5 downloads)

Keywords

Subscribe

[1] Jialin Liu, Cristina Garcia-Cardona, Brendt Wohlberg, Wotao Yin, "ONLINE CONVOLUTIONAL DICTIONARY LEARNING", IEEE SigPort, 2017. [Online]. Available: http://sigport.org/2237. Accessed: Sep. 20, 2017.
@article{2237-17,
url = {http://sigport.org/2237},
author = {Jialin Liu; Cristina Garcia-Cardona; Brendt Wohlberg; Wotao Yin },
publisher = {IEEE SigPort},
title = {ONLINE CONVOLUTIONAL DICTIONARY LEARNING},
year = {2017} }
TY - EJOUR
T1 - ONLINE CONVOLUTIONAL DICTIONARY LEARNING
AU - Jialin Liu; Cristina Garcia-Cardona; Brendt Wohlberg; Wotao Yin
PY - 2017
PB - IEEE SigPort
UR - http://sigport.org/2237
ER -
Jialin Liu, Cristina Garcia-Cardona, Brendt Wohlberg, Wotao Yin. (2017). ONLINE CONVOLUTIONAL DICTIONARY LEARNING. IEEE SigPort. http://sigport.org/2237
Jialin Liu, Cristina Garcia-Cardona, Brendt Wohlberg, Wotao Yin, 2017. ONLINE CONVOLUTIONAL DICTIONARY LEARNING. Available at: http://sigport.org/2237.
Jialin Liu, Cristina Garcia-Cardona, Brendt Wohlberg, Wotao Yin. (2017). "ONLINE CONVOLUTIONAL DICTIONARY LEARNING." Web.
1. Jialin Liu, Cristina Garcia-Cardona, Brendt Wohlberg, Wotao Yin. ONLINE CONVOLUTIONAL DICTIONARY LEARNING [Internet]. IEEE SigPort; 2017. Available from : http://sigport.org/2237

COLOR REPRESENTATION IN DEEP NEURAL NETWORKS


Convolutional neural networks are top-performers on image
classification tasks. Understanding how they make use of
color information in images may be useful for various tasks.
In this paper we analyze the representation learned by a popular
CNN to detect and characterize color-related features.
We confirm the existence of some object- and color-specific
units, as well as the effect of layer-depth on color-sensitivity
and class-invariance.

Paper Details

Authors:
Edo Collins, Sabine Süsstrunk
Submitted On:
14 September 2017 - 2:22pm
Short Link:
Type:
Event:
Presenter's Name:
Paper Code:
Document Year:
Cite

Document Files

1227_poster_IVRL_Engilberge_ICIP_2017.pdf

(7 downloads)

Keywords

Subscribe

[1] Edo Collins, Sabine Süsstrunk, "COLOR REPRESENTATION IN DEEP NEURAL NETWORKS", IEEE SigPort, 2017. [Online]. Available: http://sigport.org/2042. Accessed: Sep. 20, 2017.
@article{2042-17,
url = {http://sigport.org/2042},
author = {Edo Collins; Sabine Süsstrunk },
publisher = {IEEE SigPort},
title = {COLOR REPRESENTATION IN DEEP NEURAL NETWORKS},
year = {2017} }
TY - EJOUR
T1 - COLOR REPRESENTATION IN DEEP NEURAL NETWORKS
AU - Edo Collins; Sabine Süsstrunk
PY - 2017
PB - IEEE SigPort
UR - http://sigport.org/2042
ER -
Edo Collins, Sabine Süsstrunk. (2017). COLOR REPRESENTATION IN DEEP NEURAL NETWORKS. IEEE SigPort. http://sigport.org/2042
Edo Collins, Sabine Süsstrunk, 2017. COLOR REPRESENTATION IN DEEP NEURAL NETWORKS. Available at: http://sigport.org/2042.
Edo Collins, Sabine Süsstrunk. (2017). "COLOR REPRESENTATION IN DEEP NEURAL NETWORKS." Web.
1. Edo Collins, Sabine Süsstrunk. COLOR REPRESENTATION IN DEEP NEURAL NETWORKS [Internet]. IEEE SigPort; 2017. Available from : http://sigport.org/2042

SENTIMENT ANALYSIS WITH RECURRENT NEURAL NETWORK AND UNSUPERVISED NEURAL LANGUAGE MODEL


This paper describes a simple and efficient Neural Language Model approach for text classification that relies only on unsupervised word representation inputs. Our model employs Recurrent Neural Network Long Short-Term Memory (RNN-LSTM), on top of pre-trained word vectors for sentence-level classification tasks. In our hypothesis we argue that using word vectors obtained from an unsupervised neural language model as an extra feature with RNN-LSTM for Natural Language Processing (NLP) system can increase the performance of the system.

Paper Details

Authors:
Submitted On:
8 March 2017 - 9:34am
Short Link:
Type:
Event:
Presenter's Name:
Paper Code:

Document Files

icassp2017_poster.pdf

(383 downloads)

Keywords

Subscribe

[1] , "SENTIMENT ANALYSIS WITH RECURRENT NEURAL NETWORK AND UNSUPERVISED NEURAL LANGUAGE MODEL", IEEE SigPort, 2017. [Online]. Available: http://sigport.org/1707. Accessed: Sep. 20, 2017.
@article{1707-17,
url = {http://sigport.org/1707},
author = { },
publisher = {IEEE SigPort},
title = {SENTIMENT ANALYSIS WITH RECURRENT NEURAL NETWORK AND UNSUPERVISED NEURAL LANGUAGE MODEL},
year = {2017} }
TY - EJOUR
T1 - SENTIMENT ANALYSIS WITH RECURRENT NEURAL NETWORK AND UNSUPERVISED NEURAL LANGUAGE MODEL
AU -
PY - 2017
PB - IEEE SigPort
UR - http://sigport.org/1707
ER -
. (2017). SENTIMENT ANALYSIS WITH RECURRENT NEURAL NETWORK AND UNSUPERVISED NEURAL LANGUAGE MODEL. IEEE SigPort. http://sigport.org/1707
, 2017. SENTIMENT ANALYSIS WITH RECURRENT NEURAL NETWORK AND UNSUPERVISED NEURAL LANGUAGE MODEL. Available at: http://sigport.org/1707.
. (2017). "SENTIMENT ANALYSIS WITH RECURRENT NEURAL NETWORK AND UNSUPERVISED NEURAL LANGUAGE MODEL." Web.
1. . SENTIMENT ANALYSIS WITH RECURRENT NEURAL NETWORK AND UNSUPERVISED NEURAL LANGUAGE MODEL [Internet]. IEEE SigPort; 2017. Available from : http://sigport.org/1707

Dynamic Probabilistic Linear Discriminant Analysis for Face Recognition in Videos


Component Analysis (CA) for computer vision and machine learning comprises of a set of statistical techniques that decompose visual data to appropriate latent components that are relevant to the task-at-hand, such as alignment, clustering, segmentation, classification etc. The past few years we have witnessed an explosion of research in component analysis, introducing both novel deterministic and probabilistic models (e.g., Probabilistic Principal Component Analysis (PPCA), Probabilistic Linear Discriminant Analysis (PLDA), Probabilistic Canonical Correlation Analysis (PCCA) etc.).

Paper Details

Authors:
Alessandro Fabris, Mihalis Nicolau, Irene Kotsia, Stefanos Zafeiriou
Submitted On:
5 March 2017 - 4:58am
Short Link:
Type:
Event:
Presenter's Name:
Paper Code:
Document Year:
Cite

Document Files

Dynamic PLDA for face recognition in videos

(64 downloads)

Keywords

Subscribe

[1] Alessandro Fabris, Mihalis Nicolau, Irene Kotsia, Stefanos Zafeiriou, "Dynamic Probabilistic Linear Discriminant Analysis for Face Recognition in Videos", IEEE SigPort, 2017. [Online]. Available: http://sigport.org/1631. Accessed: Sep. 20, 2017.
@article{1631-17,
url = {http://sigport.org/1631},
author = {Alessandro Fabris; Mihalis Nicolau; Irene Kotsia; Stefanos Zafeiriou },
publisher = {IEEE SigPort},
title = {Dynamic Probabilistic Linear Discriminant Analysis for Face Recognition in Videos},
year = {2017} }
TY - EJOUR
T1 - Dynamic Probabilistic Linear Discriminant Analysis for Face Recognition in Videos
AU - Alessandro Fabris; Mihalis Nicolau; Irene Kotsia; Stefanos Zafeiriou
PY - 2017
PB - IEEE SigPort
UR - http://sigport.org/1631
ER -
Alessandro Fabris, Mihalis Nicolau, Irene Kotsia, Stefanos Zafeiriou. (2017). Dynamic Probabilistic Linear Discriminant Analysis for Face Recognition in Videos. IEEE SigPort. http://sigport.org/1631
Alessandro Fabris, Mihalis Nicolau, Irene Kotsia, Stefanos Zafeiriou, 2017. Dynamic Probabilistic Linear Discriminant Analysis for Face Recognition in Videos. Available at: http://sigport.org/1631.
Alessandro Fabris, Mihalis Nicolau, Irene Kotsia, Stefanos Zafeiriou. (2017). "Dynamic Probabilistic Linear Discriminant Analysis for Face Recognition in Videos." Web.
1. Alessandro Fabris, Mihalis Nicolau, Irene Kotsia, Stefanos Zafeiriou. Dynamic Probabilistic Linear Discriminant Analysis for Face Recognition in Videos [Internet]. IEEE SigPort; 2017. Available from : http://sigport.org/1631

A Robust FISTA-Like Algorithm


The Fast Iterative Shrinkage-Thresholding Algorithm (FISTA) is regarded as the state-of-the-art among a number of proximal gradient-based methods used for addressing large-scale optimization problems with simple but non-differentiable objective functions. However, the efficiency of FISTA in a wide range of applications is hampered by a simple drawback in the line search scheme. The local estimate of the Lipschitz constant, the inverse of which gives the step size, can only increase while the algorithm is running.

Paper Details

Authors:
Submitted On:
6 March 2017 - 8:50am
Short Link:
Type:
Event:
Presenter's Name:
Paper Code:
Document Year:
Cite

Document Files

ICASSP2017posterFloreaVorobyov.pdf

(73 downloads)

Keywords

Subscribe

[1] , "A Robust FISTA-Like Algorithm", IEEE SigPort, 2017. [Online]. Available: http://sigport.org/1568. Accessed: Sep. 20, 2017.
@article{1568-17,
url = {http://sigport.org/1568},
author = { },
publisher = {IEEE SigPort},
title = {A Robust FISTA-Like Algorithm},
year = {2017} }
TY - EJOUR
T1 - A Robust FISTA-Like Algorithm
AU -
PY - 2017
PB - IEEE SigPort
UR - http://sigport.org/1568
ER -
. (2017). A Robust FISTA-Like Algorithm. IEEE SigPort. http://sigport.org/1568
, 2017. A Robust FISTA-Like Algorithm. Available at: http://sigport.org/1568.
. (2017). "A Robust FISTA-Like Algorithm." Web.
1. . A Robust FISTA-Like Algorithm [Internet]. IEEE SigPort; 2017. Available from : http://sigport.org/1568

LEARNING TIME VARYING GRAPHS


We consider the problem of inferring the hidden structure of high-dimensional
time-varying data. In particular, we aim at capturing
the dynamic relationships by representing data as valued nodes in a
sequence of graphs. Our approach is motivated by the observation
that imposing a meaningful graph topology can help solving the generally
ill-posed and challenging problem of structure inference. To
capture the temporal evolution in the sequence of graphs, we introduce
a new prior that asserts that the graph edges change smoothly

Paper Details

Authors:
Vassilis Kalofolias, Andreas Loukas, Dorina Thanou, Pascal Frossard
Submitted On:
1 March 2017 - 8:50am
Short Link:
Type:
Event:
Presenter's Name:
Paper Code:
Document Year:
Cite

Document Files

ICASSP_graphlearning_poster-A0.pdf

(73 downloads)

Keywords

Subscribe

[1] Vassilis Kalofolias, Andreas Loukas, Dorina Thanou, Pascal Frossard, " LEARNING TIME VARYING GRAPHS", IEEE SigPort, 2017. [Online]. Available: http://sigport.org/1552. Accessed: Sep. 20, 2017.
@article{1552-17,
url = {http://sigport.org/1552},
author = {Vassilis Kalofolias; Andreas Loukas; Dorina Thanou; Pascal Frossard },
publisher = {IEEE SigPort},
title = { LEARNING TIME VARYING GRAPHS},
year = {2017} }
TY - EJOUR
T1 - LEARNING TIME VARYING GRAPHS
AU - Vassilis Kalofolias; Andreas Loukas; Dorina Thanou; Pascal Frossard
PY - 2017
PB - IEEE SigPort
UR - http://sigport.org/1552
ER -
Vassilis Kalofolias, Andreas Loukas, Dorina Thanou, Pascal Frossard. (2017). LEARNING TIME VARYING GRAPHS. IEEE SigPort. http://sigport.org/1552
Vassilis Kalofolias, Andreas Loukas, Dorina Thanou, Pascal Frossard, 2017. LEARNING TIME VARYING GRAPHS. Available at: http://sigport.org/1552.
Vassilis Kalofolias, Andreas Loukas, Dorina Thanou, Pascal Frossard. (2017). " LEARNING TIME VARYING GRAPHS." Web.
1. Vassilis Kalofolias, Andreas Loukas, Dorina Thanou, Pascal Frossard. LEARNING TIME VARYING GRAPHS [Internet]. IEEE SigPort; 2017. Available from : http://sigport.org/1552

LEARNING ROTATION INVARIANCE IN DEEP HIERARCHIES USING CIRCULAR SYMMETRIC FILTERS


Deep hierarchical models for feature learning have emerged as an effective technique for object representation and classification in recent years. Though the features learnt using deep models have shown lot of promise towards achieving invariance to data transformations, this primarily comes at the expense of using much larger training data and model size. In the proposed work we devise a novel technique to incorporate rotation invariance, while training the deep model parameters.

main.pdf

PDF icon main.pdf (261 downloads)

Paper Details

Authors:
Dhruv Kohli, Biplab Ch Das, Viswanath Gopalakrishnan, Kiran Nanjunda Iyer
Submitted On:
28 February 2017 - 11:56pm
Short Link:
Type:
Event:
Presenter's Name:
Paper Code:
Document Year:
Cite

Document Files

main.pdf

(261 downloads)

Keywords

Subscribe

[1] Dhruv Kohli, Biplab Ch Das, Viswanath Gopalakrishnan, Kiran Nanjunda Iyer, "LEARNING ROTATION INVARIANCE IN DEEP HIERARCHIES USING CIRCULAR SYMMETRIC FILTERS", IEEE SigPort, 2017. [Online]. Available: http://sigport.org/1529. Accessed: Sep. 20, 2017.
@article{1529-17,
url = {http://sigport.org/1529},
author = {Dhruv Kohli; Biplab Ch Das; Viswanath Gopalakrishnan; Kiran Nanjunda Iyer },
publisher = {IEEE SigPort},
title = {LEARNING ROTATION INVARIANCE IN DEEP HIERARCHIES USING CIRCULAR SYMMETRIC FILTERS},
year = {2017} }
TY - EJOUR
T1 - LEARNING ROTATION INVARIANCE IN DEEP HIERARCHIES USING CIRCULAR SYMMETRIC FILTERS
AU - Dhruv Kohli; Biplab Ch Das; Viswanath Gopalakrishnan; Kiran Nanjunda Iyer
PY - 2017
PB - IEEE SigPort
UR - http://sigport.org/1529
ER -
Dhruv Kohli, Biplab Ch Das, Viswanath Gopalakrishnan, Kiran Nanjunda Iyer. (2017). LEARNING ROTATION INVARIANCE IN DEEP HIERARCHIES USING CIRCULAR SYMMETRIC FILTERS. IEEE SigPort. http://sigport.org/1529
Dhruv Kohli, Biplab Ch Das, Viswanath Gopalakrishnan, Kiran Nanjunda Iyer, 2017. LEARNING ROTATION INVARIANCE IN DEEP HIERARCHIES USING CIRCULAR SYMMETRIC FILTERS. Available at: http://sigport.org/1529.
Dhruv Kohli, Biplab Ch Das, Viswanath Gopalakrishnan, Kiran Nanjunda Iyer. (2017). "LEARNING ROTATION INVARIANCE IN DEEP HIERARCHIES USING CIRCULAR SYMMETRIC FILTERS." Web.
1. Dhruv Kohli, Biplab Ch Das, Viswanath Gopalakrishnan, Kiran Nanjunda Iyer. LEARNING ROTATION INVARIANCE IN DEEP HIERARCHIES USING CIRCULAR SYMMETRIC FILTERS [Internet]. IEEE SigPort; 2017. Available from : http://sigport.org/1529

Active Regression with Compressive Sensing based Outlier Mitigation for both Small and Large Outliers


In this paper, a new active learning scheme is proposed for linear
regression problems with the objective of resolving the insufficient
training data problem and the unreliable training data labeling prob-
lem. A pool-based active regression technique is applied to select the
optimal training data to label from the overall data pool. Then, com-
pressive sensing is exploited to remove labeling errors if the errors
are sparse and have large enough magnitudes, which are called large

Paper Details

Authors:
Jian Zheng, Xiaohua Li
Submitted On:
7 December 2016 - 3:00pm
Short Link:
Type:
Event:
Presenter's Name:
Paper Code:
Document Year:
Cite

Document Files

Large and small outlier mitigation in active regression problems based on compressive sensing

(68 downloads)

Keywords

Subscribe

[1] Jian Zheng, Xiaohua Li, "Active Regression with Compressive Sensing based Outlier Mitigation for both Small and Large Outliers", IEEE SigPort, 2016. [Online]. Available: http://sigport.org/1413. Accessed: Sep. 20, 2017.
@article{1413-16,
url = {http://sigport.org/1413},
author = {Jian Zheng; Xiaohua Li },
publisher = {IEEE SigPort},
title = {Active Regression with Compressive Sensing based Outlier Mitigation for both Small and Large Outliers},
year = {2016} }
TY - EJOUR
T1 - Active Regression with Compressive Sensing based Outlier Mitigation for both Small and Large Outliers
AU - Jian Zheng; Xiaohua Li
PY - 2016
PB - IEEE SigPort
UR - http://sigport.org/1413
ER -
Jian Zheng, Xiaohua Li. (2016). Active Regression with Compressive Sensing based Outlier Mitigation for both Small and Large Outliers. IEEE SigPort. http://sigport.org/1413
Jian Zheng, Xiaohua Li, 2016. Active Regression with Compressive Sensing based Outlier Mitigation for both Small and Large Outliers. Available at: http://sigport.org/1413.
Jian Zheng, Xiaohua Li. (2016). "Active Regression with Compressive Sensing based Outlier Mitigation for both Small and Large Outliers." Web.
1. Jian Zheng, Xiaohua Li. Active Regression with Compressive Sensing based Outlier Mitigation for both Small and Large Outliers [Internet]. IEEE SigPort; 2016. Available from : http://sigport.org/1413

Sketching for Large-Scale Learning of Mixture Models


Learning parameters from voluminous data can be prohibitive in terms of memory and computational requirements. We propose a "compressive learning" framework where we first sketch the data by computing random generalized moments of the underlying probability distribution, then estimate mixture model parameters from the sketch using an iterative algorithm analogous to greedy sparse signal recovery. We exemplify our framework with the sketched estimation of Gaussian Mixture Models (GMMs).

pres.pdf

PDF icon pres.pdf (167 downloads)

Paper Details

Authors:
Anthony Bourrier, Remi Gribonval, Patrick Perez
Submitted On:
21 March 2016 - 4:00am
Short Link:
Type:
Event:
Presenter's Name:
Paper Code:
Document Year:
Cite

Document Files

pres.pdf

(167 downloads)

Keywords

Subscribe

[1] Anthony Bourrier, Remi Gribonval, Patrick Perez, "Sketching for Large-Scale Learning of Mixture Models", IEEE SigPort, 2016. [Online]. Available: http://sigport.org/911. Accessed: Sep. 20, 2017.
@article{911-16,
url = {http://sigport.org/911},
author = {Anthony Bourrier; Remi Gribonval; Patrick Perez },
publisher = {IEEE SigPort},
title = {Sketching for Large-Scale Learning of Mixture Models},
year = {2016} }
TY - EJOUR
T1 - Sketching for Large-Scale Learning of Mixture Models
AU - Anthony Bourrier; Remi Gribonval; Patrick Perez
PY - 2016
PB - IEEE SigPort
UR - http://sigport.org/911
ER -
Anthony Bourrier, Remi Gribonval, Patrick Perez. (2016). Sketching for Large-Scale Learning of Mixture Models. IEEE SigPort. http://sigport.org/911
Anthony Bourrier, Remi Gribonval, Patrick Perez, 2016. Sketching for Large-Scale Learning of Mixture Models. Available at: http://sigport.org/911.
Anthony Bourrier, Remi Gribonval, Patrick Perez. (2016). "Sketching for Large-Scale Learning of Mixture Models." Web.
1. Anthony Bourrier, Remi Gribonval, Patrick Perez. Sketching for Large-Scale Learning of Mixture Models [Internet]. IEEE SigPort; 2016. Available from : http://sigport.org/911

Dictionary Learning from Phaseless Measurements


We propose a new algorithm to learn a dictionary along with sparse representations from signal measurements without phase. Specifically, we consider the task of reconstructing a two-dimensional image from squared-magnitude measurements of a complex-valued linear transformation of the original image. Several recent phase retrieval algorithms exploit underlying sparsity of the unknown signal in order to improve recovery performance.

Paper Details

Authors:
Andreas M. Tillmann, Yonina C. Eldar, Julian Mairal
Submitted On:
16 March 2016 - 5:30am
Short Link:
Type:
Event:
Presenter's Name:
Document Year:
Cite

Document Files

DOLPHIn_ICASSP_Poster.pdf

(240 downloads)

Keywords

Subscribe

[1] Andreas M. Tillmann, Yonina C. Eldar, Julian Mairal, "Dictionary Learning from Phaseless Measurements", IEEE SigPort, 2016. [Online]. Available: http://sigport.org/710. Accessed: Sep. 20, 2017.
@article{710-16,
url = {http://sigport.org/710},
author = {Andreas M. Tillmann; Yonina C. Eldar; Julian Mairal },
publisher = {IEEE SigPort},
title = {Dictionary Learning from Phaseless Measurements},
year = {2016} }
TY - EJOUR
T1 - Dictionary Learning from Phaseless Measurements
AU - Andreas M. Tillmann; Yonina C. Eldar; Julian Mairal
PY - 2016
PB - IEEE SigPort
UR - http://sigport.org/710
ER -
Andreas M. Tillmann, Yonina C. Eldar, Julian Mairal. (2016). Dictionary Learning from Phaseless Measurements. IEEE SigPort. http://sigport.org/710
Andreas M. Tillmann, Yonina C. Eldar, Julian Mairal, 2016. Dictionary Learning from Phaseless Measurements. Available at: http://sigport.org/710.
Andreas M. Tillmann, Yonina C. Eldar, Julian Mairal. (2016). "Dictionary Learning from Phaseless Measurements." Web.
1. Andreas M. Tillmann, Yonina C. Eldar, Julian Mairal. Dictionary Learning from Phaseless Measurements [Internet]. IEEE SigPort; 2016. Available from : http://sigport.org/710

Pages