Joint discriminative Bayesian dictionary and classifier learning

Naveed Akhtar, Ajmal Mian, Fatih Porikli

Research output: Chapter in Book/Conference paperConference paper

7 Citations (Scopus)

Abstract

We propose to jointly learn a Discriminative Bayesian dictionary along a linear classifier using coupled Beta-Bernoulli Processes. Our representation model uses separate base measures for the dictionary and the classifier, but associates them to the class-specific training data using the same Bernoulli distributions. The Bernoulli distributions control the frequency with which the factors (e.g. dictionary atoms) are used in data representations, and they are inferred while accounting for the class labels in our approach. To further encourage discrimination in the dictionary, our model uses separate (sets of) Bernoulli distributions to represent data from different classes. Our approach adaptively learns the association between the dictionary atoms and the class labels while tailoring the classifier to this relation with a joint inference over the dictionary and the classifier. Once a test sample is represented over the dictionary, its representation is accurately labeled by the classifier due to the strong coupling between the dictionary and the classifier. We derive the Gibbs Sampling equations for our joint representation model and test our approach for face, object, scene and action recognition to establish its effectiveness.

Original languageEnglish
Title of host publicationProceedings - 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017
Place of PublicationNew York
PublisherIEEE, Institute of Electrical and Electronics Engineers
Pages3919-3928
Number of pages10
Volume2017-January
ISBN (Electronic)9781538604571
DOIs
Publication statusPublished - 6 Nov 2017
Event30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017 - Honolulu, United States
Duration: 21 Jul 201726 Jul 2017

Conference

Conference30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017
CountryUnited States
CityHonolulu
Period21/07/1726/07/17

Fingerprint

Glossaries
Classifiers
Labels
Atoms
Sampling

Cite this

Akhtar, N., Mian, A., & Porikli, F. (2017). Joint discriminative Bayesian dictionary and classifier learning. In Proceedings - 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017 (Vol. 2017-January, pp. 3919-3928). New York: IEEE, Institute of Electrical and Electronics Engineers. https://doi.org/10.1109/CVPR.2017.417
Akhtar, Naveed ; Mian, Ajmal ; Porikli, Fatih. / Joint discriminative Bayesian dictionary and classifier learning. Proceedings - 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017. Vol. 2017-January New York : IEEE, Institute of Electrical and Electronics Engineers, 2017. pp. 3919-3928
@inproceedings{a16cb386cd3641939e8aba8b1d2a7916,
title = "Joint discriminative Bayesian dictionary and classifier learning",
abstract = "We propose to jointly learn a Discriminative Bayesian dictionary along a linear classifier using coupled Beta-Bernoulli Processes. Our representation model uses separate base measures for the dictionary and the classifier, but associates them to the class-specific training data using the same Bernoulli distributions. The Bernoulli distributions control the frequency with which the factors (e.g. dictionary atoms) are used in data representations, and they are inferred while accounting for the class labels in our approach. To further encourage discrimination in the dictionary, our model uses separate (sets of) Bernoulli distributions to represent data from different classes. Our approach adaptively learns the association between the dictionary atoms and the class labels while tailoring the classifier to this relation with a joint inference over the dictionary and the classifier. Once a test sample is represented over the dictionary, its representation is accurately labeled by the classifier due to the strong coupling between the dictionary and the classifier. We derive the Gibbs Sampling equations for our joint representation model and test our approach for face, object, scene and action recognition to establish its effectiveness.",
author = "Naveed Akhtar and Ajmal Mian and Fatih Porikli",
year = "2017",
month = "11",
day = "6",
doi = "10.1109/CVPR.2017.417",
language = "English",
volume = "2017-January",
pages = "3919--3928",
booktitle = "Proceedings - 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017",
publisher = "IEEE, Institute of Electrical and Electronics Engineers",
address = "United States",

}

Akhtar, N, Mian, A & Porikli, F 2017, Joint discriminative Bayesian dictionary and classifier learning. in Proceedings - 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017. vol. 2017-January, IEEE, Institute of Electrical and Electronics Engineers, New York, pp. 3919-3928, 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, Honolulu, United States, 21/07/17. https://doi.org/10.1109/CVPR.2017.417

Joint discriminative Bayesian dictionary and classifier learning. / Akhtar, Naveed; Mian, Ajmal; Porikli, Fatih.

Proceedings - 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017. Vol. 2017-January New York : IEEE, Institute of Electrical and Electronics Engineers, 2017. p. 3919-3928.

Research output: Chapter in Book/Conference paperConference paper

TY - GEN

T1 - Joint discriminative Bayesian dictionary and classifier learning

AU - Akhtar, Naveed

AU - Mian, Ajmal

AU - Porikli, Fatih

PY - 2017/11/6

Y1 - 2017/11/6

N2 - We propose to jointly learn a Discriminative Bayesian dictionary along a linear classifier using coupled Beta-Bernoulli Processes. Our representation model uses separate base measures for the dictionary and the classifier, but associates them to the class-specific training data using the same Bernoulli distributions. The Bernoulli distributions control the frequency with which the factors (e.g. dictionary atoms) are used in data representations, and they are inferred while accounting for the class labels in our approach. To further encourage discrimination in the dictionary, our model uses separate (sets of) Bernoulli distributions to represent data from different classes. Our approach adaptively learns the association between the dictionary atoms and the class labels while tailoring the classifier to this relation with a joint inference over the dictionary and the classifier. Once a test sample is represented over the dictionary, its representation is accurately labeled by the classifier due to the strong coupling between the dictionary and the classifier. We derive the Gibbs Sampling equations for our joint representation model and test our approach for face, object, scene and action recognition to establish its effectiveness.

AB - We propose to jointly learn a Discriminative Bayesian dictionary along a linear classifier using coupled Beta-Bernoulli Processes. Our representation model uses separate base measures for the dictionary and the classifier, but associates them to the class-specific training data using the same Bernoulli distributions. The Bernoulli distributions control the frequency with which the factors (e.g. dictionary atoms) are used in data representations, and they are inferred while accounting for the class labels in our approach. To further encourage discrimination in the dictionary, our model uses separate (sets of) Bernoulli distributions to represent data from different classes. Our approach adaptively learns the association between the dictionary atoms and the class labels while tailoring the classifier to this relation with a joint inference over the dictionary and the classifier. Once a test sample is represented over the dictionary, its representation is accurately labeled by the classifier due to the strong coupling between the dictionary and the classifier. We derive the Gibbs Sampling equations for our joint representation model and test our approach for face, object, scene and action recognition to establish its effectiveness.

UR - http://www.scopus.com/inward/record.url?scp=85044315971&partnerID=8YFLogxK

UR - http://cvpr2017.thecvf.com/

U2 - 10.1109/CVPR.2017.417

DO - 10.1109/CVPR.2017.417

M3 - Conference paper

VL - 2017-January

SP - 3919

EP - 3928

BT - Proceedings - 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017

PB - IEEE, Institute of Electrical and Electronics Engineers

CY - New York

ER -

Akhtar N, Mian A, Porikli F. Joint discriminative Bayesian dictionary and classifier learning. In Proceedings - 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017. Vol. 2017-January. New York: IEEE, Institute of Electrical and Electronics Engineers. 2017. p. 3919-3928 https://doi.org/10.1109/CVPR.2017.417