Boltzmann machines for image denoising

Kyunghyun Cho

doi:10.1007/978-3-642-40728-4_76

Boltzmann machines for image denoising

Kyunghyun Cho

Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

Image denoising based on a probabilistic model of local image patches has been employed by various researchers, and recently a deep denoising autoencoder has been proposed in [2] and [17] as a good model for this. In this paper, we propose that another popular family of models in the field of deep learning, called Boltzmann machines, can perform image denoising as well as, or in certain cases of high level of noise, better than denoising autoencoders. We empirically evaluate these two models on three different sets of images with different types and levels of noise. The experiments confirmed our claim and revealed that the denoising performance can be improved by adding more hidden layers, especially when the level of noise is high.

Original language	English (US)
Title of host publication	Artificial Neural Networks and Machine Learning, ICANN 2013 - 23rd International Conference on Artificial Neural Networks, Proceedings
Pages	611-618
Number of pages	8
DOIs	https://doi.org/10.1007/978-3-642-40728-4_76
State	Published - 2013
Event	23rd International Conference on Artificial Neural Networks, ICANN 2013 - Sofia, Bulgaria Duration: Sep 10 2013 → Sep 13 2013

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	8131 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Other

Other	23rd International Conference on Artificial Neural Networks, ICANN 2013
Country/Territory	Bulgaria
City	Sofia
Period	9/10/13 → 9/13/13

Keywords

Deep Boltzmann Machine
Deep Learning
Image Denoising
Restricted Boltzmann Machine

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-642-40728-4_76

Cite this

Cho, K. (2013). Boltzmann machines for image denoising. In Artificial Neural Networks and Machine Learning, ICANN 2013 - 23rd International Conference on Artificial Neural Networks, Proceedings (pp. 611-618). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8131 LNCS). https://doi.org/10.1007/978-3-642-40728-4_76

Boltzmann machines for image denoising. / Cho, Kyunghyun.
Artificial Neural Networks and Machine Learning, ICANN 2013 - 23rd International Conference on Artificial Neural Networks, Proceedings. 2013. p. 611-618 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8131 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Cho, K 2013, Boltzmann machines for image denoising. in Artificial Neural Networks and Machine Learning, ICANN 2013 - 23rd International Conference on Artificial Neural Networks, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 8131 LNCS, pp. 611-618, 23rd International Conference on Artificial Neural Networks, ICANN 2013, Sofia, Bulgaria, 9/10/13. https://doi.org/10.1007/978-3-642-40728-4_76

Cho K. Boltzmann machines for image denoising. In Artificial Neural Networks and Machine Learning, ICANN 2013 - 23rd International Conference on Artificial Neural Networks, Proceedings. 2013. p. 611-618. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-642-40728-4_76

@inproceedings{93c6f343b12f45fd994031e57da48cc9,

title = "Boltzmann machines for image denoising",

abstract = "Image denoising based on a probabilistic model of local image patches has been employed by various researchers, and recently a deep denoising autoencoder has been proposed in [2] and [17] as a good model for this. In this paper, we propose that another popular family of models in the field of deep learning, called Boltzmann machines, can perform image denoising as well as, or in certain cases of high level of noise, better than denoising autoencoders. We empirically evaluate these two models on three different sets of images with different types and levels of noise. The experiments confirmed our claim and revealed that the denoising performance can be improved by adding more hidden layers, especially when the level of noise is high.",

keywords = "Deep Boltzmann Machine, Deep Learning, Image Denoising, Restricted Boltzmann Machine",

author = "Kyunghyun Cho",

year = "2013",

doi = "10.1007/978-3-642-40728-4_76",

language = "English (US)",

isbn = "9783642407277",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

pages = "611--618",

booktitle = "Artificial Neural Networks and Machine Learning, ICANN 2013 - 23rd International Conference on Artificial Neural Networks, Proceedings",

note = "23rd International Conference on Artificial Neural Networks, ICANN 2013 ; Conference date: 10-09-2013 Through 13-09-2013",

}

TY - GEN

T1 - Boltzmann machines for image denoising

AU - Cho, Kyunghyun

PY - 2013

Y1 - 2013

N2 - Image denoising based on a probabilistic model of local image patches has been employed by various researchers, and recently a deep denoising autoencoder has been proposed in [2] and [17] as a good model for this. In this paper, we propose that another popular family of models in the field of deep learning, called Boltzmann machines, can perform image denoising as well as, or in certain cases of high level of noise, better than denoising autoencoders. We empirically evaluate these two models on three different sets of images with different types and levels of noise. The experiments confirmed our claim and revealed that the denoising performance can be improved by adding more hidden layers, especially when the level of noise is high.

AB - Image denoising based on a probabilistic model of local image patches has been employed by various researchers, and recently a deep denoising autoencoder has been proposed in [2] and [17] as a good model for this. In this paper, we propose that another popular family of models in the field of deep learning, called Boltzmann machines, can perform image denoising as well as, or in certain cases of high level of noise, better than denoising autoencoders. We empirically evaluate these two models on three different sets of images with different types and levels of noise. The experiments confirmed our claim and revealed that the denoising performance can be improved by adding more hidden layers, especially when the level of noise is high.

KW - Deep Boltzmann Machine

KW - Deep Learning

KW - Image Denoising

KW - Restricted Boltzmann Machine

UR - http://www.scopus.com/inward/record.url?scp=84884963682&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84884963682&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-40728-4_76

DO - 10.1007/978-3-642-40728-4_76

M3 - Conference contribution

AN - SCOPUS:84884963682

SN - 9783642407277

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 611

EP - 618

BT - Artificial Neural Networks and Machine Learning, ICANN 2013 - 23rd International Conference on Artificial Neural Networks, Proceedings

T2 - 23rd International Conference on Artificial Neural Networks, ICANN 2013

Y2 - 10 September 2013 through 13 September 2013

ER -

Boltzmann machines for image denoising

Abstract

Publication series

Other

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this