Boltzmann machines for image denoising

Research output: Chapter in Book/Report/Conference proceedingConference 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 languageEnglish (US)
Title of host publicationArtificial Neural Networks and Machine Learning, ICANN 2013 - 23rd International Conference on Artificial Neural Networks, Proceedings
Pages611-618
Number of pages8
Volume8131 LNCS
DOIs
StatePublished - 2013
Event23rd International Conference on Artificial Neural Networks, ICANN 2013 - Sofia, Bulgaria
Duration: Sep 10 2013Sep 13 2013

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8131 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other23rd International Conference on Artificial Neural Networks, ICANN 2013
CountryBulgaria
CitySofia
Period9/10/139/13/13

Fingerprint

Boltzmann Machine
Image denoising
Image Denoising
Denoising
Probabilistic Model
Patch
Model
Evaluate
Experiment
Experiments

Keywords

  • Deep Boltzmann Machine
  • Deep Learning
  • Image Denoising
  • Restricted Boltzmann Machine

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

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

}

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

SN - 9783642407277

VL - 8131 LNCS

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

ER -