Simple sparsification improves sparse denoising autoencoders in denoising highly noisy images

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Recently Burger et al. (2012) and Xie et al. (2012) proposed to use a denoising autoencoder (DAE) for denoising noisy images. They showed that a plain, deep DAE can denoise noisy images as well as the conventional methods such as BM3D and KSVD. Both of them approached image denoising by denoising small, image patches of a larger image and combining them to form a clean image. In this setting, it is usual to use the encoder of the DAE to obtain the latent representation and subsequently apply the decoder to get the clean patch. We propose that a simple sparsification of the latent representation found by the encoder improves denoising performance, both when the DAE was trained with and without sparsity regularization. The experiments confirm that the proposed sparsification indeed helps both denoising a small image patch and denoising a larger image consisting of those patches. Furthermore, it is found out that the proposed method improves even classification performance when test samples are corrupted with noise.

Original languageEnglish (US)
Title of host publication30th International Conference on Machine Learning, ICML 2013
PublisherInternational Machine Learning Society (IMLS)
Pages1469-1477
Number of pages9
EditionPART 2
StatePublished - 2013
Event30th International Conference on Machine Learning, ICML 2013 - Atlanta, GA, United States
Duration: Jun 16 2013Jun 21 2013

Other

Other30th International Conference on Machine Learning, ICML 2013
CountryUnited States
CityAtlanta, GA
Period6/16/136/21/13

Fingerprint

Image denoising
Experiments
performance
experiment

ASJC Scopus subject areas

  • Human-Computer Interaction
  • Sociology and Political Science

Cite this

Cho, K. (2013). Simple sparsification improves sparse denoising autoencoders in denoising highly noisy images. In 30th International Conference on Machine Learning, ICML 2013 (PART 2 ed., pp. 1469-1477). International Machine Learning Society (IMLS).

Simple sparsification improves sparse denoising autoencoders in denoising highly noisy images. / Cho, Kyunghyun.

30th International Conference on Machine Learning, ICML 2013. PART 2. ed. International Machine Learning Society (IMLS), 2013. p. 1469-1477.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Cho, K 2013, Simple sparsification improves sparse denoising autoencoders in denoising highly noisy images. in 30th International Conference on Machine Learning, ICML 2013. PART 2 edn, International Machine Learning Society (IMLS), pp. 1469-1477, 30th International Conference on Machine Learning, ICML 2013, Atlanta, GA, United States, 6/16/13.
Cho K. Simple sparsification improves sparse denoising autoencoders in denoising highly noisy images. In 30th International Conference on Machine Learning, ICML 2013. PART 2 ed. International Machine Learning Society (IMLS). 2013. p. 1469-1477
Cho, Kyunghyun. / Simple sparsification improves sparse denoising autoencoders in denoising highly noisy images. 30th International Conference on Machine Learning, ICML 2013. PART 2. ed. International Machine Learning Society (IMLS), 2013. pp. 1469-1477
@inproceedings{bd47565410ee4222abd3e1ab36efd2fb,
title = "Simple sparsification improves sparse denoising autoencoders in denoising highly noisy images",
abstract = "Recently Burger et al. (2012) and Xie et al. (2012) proposed to use a denoising autoencoder (DAE) for denoising noisy images. They showed that a plain, deep DAE can denoise noisy images as well as the conventional methods such as BM3D and KSVD. Both of them approached image denoising by denoising small, image patches of a larger image and combining them to form a clean image. In this setting, it is usual to use the encoder of the DAE to obtain the latent representation and subsequently apply the decoder to get the clean patch. We propose that a simple sparsification of the latent representation found by the encoder improves denoising performance, both when the DAE was trained with and without sparsity regularization. The experiments confirm that the proposed sparsification indeed helps both denoising a small image patch and denoising a larger image consisting of those patches. Furthermore, it is found out that the proposed method improves even classification performance when test samples are corrupted with noise.",
author = "Kyunghyun Cho",
year = "2013",
language = "English (US)",
pages = "1469--1477",
booktitle = "30th International Conference on Machine Learning, ICML 2013",
publisher = "International Machine Learning Society (IMLS)",
edition = "PART 2",

}

TY - GEN

T1 - Simple sparsification improves sparse denoising autoencoders in denoising highly noisy images

AU - Cho, Kyunghyun

PY - 2013

Y1 - 2013

N2 - Recently Burger et al. (2012) and Xie et al. (2012) proposed to use a denoising autoencoder (DAE) for denoising noisy images. They showed that a plain, deep DAE can denoise noisy images as well as the conventional methods such as BM3D and KSVD. Both of them approached image denoising by denoising small, image patches of a larger image and combining them to form a clean image. In this setting, it is usual to use the encoder of the DAE to obtain the latent representation and subsequently apply the decoder to get the clean patch. We propose that a simple sparsification of the latent representation found by the encoder improves denoising performance, both when the DAE was trained with and without sparsity regularization. The experiments confirm that the proposed sparsification indeed helps both denoising a small image patch and denoising a larger image consisting of those patches. Furthermore, it is found out that the proposed method improves even classification performance when test samples are corrupted with noise.

AB - Recently Burger et al. (2012) and Xie et al. (2012) proposed to use a denoising autoencoder (DAE) for denoising noisy images. They showed that a plain, deep DAE can denoise noisy images as well as the conventional methods such as BM3D and KSVD. Both of them approached image denoising by denoising small, image patches of a larger image and combining them to form a clean image. In this setting, it is usual to use the encoder of the DAE to obtain the latent representation and subsequently apply the decoder to get the clean patch. We propose that a simple sparsification of the latent representation found by the encoder improves denoising performance, both when the DAE was trained with and without sparsity regularization. The experiments confirm that the proposed sparsification indeed helps both denoising a small image patch and denoising a larger image consisting of those patches. Furthermore, it is found out that the proposed method improves even classification performance when test samples are corrupted with noise.

UR - http://www.scopus.com/inward/record.url?scp=84897512451&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84897512451&partnerID=8YFLogxK

M3 - Conference contribution

SP - 1469

EP - 1477

BT - 30th International Conference on Machine Learning, ICML 2013

PB - International Machine Learning Society (IMLS)

ER -