Composite denoising autoencoders

Krzysztof Geras, Charles Sutton

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

Abstract

In representation learning, it is often desirable to learn features at different levels of scale. For example, in image data, some edges will span only a few pixels, whereas others will span a large portion of the image. We introduce an unsupervised representation learning method called a composite denoising autoencoder (CDA) to address this. We exploit the observation from previous work that in a denoising autoencoder, training with lower levels of noise results in more specific,
fine-grained features. In a CDA, different parts of the network are trained with different versions of the same input, corrupted at different noise levels. We introduce a novel cascaded training procedure which is designed to avoid types of bad solutions that are specific to CDAs. We show that CDAs learn effective representations on two different image data sets.
Original languageEnglish
Title of host publicationMachine Learning and Knowledge Discovery in Databases
Subtitle of host publicationEuropean Conference on Machine Learning and Knowledge Discovery in Databases ECML PKDD 2016
PublisherSpringer International Publishing
Pages681-696
Number of pages16
ISBN (Electronic)978-3-319-46128-1
ISBN (Print)978-3-319-46127-4
DOIs
Publication statusPublished - 4 Sep 2016
EventEuropean Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases 2016 - Riva del Garda, Italy
Duration: 19 Sep 201623 Sep 2016
http://www.ecmlpkdd2016.org/

Publication series

NameLecture Notes in Computer Science (LNCS)
PublisherSpringer International Publishing
Volume9851
ISSN (Print)0302-9743

Conference

ConferenceEuropean Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases 2016
Abbreviated titleECML-PKDD 2016
Country/TerritoryItaly
CityRiva del Garda
Period19/09/1623/09/16
Internet address

Fingerprint

Dive into the research topics of 'Composite denoising autoencoders'. Together they form a unique fingerprint.

Cite this