Guiding Generative Models via Class Label Information

dc.contributor.advisorTaylor, Graham W.
dc.contributor.authorRudy, Jan
dc.date.accessioned2016-01-22T18:04:00Z
dc.date.available2016-01-22T18:04:00Z
dc.date.copyright2016-01
dc.date.created2015-12-22
dc.date.issued2016-01-22
dc.degree.departmentSchool of Engineeringen_US
dc.degree.grantorUniversity of Guelphen_US
dc.degree.nameMaster of Applied Scienceen_US
dc.degree.programmeEngineeringen_US
dc.description.abstractGiven a finite number of samples from some high-dimensional distribution, the task of efficiently and accurately modeling the distribution can be challenging. Some datasets, however, provide additional information (e.g. categorical class labels) for each input. When class labels are available, can they be used to better model the data distribution? A conditional modeling and training procedure is introduced for a type of generative model (the generalized denoising autoencoder) and two methods of injecting class label information are presented (additive vs. multiplicative). When trained on natural images, models with access to class information generate samples of higher visual fidelity than those trained on images alone. Additionally, with higher dimensional data, multiplicative architectures outperform their additive counterparts. Finally, experimental results confirm recent findings that Parzen likelihood estimates are a poor measure of visual sample quality.en_US
dc.description.sponsorshipNatural Sciences and Engineering Research Council of Canada Discovery Grant #436126
dc.identifier.urihttp://hdl.handle.net/10214/9498
dc.language.isoenen_US
dc.publisherUniversity of Guelphen_US
dc.rightsAttribution-NonCommercial-ShareAlike 2.5 Canada*
dc.rights.urihttp://creativecommons.org/licenses/by-nc-sa/2.5/ca/*
dc.subjectDeep Learningen_US
dc.subjectGenerative modelsen_US
dc.subjectneural networksen_US
dc.subjectartificial intelligenceen_US
dc.subjectmachine learningen_US
dc.subjectgated modelsen_US
dc.titleGuiding Generative Models via Class Label Informationen_US
dc.typeThesisen_US

Files

Original bundle
Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
Rudy_Jan_201601_MASc.pdf
Size:
99.02 MB
Format:
Adobe Portable Document Format
Description: