DC Field | Value | Language |
---|---|---|
dc.contributor.author | Joo, Weonyoung | ko |
dc.contributor.author | Lee, Wonsung | ko |
dc.contributor.author | Park, Sungrae | ko |
dc.contributor.author | Moon, Il-Chul | ko |
dc.date.accessioned | 2021-03-26T01:34:18Z | - |
dc.date.available | 2021-03-26T01:34:18Z | - |
dc.date.created | 2020-08-19 | - |
dc.date.created | 2020-08-19 | - |
dc.date.created | 2020-08-19 | - |
dc.date.issued | 2020-11 | - |
dc.identifier.citation | PATTERN RECOGNITION, v.107 | - |
dc.identifier.issn | 0031-3203 | - |
dc.identifier.uri | http://hdl.handle.net/10203/281842 | - |
dc.description.abstract | This paper proposes Dirichlet Variational Autoencoder (DirVAE) using a Dirichlet prior. To infer the parameters of DirVAE, we utilize the stochastic gradient method by approximating the inverse cumulative distribution function of the Gamma distribution, which is a component of the Dirichlet distribution. This approximation on a new prior led an investigation on the component collapsing, and DirVAE revealed that the component collapsing originates from two problem sources: decoder weight collapsing and latent value collapsing. The experimental results show that 1) DirVAE generates the result with the best log-likelihood compared to the baselines; 2) DirVAE produces more interpretable latent values with no collapsing issues which the baselines suffer from; 3) the latent representation from DirVAE achieves the best classification accuracy in the (semi-)supervised classification tasks on MNIST, OMNIGLOT, COIL-20, SVHN, and CIFAR-10 compared to the baseline VAEs; and 4) the DirVAE augmented topic models show better performances in most cases. | - |
dc.language | English | - |
dc.publisher | ELSEVIER SCI LTD | - |
dc.title | Dirichlet Variational Autoencoder | - |
dc.type | Article | - |
dc.identifier.wosid | 000552866000055 | - |
dc.identifier.scopusid | 2-s2.0-85087402885 | - |
dc.type.rims | ART | - |
dc.citation.volume | 107 | - |
dc.citation.publicationname | PATTERN RECOGNITION | - |
dc.identifier.doi | 10.1016/j.patcog.2020.107514 | - |
dc.contributor.localauthor | Moon, Il-Chul | - |
dc.contributor.nonIdAuthor | Lee, Wonsung | - |
dc.contributor.nonIdAuthor | Park, Sungrae | - |
dc.description.isOpenAccess | N | - |
dc.type.journalArticle | Article | - |
dc.subject.keywordAuthor | Representation learning | - |
dc.subject.keywordAuthor | Variational autoencoder | - |
dc.subject.keywordAuthor | Deep generative model | - |
dc.subject.keywordAuthor | Multi-modal latent representation | - |
dc.subject.keywordAuthor | Component collapse | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.