Presentation | 2015-09-15 Improvement of nonnegative matrix factorization for probabilistic latent semantic analysis Toshio Uchiyama, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Nonnegative Matrix Factorization (NMF) is equivalent to PLSA (Probabilistic Latent Semantic Analysis), when it uses generarized KL divergence as target function. This paper proposes a novel method for initializing base matrix using smoothing to improve algorithms for NMF and PLSA. It also shows that perplexity of training data and test data are correlated. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | NMF(Nonnegative Matrix Factorization) / competitive learning / clustering / topic model / PLSA |
Paper # | PRMU2015-83,IBISML2015-43 |
Date of Issue | 2015-09-07 (PRMU, IBISML) |
Conference Information | |
Committee | PRMU / IBISML / IPSJ-CVIM |
---|---|
Conference Date | 2015/9/14(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | Eisaku Maeda(NTT) / Takashi Washio(Osaka Univ.) |
Vice Chair | Shuji Senda(NEC) / Seiichi Uchida(Kyushu Univ.) / Kenji Fukumizu(ISM) / Masashi Sugiyama(Tokyo Inst. of Tech.) |
Secretary | Shuji Senda(Mie Univ.) / Seiichi Uchida(DENSO IT Lab.) / Kenji Fukumizu(ISM) / Masashi Sugiyama(Kyoto Univ.) |
Assistant | Kazuaki Kondo(Kyoto Univ.) / Akisato Kimura(NTT) / Koji Tsuda(Univ. of Tokyo) / Hisashi Kashima(Kyoto Univ.) |
Paper Information | |
Registration To | Technical Committee on Pattern Recognition and Media Understanding / Technical Committee on Infomation-Based Induction Sciences and Machine Learning / Special Interest Group on Computer Vision and Image Media |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Improvement of nonnegative matrix factorization for probabilistic latent semantic analysis |
Sub Title (in English) | |
Keyword(1) | NMF(Nonnegative Matrix Factorization) |
Keyword(2) | competitive learning |
Keyword(3) | clustering |
Keyword(4) | topic model |
Keyword(5) | PLSA |
1st Author's Name | Toshio Uchiyama |
1st Author's Affiliation | Hokkaido Information University(HIU) |
Date | 2015-09-15 |
Paper # | PRMU2015-83,IBISML2015-43 |
Volume (vol) | vol.115 |
Number (no) | PRMU-224,IBISML-225 |
Page | pp.pp.117-122(PRMU), pp.117-122(IBISML), |
#Pages | 6 |
Date of Issue | 2015-09-07 (PRMU, IBISML) |