Presentation 2015-03-19
Semi-Supervised Low-rank GMM Learning for Multimodal Distribution
Eita AOKI, Tetsuya MATSUMOTO, Noboru OHNIHSHI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In this study, we examine the learning method of each category distribution under underdetermined environment that category information of few data are known. Because we use multimodal data distribution, learning for each category distribution is difficult. Therefore, by using the given category information, and projecting the data into low-dimensional subspace suitable for category classification, we propose a method to GMM approximation. Specifically, we consider each category distribution as a linear combination of subclasses of the multidimensional normal distribution, and determine the projection space by appropriate criteria. In the experiment, we compared the proposed method with the conventional method by changing the parameters. In some cases, the proposed method showed higher accuracy than the conventional method.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Semi-Supervised Learning / Dimensionality Reduction / Multi-Modality
Paper # BioX2014-60,PRMU2014-180
Date of Issue

Conference Information
Committee PRMU
Conference Date 2015/3/12(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Pattern Recognition and Media Understanding (PRMU)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Semi-Supervised Low-rank GMM Learning for Multimodal Distribution
Sub Title (in English)
Keyword(1) Semi-Supervised Learning
Keyword(2) Dimensionality Reduction
Keyword(3) Multi-Modality
1st Author's Name Eita AOKI
1st Author's Affiliation Graduate School of Information Science, Nagoya University()
2nd Author's Name Tetsuya MATSUMOTO
2nd Author's Affiliation Graduate School of Information Science, Nagoya University
3rd Author's Name Noboru OHNIHSHI
3rd Author's Affiliation Graduate School of Information Science, Nagoya University
Date 2015-03-19
Paper # BioX2014-60,PRMU2014-180
Volume (vol) vol.114
Number (no) 521
Page pp.pp.-
#Pages 6
Date of Issue