Presentation 2017-09-08
On Two Part Coding of Gaussian Mixture Models
Kohei Miyamoto, Masanori Kawakita, Jun'ichi Takeuchi,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) The final goal of this work is model sellection for gaussian mixturemodels(GMM) based on the minimum descriptionlength(MDL) principle. In this paper, we discuss some issues about atwo part coding for GMM for this purpose. Firstwe consider mixture models which have known distributions as theircomponents. We introduce a two part codesfor that mixture models and derive a risk bound for typical sequencesand a probabilistic loss bound when the R'enyidivergence is used as a loss function. Next we examine their Fisherinformation. its determinant and Jeffrey's priorwhich is related to the restriction on parameters. These works will beuseful when we consider a two part coding ofGMM itself.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Gaussian mixture models / two part coding / MDL principle
Paper # IT2017-47
Date of Issue 2017-09-01 (IT)

Conference Information
Committee IT
Conference Date 2017/9/8(1days)
Place (in Japanese) (See Japanese page)
Place (in English) Centcore Yamaguchi Hotel
Topics (in Japanese) (See Japanese page)
Topics (in English) error correcting codes, general
Chair Masayoshi Ohashi(Fukuoka Univ.)
Vice Chair Jun Muramatsu(NTT)
Secretary Jun Muramatsu(Yokohama College of Commerce)
Assistant Sigeaki Kuzuoka(wakayama univ.)

Paper Information
Registration To Technical Committee on Information Theory
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) On Two Part Coding of Gaussian Mixture Models
Sub Title (in English)
Keyword(1) Gaussian mixture models
Keyword(2) two part coding
Keyword(3) MDL principle
1st Author's Name Kohei Miyamoto
1st Author's Affiliation Kyushu University(Kyushu Univ.)
2nd Author's Name Masanori Kawakita
2nd Author's Affiliation Kyushu University(Kyushu Univ.)
3rd Author's Name Jun'ichi Takeuchi
3rd Author's Affiliation Kyushu University(Kyushu Univ.)
Date 2017-09-08
Paper # IT2017-47
Volume (vol) vol.117
Number (no) IT-208
Page pp.pp.49-54(IT),
#Pages 6
Date of Issue 2017-09-01 (IT)