Presentation 2013-07-12
Variational Bayes in Directional Statistics
Toshihisa TANAKA, Masayuki KOBAYASHI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Estimation of the parameters of distribution with observed data is crucial in signal processing and machine learning. A well-known estimation method is Bayesian estimation, which is useful for restraining overfitting and less affected by outliers compared to classical maximum likelihood estimation. This article derives an estimator of the parameters in a von Mises mixture model, which is an important distribution in directional statistics. More specifically, we focus on the fact that the von Mises distribution is derived from the bivariate Gaussian distribution. Based on this, a variational Bayesian method for a mixture Gaussian model is applied to estimating the parameters of the mixture von Mises model. Experimental results support the analysis.
Keyword(in Japanese) (See Japanese page)
Keyword(in English)
Paper # CAS2013-21,VLD2013-31,SIP2013-51,MSS2013-21
Date of Issue

Conference Information
Committee VLD
Conference Date 2013/7/4(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 VLSI Design Technologies (VLD)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Variational Bayes in Directional Statistics
Sub Title (in English)
Keyword(1)
1st Author's Name Toshihisa TANAKA
1st Author's Affiliation Department of Electrical and Electronic Engineering Tokyo University of Agriculture and Technology()
2nd Author's Name Masayuki KOBAYASHI
2nd Author's Affiliation Department of Electrical and Electronic Engineering Tokyo University of Agriculture and Technology
Date 2013-07-12
Paper # CAS2013-21,VLD2013-31,SIP2013-51,MSS2013-21
Volume (vol) vol.113
Number (no) 119
Page pp.pp.-
#Pages 6
Date of Issue