Presentation 1998/3/20
EM Algorithm with Split and Merge Operations and its application to Mixture Density Estimation
Naonori UEDA, Ryohei NAKANO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) The EM algorithm has been been extensively used for a wide variety of parameter estimation problems with hidden or latent variables to compute the maximum likelihood estimates. The algorithm, however, suffers from the local maxima problem in practice. In this report, focusing on the case of discrete hidden variables, we present a split and merge EM algorithm to overcome the local maxima problem. We apply the proposed algorithm to Gaussian mixture density estimation problem and show that the algorithm could impressively improve likelihood values.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) EM algorithm / Maximum likelihood estimation / Gaussian mixture estimation problem / Split& Merge operations
Paper #
Date of Issue

Conference Information
Committee NC
Conference Date 1998/3/20(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 Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) EM Algorithm with Split and Merge Operations and its application to Mixture Density Estimation
Sub Title (in English)
Keyword(1) EM algorithm
Keyword(2) Maximum likelihood estimation
Keyword(3) Gaussian mixture estimation problem
Keyword(4) Split& Merge operations
1st Author's Name Naonori UEDA
1st Author's Affiliation NTT Communication Science Laboratories()
2nd Author's Name Ryohei NAKANO
2nd Author's Affiliation NTT Communication Science Laboratories
Date 1998/3/20
Paper #
Volume (vol) vol.97
Number (no) 624
Page pp.pp.-
#Pages 8
Date of Issue