Presentation | 2002/12/6 Effective Temperature Scheduling for Multi-thread DAEM Masaharu TAKADA, Ryohei NAKANO, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | The EM algorithm is an efficient algorithm to obtain the ML estimate for incomplete data, but has the local optimality problem. The DAEM algorithm was once proposed to solve the problem, but is not guaranteed to obtain the global optimum since it employs a single token search. Then the multi-thread DAEM (m-DAEM) algorithm was proposed by incorporating a search framework of multiple tokens, giving further improvement of solution quality with a heavy computing cost. This paper proposes a very light version of m-DAEM, called ε-DAEM, by introducing an effective annealing where the Hessian information is made good use of. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | maximum likelihood / EM algorithm / DAEM algorithm / Gaussian mixture / annealing |
Paper # | NC2002-88 |
Date of Issue |
Conference Information | |
Committee | NC |
---|---|
Conference Date | 2002/12/6(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) | Effective Temperature Scheduling for Multi-thread DAEM |
Sub Title (in English) | |
Keyword(1) | maximum likelihood |
Keyword(2) | EM algorithm |
Keyword(3) | DAEM algorithm |
Keyword(4) | Gaussian mixture |
Keyword(5) | annealing |
1st Author's Name | Masaharu TAKADA |
1st Author's Affiliation | Department of Intelligence and Computer Science, Nagoya Institute of Technology() |
2nd Author's Name | Ryohei NAKANO |
2nd Author's Affiliation | Department of Intelligence and Computer Science, Nagoya Institute of Technology |
Date | 2002/12/6 |
Paper # | NC2002-88 |
Volume (vol) | vol.102 |
Number (no) | 508 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |