DocumentCode :
464003
Title :
A Q-EM Based Simulated Annealing Algorithm for Finite Mixture Estimation
Author :
Wenbin Guo ; Shuguang Cui
Author_Institution :
Dept. of ECE, Arizona Univ., Tucson, AZ, USA
Volume :
3
fYear :
2007
fDate :
15-20 April 2007
Abstract :
We develop a q-expectation maximization (q-EM) simulated annealing method for parameter estimation. The g-EM algorithm is a one-parameter generalization of normal expectation maximization (EM) algorithm based on Tsallis entropy. By incorporating the simulated annealing method, we propose the q-deterministic annealing expectation maximization (q-DAEM) algorithm. Given the inherent connection between a physical annealing process and statistical mechanics, we show that the proposed algorithm actually minimizes a counterpart of the free energy in statistical mechanics by controlling an effective temperature. Simulations of mixed Gaussian parameter estimation show that the proposed method is much less initialization-dependent than the standard EM algorithm and converges dramatically faster than the DAEM algorithm.
Keywords :
Gaussian processes; expectation-maximisation algorithm; simulated annealing; statistical mechanics; Tsallis entropy; finite mixture estimation; mixed Gaussian parameter estimation; one-parameter generalization; physical annealing process; q-EM based simulated annealing algorithm; q-deterministic annealing expectation maximization algorithm; q-expectation maximization simulated annealing method; statistical mechanics; Clustering algorithms; Entropy; Parameter estimation; Physics; Probability; Simulated annealing; Solids; Stochastic processes; Temperature control; Temperature distribution; DAEM; Estimation; Expectation Maximization; Simulated Annealing; Tsallis Entropy;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Acoustics, Speech and Signal Processing, 2007. ICASSP 2007. IEEE International Conference on
Conference_Location :
Honolulu, HI
ISSN :
1520-6149
Print_ISBN :
1-4244-0727-3
Type :
conf
DOI :
10.1109/ICASSP.2007.366879
Filename :
4217909
Link To Document :
بازگشت