作者: Yann Guédon
DOI: 10.1002/(SICI)1526-4025(199907/09)15:3<195::AID-ASMB376>3.0.CO;2-F
关键词:
摘要: We propose a computational approach for implementing discrete hidden semi-Markov chains. A chain is composed of non-observable or process which finite and observable process. Hidden chains possess both the flexibility Markov approximating complex probability distributions representing temporal structures. Efficient algorithms computing characteristic organized according to intensity, interval counting points view are described. The proposed in conjunction with statistical inference previously makes powerful model analysis samples non-stationary sequences. (Resume d'auteur)