An extended learning algorithm based on duration HMM

DU Shi-ping · jOURNAL OF southwest University for Nationalities · 2008

In This paper,new probability of state duration is defined,and the structure of modified Hidden Markov Models(HMM) with duration is discussed.New forward-backward variables are defined and also recursive formulas are given about forward-backward algorithm based on traditional HMM.In addition,the reestimation formulas of the parameters about new models are brought forward using new forward-backward algorithm.

Read the paper · More papers on PaperTik