二阶隐马尔可夫模型的学习算法及其与一阶隐马尔可夫模型的关系

应用科学学报 ›› 2001, Vol. 19 ›› Issue (1): 29-32.

二阶隐马尔可夫模型的学习算法及其与一阶隐马尔可夫模型的关系

史笑兴, 王太君, 何振亚

东南大学无线电工程系, 江苏南京 210096

收稿日期:1999-07-25 修回日期:2000-01-06 出版日期:2001-03-31 发布日期:2001-03-31
作者简介:史笑兴(1970-),男,江苏镇江人,博士;何振亚(1923-),男,江苏盐城人,教授,博导.

The Learning Algorithm of the Second Order HMM and Its Relationship with the First Order HMM

SHI Xiao-xing, WANG Tai-jun, HE Zhen-ya

Department of Radio Engineering, Southeast University, Nanjing 210096, China

Received:1999-07-25 Revised:2000-01-06 Online:2001-03-31 Published:2001-03-31

摘要/Abstract

摘要： 研究和推导了二阶隐马尔可夫模型(HMM2)的学习算法,其中包括前向、后向变量的定义与计算,Viter-bi算法,以及Baum-Welch算法.同时还研究了HMM2与一阶隐马尔可夫模型(HMM1)之间的关系,发现对任给的HMM2,总存在一个HMM1与之等价,提出了HMM2与HMM1的等价性定理,并给出了定理的证明.

关键词: 一阶隐马尔可夫模型, 二阶隐马尔可夫模型, 学习算法

Abstract: In this paper, we investigate the learning algorithm of the second order HMM (HMM2), which includes the forward variable algorithm, backward variable algorithm, Viterbi algorithm and Baum-Welch algorithm. Meanwhile, we study the relationship between the HMM2 and the first order HMM (HMM1), and find that for any HMM2 there always exists a HMM1 which is equal to the HMM2. We propose and prove the equivalent theorem of the HMM2 and HMM1.

Key words: the second order HMM (HMM2), learning algorithm, the first order HMM (HMM1)

中图分类号:

TN912.34

史笑兴, 王太君, 何振亚. 二阶隐马尔可夫模型的学习算法及其与一阶隐马尔可夫模型的关系[J]. 应用科学学报, 2001, 19(1): 29-32.

SHI Xiao-xing, WANG Tai-jun, HE Zhen-ya. The Learning Algorithm of the Second Order HMM and Its Relationship with the First Order HMM[J]. Journal of Applied Sciences, 2001, 19(1): 29-32.

[1]	黄杰，倪鹏宇. 基于人体关键部位的不良图像过滤[J]. 应用科学学报, 2014, 32(4): 416-422.
[2]	冯　超;李柠;李少远. 递增的稀疏神经网络研究 [J]. 应用科学学报, 2008, 26(2): 194-194 .