计算机工程与应用 ›› 2012, Vol. 48 ›› Issue (18): 57-60.

• 研究、探讨 • 上一篇    下一篇

HMM参数估计的Gibbs抽样算法

朱成文1,李  兵2,胡  奎3   

  1. 1.武汉军械士官学校,武汉 430075
    2.国防科技大学 理学院,长沙 410073
    3.中国人民解放军92941部队96分队
  • 出版日期:2012-06-21 发布日期:2012-06-20

Algorithm of parameter estimation of HMM via Gibbs sampling

ZHU Chengwen1, LI Bing2, HU Kui3   

  1. 1.Wuhan Ordnance Non-commissioned Officer Academy PLA, Wuhan 430075, China
    2.College of Science, National University of Defense Technology, Changsha 410073, China
    3.Unit 92941 of PLA, China
  • Online:2012-06-21 Published:2012-06-20

摘要: 隐马氏模型(HMM)的参数估计是隐马氏模型各种应用的关键。经典的Baum-Welch算法容易陷入局部最优,对初始参数的要求苛刻。HMM参数估计的Gibbs抽样法,充分利用模型先验信息,借助马氏链蒙特卡洛方法(MCMC)的强大计算功能,避免了陷入局部最优,有更好的效果。

关键词: 隐马氏模型, Gibbs抽样, 共轭先验

Abstract: The parameter estimation of Hidden Markov Model(HMM) is critical to all its applications. The classic Baum-Welch algorithm is not flexible with the initial parameters and is easy to fall into the local optimal solution. The great computational power of MCMC is employed. The algorithm of parameter estimation of HMM via Gibbs sampling avoids the local optimal solution and can be more effective.

Key words: Hidden Markov Models(HMM), Gibbs sampling, conjugate priors