Optimization of Discrete Hidden Markov Model Based on an Improved Tabu Search Algorithm
Jianghua Liu, Jiapin Chen, Junshi Cheng · Computer Engineering and Applications Journal · 2003
隐马尔可夫模型(HMM,Hidden Markov Model)是语音识别和手势识别中广泛使用的统计模式识别方法.文章提出了一种改进的禁忌搜索(ITS,Improved Tabu Search)优化HMM的参数.传统的Tanu Search(TS)与局部搜索算法(极大似然法)交替进行,从而加快了算法的收敛速度,并得到优化解.分别用TS及ITS训练隐马尔可夫模型进行动态手势识别.结果表明ITS可获得更高的识别率,且能达到全局优化.