摘要
根据隐马尔可夫模型HMM的基本理论和算法设计了一个情感模型。该模型由三种心理情绪状态组成。外部刺激经过HMM模型识别,确定情感输出。Baum-Welch算法在训练时存在溢出问题,需要人工介入来调整。提出一种能消除HMM情感模型上下溢问题的算法,不仅不需人工介入,还保证了计算精度,从而提高了模型的准确性。仿真结果表明这种方案可行,且识别效果较好。
This paper deals with the design of the affecfive model based on the principal theory and algorithm of the Hidden Markov Model .The model consists of three emotional states . Through the HMM ,the affective models are identified and the accuracy of the model is enhanced. In this paper, an algorithm is presented to avoid the overflow completely o With this algorithm manual adjustment is not needed and the computation accuracy is guaranteed. Experiment results show that the method has higher classification accuracy.
出处
《辽宁工程技术大学学报（自然科学版）》
EI
CAS
北大核心
2006年第4期635-637,共3页
Journal of Liaoning Technical University (Natural Science)
基金
国家自然科学基金资助项目(60573059)
关键词
情感模型
隐马尔可夫模型
算法
affective model
hidden markov model
algorithm
