摘要
针对用生理信号识别情感中的最优情感特征组合的选择这一组合优化问题,将遗传算法中的交叉、变异操作加以改进形成新的算法。该算法用来选择最能代表相应情感状态的最优特征组合,并以最近邻法的分类正确率作为当前搜索到的最优特征组合的评价准则,对joy、anger、pleasure、sadness这4种情感状态进行识别,得到了较好的情感识别效果。仿真实验表明了该方法的可行性和有效性。
Taking into account the combinational optimization problem of emotion recognition with the physiological signals, a new algorithm that could adaptively adjust the probabilities of crossover and mutation is presented, the new algorithm is used to search the optimal feature subset which represented exactly the relevant affective states, the classified accuracy of the nearest neighbor is as the evaluation criterion that has searched the optimal feature subset. The new algorithm which has a better emotion recognition effect is used to recognize the following four affective states including Joy, anger, pleasure, sadness. The experimental results show that the method is feasible and effective.
出处
《计算机工程与设计》
CSCD
北大核心
2008年第14期3726-3728,3731,共4页
Computer Engineering and Design
基金
重庆市科委基金项目(CSTC
2006BB2028)
关键词
自适应遗传算法
最近邻法
最优情感特征组合
交叉
变异
adaptive genetic algorithm
nearest neighbor classification
optimal emotion feature subset
crossover
mutation