用自适应蚁群算法的生理信号情感状态识别

Emotion recognition from physiological signal based on adaptive ACO

针对生理信号的情感识别问题,将蚁群优化算法用于情感生理信号特征选择,并采用自适应的适应度参数值、变异策略和临近位置交换策略对算法进行改进,使用K近邻法进行情感分类,以获得较高的识别率和有效特征组合。通过四种生理信号(EMG、SC、ECG、RSP)来识别四种情感(joy、anger、sadness、pleasure),实验仿真结果表明,将蚁群优化算法引入情感识别的研究是可行的。

Aiming at the problem about emotion recognition of physiological signals, a method of feature selection was based on Ant Colony Optimization （ACO）. The paper used adaptive fitness parameter, mutation strategy and neighbor position exchanged to improve the ACO algorithm, used k-nearest for recognizing emotion classify to obtain upper correct recognizing rate and the effective feature subset. The survey recognized 4 emotion states （joy, anger, sadness, pleasure） using 4 physiological signals（ EMG, SC, ECG, RSP）, Simulation shows that it is feasible to recognize emotion using Ant Colony Optimization algorithm.