Applying dual-structure particle swarm optimization and KNN to identify affective states ground on physiological signals
双重结构粒子群和KNN在生理信号情感识别中的应用

Dual-Structure Particle Swarm Optimization (DSPSO) was applied to select emotion features of physiological signals, which improved the effect of feature selection and the correct rate of affective classification. Incremental K algorithm was proposed to avoid indivisibility for multi-classification, which also advanced multi-identification effect. Compared with the results of traditional SFFS and BPSO algorithms, the proposed method obtained better effect in recognizing four affective states (joy, anger, sadness, pleasure) from four physiological signals (EMG, SC, ECG, RSP). Simulation results show, based on physiological signals, DSPSO can select feature well.