摘要
针对心电(ECG)信号情感识别中特征选择的问题,首先运用相关性分析方法,去除原始特征集中的高相关度特征,实现原始特征集的降维;其次,为了在降维后的特征空间中进行有效的特征选择,提出了一种改进的二进制量子粒子群算法(SBQPSO)。实验结果表明,基于本算法结合Fisher分类器建立的ECG信号情感识别系统能够对高兴、惊奇、厌恶、悲伤、愤怒和恐惧6种情感达到良好的识别效果。
This paper discussed the feature selection from ECG signal in affective recognition. At first, the original features with high correlation were deleted to reduce dimensionality of original feature set by correlation analysis. And then, an improved quantum-behaved particle swarm optimization with binary encoding algorithm was proposed to achieve effective feature selection in the feature space with reduced dimension. The experimental results shows that the affective recognition system based on this algorithm and fisher classifier recognize the anger, disgust, fear, grief, joy and surprise successfully.
出处
《计算机科学》
CSCD
北大核心
2012年第3期209-211,221,共4页
Computer Science
基金
国家自然科学基金(60873143)
国家重点学科基础心理学科研基金(NKFS07003)
中央高校基本科研业务费专项资金(XDJK2009B008)资助
