关键词: 特征选择, 遗传算法, 两步式, 轮询式, 多准则

Abstract: For the low precision of feature selection under filter mode and high time cost of feature selection under wrapper mode, one new poll multi-criteria feature selection algorithm was proposed. This algorithm adopted chain-like agent genetic algorithm as searching algorithm, introduced principal criteria strategy to ensure the order of poll, thereby realizing multi-criteria feature selection algorithm. The experiments were conducted to compare this algorithm and several other feature selection algorithms. The experimental results show that this algorithm can obtain better precise selection result than several single evaluation criterion feature selection algorithms under filter mode, and less selection time cost than feature selection algorithm under wrapper mode.

Key words: feature selection, genetic algorithm, two-step mode, poll mode, multi-criteria