A feature selection method,FS_FGCEIFA,was proposed based on fuzzy granularity conditional entropy and improved firefly algorithm(FA).By combining the knowledge granularity and λ-conditional entropy in fuzzy rough sets,a new information entropy model-fuzzy granularity conditional entropy was put forward;the fuzzy granularity conditional entropy was applied to FA,and a fitness function based on fuzzy granularity conditional entropy was proposed;the calculation strategy of particle inertia mass and universal gravity in the gravity search algorithm was adopted to adjust the brightness and attractiveness of fireflies in FA,and the adaptive step size factor based on gravity search was integrated into the position update of FA.Experiments on multiple UCI datasets and software defect prediction datasets show that FS_FGCEIFA could achieve better classification performance.
关键词
特征选择/萤火虫算法/引力搜索算法/模糊粒度条件熵/模糊粗糙集
Key words
feature selection/firefly algorithm/gravity search algorithm/fuzzy granularity conditional entropy/fuzzy rough set
维普
万方数据