基于ACO及PSO的特征选择算法

Feature Selection Algorithm Based on ACO and PSO

在属性约简的进化算法中,算法时间存在复杂度高、搜索空间大等不足.为此文中引入最小冗余度的属性重要性后,提出一种基于蚁群优化(ACO)和粒子群优化(PSO)的进化特征选择算法,利用PSO算法的快速简洁等优点得到ACO的初始路径,以此减少迭代次数,加快算法的收敛速度;同时,利用蚂蚁之间的可并行性,采用分布式技术实现蚂蚁之间的并行搜索,改进了算法的效率.理论分析及实验结果表明,文中的算法是有效可行的.

Currently, there are many evolutionary algorithms available for attribute reduction, but high time complexity and wide search space are their common disadvandages. Therefore, the paper,after introducing the strategy of evaluating the＂ significance of attribute based on minimum reducdant degree, presents a novel evolutionary .algorithm based on ACO and PSO for attribute reduction. By employing the merits of PSO algorithm for its high efficiency and concision, the proposed algorithm can obtain efficient intitial paths, whereby reducing iterative times and accelenating convergence. At the same time, using the parallelizability of ants and distributed parapllelized search technology,the performance of the algorithm can be efficiently improved. The theoretical analysis and experimental results show that the algorithm of this paper is flexible.