【摘要】 针对离散粒子群优化算法进行基因特征选择容易陷入局部最优解的问题,提出一种基于离散粒子群优化和邻域约简的组合优化算法。利用邻域约简挖掘基因数据本身蕴含知识的特点,依据决策属性对条件子集的依赖度构造离散粒子群优化算法中的优化函数,根据优化函数值的大小引导粒子搜索最优基因特征子集,从而解决局部最优的问题。实验结果表明,与粒子群优化和遗传算法的混合优化算法、优化的邻域粗糙集等算法相比,该算法能够获得较高的分类准确度。更多还原

【Abstract】 Aiming at the problem that the gene feature selection based on discrete Particle Swarm Optimization( PSO)algorithm is easy to fall into local optimal solution,a combined optimization algorithm based on discrete PSO and neighborhood reduction is proposed. Neighborhood reduction can effectively excavate gene data. The optimization function of the discrete PSO algorithm is constructed based on the dependence of the conditional subset of decision attributes. According to the optimization function value,it guides the particle to search the optimal gene feature subset,so as to solve the problem of local optimum. Experimental results showthat the algorithm can obtain higher classification accuracy compared with hybrid optimization algorithm based on PSO and genetic algorithm,optimized neighborhood rough set,etc.更多还原