基于模拟退火及蜂群算法的优化特征选择算法

摘要

为了解决中文文本分类中初始特征空间维数过高带来的“维数灾难”问题,提高分类精度和分类效率,提出了一种基于模拟退火及蜂群算法的优化特征选择算法.该算法中,以蜂群算法流程为主体,根据蜜蜂群体觅食的特点快速寻找最优解,并且针对蜂群算法容易陷入局部最优解的问题,把模拟退火算法机制引入其中.该算法既保留了蜂群算法群体寻优的特点,又可以有效地避免陷入局部最优解.通过选择合适的收益率函数和温度下降函数,用实验的方法与卡方统计、信息增益和互信息等算法进行比较,表明了该算法的可行性和有效性.%In order to solve the "dimension disaster" caused by high-dimensional initial feature space in Chinese text classification,and improve the classification accuracy and efficiency,an algorithm based on the simulated annealing algorithm and artificial bee colony algorithm came up.In this algorithm,artificial bee colony algorithm process is used as the main body to find the optimal solution quickly according to the characteristics of the artificial bee colony algorithm,and the simulated annealing algorithm is introduced in order to make up for the shortcomings of artificial bee colony algorithm.This algorithm not only keeps the advantages of artificial bee colony algorithm but also effectively avoid local optimal solution.This algorithm is proved to be feasible and effective by choosing appropriate yield function and temperature decreasing function and comparing with Chi-square,information gain and mutual information algorithm through the experimental method.