一种结合最小熵信息度量和粒子群优化算法的基因选取方法

A GENE SELECTION METHOD COMBINING MINIMUM ENTROPY INFORMATION METRIC AND PSO

提出一种结合最小熵信息度量和粒子群优化算法进行基因选取的方法。将每个粒子所代表的基因组合的信息度值作为该粒子的适应度函数值,通过粒子群的进化来获取最优基因子集,从而有利于实现样本分类。实验结果表明,该方法能够获取低冗余的信息基因,并在该方法选出的基因子集上,获得优于经典方法的分类准确率。

In the paper we propose a method to select gene which combines minimum entropy information metric and particle swarm optimi-sation.It takes the informative degree value of gene combination represented by every particle as the fitness function value of the particle, and obtains optimal gene subsets through particle swarm evolution, therefore it is beneficial to the implementation of sample classification.Experi-ment result shows, this method can obtain gene information with low redundancy as well as the classification accuracy rate better than classic methods based on the gene subsets selected by it.