基于邻域熵与蚁群优化的基因选择算法

Gene selection algorithm based on neighborhood entropy and ant colony optimization

针对基因表达数据集的基因选择问题,采用邻域熵度量与蚁群优化原理,提出一种基因选择方法.首先,引入邻域粗糙集模型对基因数据进行邻域粒化,定义邻域熵度量用于剔除冗余基因构成预选择基因子集;然后,采用邻域熵构造基因重要度作为启发式信息,发挥蚁群优化算法的分布式、正反馈及全局寻优的优势,运用蚁群优化算法从预选择基因子集中搜索出最小基因子集;最后,在选取的最小基因子集上进行分类测试.实验表明:建立在该最小基因子集上的分类器具有良好的分类性能.

To deal with the problem of selecting gene subset in a gene dataset,a novel gene selection method is proposed. Firstly,a neighborhood rough set model is introduced to granulate the gene data.The neighborhood entropy is defined for measuring the uncertainty of gene data and removing the redundant genes to constitute a pre-selected subset. Furthermore,the neighborhood entropy based gene importance is constructed as the heuristic information in the proposed ACO algorithm,which has the advantages of distributed,positive feedback and global optimization. The proposed algorithm has a good ability for finding the minimum critical gene subset from the pre-selected set. Finally,the classification experiments are carried out on the selected genes. The results show that the classifier constructed on the selected genes has a good classification performance.