基于遗传算法与支持向量机的癌症特征基因提取  

Cancer Feature Gene Extraction Based on Genetic Algorithm and Support Vector Machine

作　　者：唐铭一[1] 郑虹[1] 韩立权[1] TANG Ming-yi;ZHENG Hong;HAN Li-quan(Changchun University of Technology,Changchun 130012,China)

机构地区：[1]长春工业大学,吉林长春130012

出　　处：《电脑知识与技术》2020年第26期10-12,16,共4页Computer Knowledge and Technology

基　　金：吉林省教育厅项目(JJKH20181046KJ)。

摘　　要：针对癌症基因特征提取问题,根据遗传算法中不同迭代时期的种群特性,设计了新的突变方法。多突变基因库与种群代数相关的设计,使得算法能够较快地收敛到最优解而又避免其过早陷入局部最优解中;选择算子中包括个体对种群的基因丰富度贡献;针对种群中大量的重复个体,加入重复控制,去除重复个体,提高个体与种群基因的多样性。算法在几种实验数据集上均取得了较好的结果。A new mutation method was designed according to the population characteristics of different iteration periods in genetic algorithm to solve the problem of cancer gene extraction.Multi-mutation gene bank is designed to related with population algebra,so the algorithm converge to the optimal solution quickly and avoid falling into the local optimal solution too early.Selection opera⁃tor is designed to relate with population,including the contribution of individuals to the genetic richness of the population,the MIC evaluation of individuals,and the redundancy of genes within individuals,which makes the algorithm pay attention to both the pop⁃ulation and the characteristics of individuals.The genetic diversity of individuals and populations are improved by eliminating du⁃plicates.

关 键 词：遗传算法 支持向量机 特征提取 选择算子 变异算子 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]