基于蚁群算法的基因联接学习遗传算法

论文提出了一种基于蚁群算法的基因联接学习遗传算法。在该算法中遗传算法的种群对应于蚁群,遗传算法的染色体同时是蚁群算法的一只蚂蚁。在每一次进行交叉或突变操作时,算法首先根据蚁群算法的信息素矩阵计算父代个体的基因间联接强度,然后根据该联接强度选择交叉和突变位点。这样可以避免积木块过多地被遗传操作所破坏,减少遗传算法的搜索空间,并指引寻优的方向。联接学习在该算法中是并行进行的,而在Harik的算法中是串行进行的;该算法的编码长度不会随着等位基因数量的增加而成倍地增加。文章通过有界难度问题和TSP问题的实验研究验证了算法的有效性。

An ACO-based Linkage Learning Genetic Algorithm

This pa per proposes an Ant Colony Optimization(ACO)based Linkage Learning Genetic Al gorithm(LLGA).In this approach,every individual(chromosome )of GA is at th e same time an ant of ACO.Whenever GA performs the oper-ation of crossover and mutation,the linkage strength of parent chromosome (s)is computed accor ding to the pheromone matrix of ACO and is used to guide the selection of crosso ver or mutation point (s).The stronger linkage strength of two neighboring g enes is,the less probably these genes will be separated by genetic operations. Different form Harik's LLGA,this approach learns the linkage parallel and the chromosome length will not multiple as the number of allele increase.Experimen ts have validated the scheme.