论文部分内容阅读
针对无监督分类问题,提出一种多尺度并行免疫克隆优化聚类算法.算法中,进化在多个子群之间并行进行,不同子群的抗体根据子群适应度采用不同变异尺度.进化初期,利用大尺度变异子群实现全局最优解空间的快速定位,同时变异尺度随着适应值的提升逐渐降低;进化后期,利用小尺度变异子群完成局部解空间的精确搜索.将新算法与其他聚类算法进行比较,所得结果表明新算法具有较好的聚类性能和鲁棒性.
Aiming at the problem of unsupervised classification, a multi-scale parallel immune clonal clustering algorithm is proposed. In the algorithm, evolution proceeds in parallel between multiple subgroups, and different subgroups adopt different mutation scales according to subgroup fitness.At the early stage of evolution, In the later stage of evolution, the accurate search of the local solution space is accomplished by using the subgroups of large-scale mutation, and the new algorithm is compared with other algorithms The results show that the new algorithm has better clustering performance and robustness.