基于动态概率引导的人工蜂群聚类算法

Artificial Bee Colony Clustering Algorithm Based on Dynamic Probability Guidance

针对传统人工蜂群算法中搜索策略开发能力不足、单一的搜索策略难以适用于算法运行的各个阶段等问题,提出了一种搜索策略动态调整的人工蜂群算法,该算法搜索策略由基于反馈的动态概率引导以平衡算法的探索能力和开发能力;为增强蜜源结构相似性的联系与优秀蜜源的影响,引入局部最优蜜源引导下一代的产生。在此基础上,针对K-means算法初始聚类中心敏感、全局搜索能力不足等问题提出了基于改进蜂群算法的K-means算法,在多个标准测试函数和UCI数据集上测试验证所提出算法的性能。

Aiming at the problem that the traditional artificial bee colony algorithm has insufficient development ability of search strategy and the single search strategy is difficult to apply to all stages of algorithm operation,an artificial bee colony algorithm based on dynamic adjustment of search strategy is proposed.The search strategy is based on feedback.The dynamic probabilistic guidance algorithm is adjusted to balance the exploration ability and development ability of the algorithm;to enhance the relationship between the similarity of the honey source structure and the influence of the excellent honey source,the local optimal honey source is introduced to guide the generation of the next generation.On this basis,the K-means algorithm based on improved bee colony algorithm is proposed for the K-means algorithm,which is sensitive to initial clustering center and insufficient global search ability.It is proposed by test and verification on multiple standard test functions and UCI data sets.