In order to solve the problem that the krill herd(KH)algorithm is premature due to the decrease of population diversity,a new hybrid vortex search KH(VSKH)algorithm has been developed to deal with the global optimizat...In order to solve the problem that the krill herd(KH)algorithm is premature due to the decrease of population diversity,a new hybrid vortex search KH(VSKH)algorithm has been developed to deal with the global optimization problem.The improvement is that a new hybrid vortex search(HVS)operator is added into the updating process of the krill for the purpose of dealing with optimization problems more efficiently.Using 20 benchmark functions for comparison experiments,the results show that the VSKH algorithm has higher accuracy.展开更多
基金Supported by the National Natural Science Foundation of China(11871383,71471140)。
文摘In order to solve the problem that the krill herd(KH)algorithm is premature due to the decrease of population diversity,a new hybrid vortex search KH(VSKH)algorithm has been developed to deal with the global optimization problem.The improvement is that a new hybrid vortex search(HVS)operator is added into the updating process of the krill for the purpose of dealing with optimization problems more efficiently.Using 20 benchmark functions for comparison experiments,the results show that the VSKH algorithm has higher accuracy.
文摘针对核模糊C-均值算法(kernel fuzzy C-means,KFCM)随机选择初始聚类中心而不能获得全局最优且在聚类中心较近或重合时易产生一致性聚类等问题,提出一种改进算法。改进算法在原目标函数中引入中心极大化约束项来调控簇间分离度,从而避免算法出现一致性聚类结果。利用磷虾群算法对基于新目标函数的KFCM算法进行优化,使算法不再依赖初始聚类中心,提高算法的稳定性。基于距离最大最小原则产生多组较优的聚类中心作为初始磷虾群体并在算法迭代过程中融合一种新的精英保留策略,从而确保算法收敛到全局极值;通过对个体随机扩散活动进行分段式Logistic混沌扰动,提高算法全局寻优能力。使用KDD Cup 99入侵检测数据进行仿真实验表明,改进算法具有更好的检测性能,解决了传统的聚类算法在入侵检测中稳定性差、检测准确率低的问题。
