Adaptive affinity propagation method based on improved cuckoo search
作者：
Highlights：
• The proposed CAAP method can improve the clustering performances of AP and adaptive AP methods.
• The parameter “preference” of AP can be adapted by the silhouette index.
• In CAAP, the damping factor can be adjusted to eliminate the oscillation of AP.
• An improved CS algorithm that has strong ability to explore global minima is proposed.
• A coefficient about evolutional ratio to modify the step length and discovery probability of CS.
摘要
•The proposed CAAP method can improve the clustering performances of AP and adaptive AP methods.•The parameter “preference” of AP can be adapted by the silhouette index.•In CAAP, the damping factor can be adjusted to eliminate the oscillation of AP.•An improved CS algorithm that has strong ability to explore global minima is proposed.•A coefficient about evolutional ratio to modify the step length and discovery probability of CS.
论文关键词：Clustering,Affinity propagation,Cuckoo search,Quaternion,Optimization
论文评审过程：Received 3 February 2016, Revised 8 July 2016, Accepted 12 July 2016, Available online 2 August 2016, Version of Record 23 September 2016.
论文官网地址：https://doi.org/10.1016/j.knosys.2016.07.039
