Elastic Network Algorithm for Clustering Based on Cluster Center Shift
Junyan Yi, Du Xiaopeng · 2020
The clustering problem is a hot issue in recent years. Mean-Shift is an effective method for the clustering problem without the prior knowledge of the number of clusters. In this paper, Mean-Shift and Elastic Network Algorithm (ENA) are combined. Further, an Elastic Network Algorithm for Clustering based on cluster center shift (ENACS) for cluster analysis is proposed, which can be used for the optimization of cluster stability and cluster effectiveness. In the experiments, we show the effectiveness of ENACS. In comparison with traditional algorithms, the ENACS, which didn't need to give the number of clusters, could consistently and effectively converge to the approximate optimal value corresponding to the given data. Experiments used several of the standard datasets from UCI machine learning repository.