A Multi-view Fuzzy Compactness and Separation Co-clustering Algorithm

With the advent of the era of big data, although the relationship between samples can be found from individual clustering of each view, data in the real world generally has multiple representations, and different data representations can complement each other. Clustering performance can be made more accurate by mining the information hidden between various multi-view data, making real-word data more interpretable. In view of the above problems, the multi-view clustering is studied, and a multi-view fuzzy compactness and separation co-clustering(MvFCSCC)algorithm is proposed. Based on the fuzzy compactness and separation co-clustering (FCSCC), the algorithm clusters the two dimensions of data objects and features for each view, and then optimizes and assigns different weights to different views by iterating the weight vectors, and finally performs collaborative clustering according to the importance of different views. In order to evaluate clustering effectiveness, experiments were carried out on four multi-view datasets to compare the MvFCSCC with the other three clustering algorithms. The experimental results show that this algorithm not only effectively improves the clustering effect, but also has a stronger robustness to noisy data.