Horizontal collaborative fuzzy clustering based on similarity matrix

Abstract

The collaborative information in horizontal collaborative fuzzy clustering is transmitted by partition matrix, which requires that the dimensions of collaborating partition matrix and collaborated partition matrix must be the same. It requires that the collaborative datasets are clustered into the same number of clusters, but in many cases it is not suitable or difficult to do. In this paper, a new collaborative information transfer mechanism is proposed, which utilizes the similarity matrix to transfer the collaborative information. In this way, when the objective function is designed, the operation of the partition matrix can be transformed into the operation of the similarity matrix, so as to realize the horizontal collaborative fuzzy clustering in the case of different number of clusters. The experimental results show that the proposed algorithm is effective, which extends the application range of horizontal collaborative clustering.