Laplacian Eigenmaps based Semi-supervised Metric Fuzzy Clustering algorithm

Abstract

Semi-supervised Metric Fuzzy Clustering (SMUC) is known for taking advantage of prior information of membership to guide clustering. However, SMUC has the following problem: it is easy for SMUC to reduce the effectiveness of priori information of membership guidance because of the sensitivity of algorithm to random noise, which has a negative impact on the performance of SMUC algorithm. In order to solve the problem, we propose a Laplacian Eigenmaps based Semi-supervised Metric Fuzzy Clustering algorithm (LESMUC). Firstly, K nearest neighbors are selected in the data to construct the connected graph; secondly, the weight of the graph is calculated; finally, the objective function is minimized to get the mapping matrix, and the mapping matrix is used to map the data to a new space. This process can reduce the influence of random noise in the data set on the prior information and achieve better clustering effect. Experiments on UCI data and COVID-19 CT images show the effectiveness of the proposed clustering algorithm.