A SEMI-SUPERVISED FUZZY CLUSTERING METHOD FOR DATA PARTITION WITH CONFIDENCE PROBLEM BASED ON PICTURE FUZZY CLUSTERING

Abstract

Data clustering and applications have received much research attention in recent years. During data collection, it is possible that some data with lower confidence (wrong value, incorrect attribute, etc.). This will reduce the clustering performance with possible outliers and noises. Several research directions have been proposed to solve this problem. First, for data elements with wrong values or wrong attributes can use Safe semi-supervised fuzzy clustering methods. Secondly, for noisy data elements, the concept of Picture Fuzzy Set can be used, although there are some related studies to reduce noices and increase the quality of clustering, it is only on the traditional fuzzy set. In this paper, we propose a new algorithm named as PT2FCM, to handle the problem of data partition with confidence problem. The proposed method is implemented and experimentally compared against the related methods, including the standard Picture fuzzy clustering (FCPFS), and the Confidence-weighted safe semi-supervised clustering (CS3FCM), etc. The experimental results show that the proposed method has better performance comparing to selected methods on the same datasets.