An Efficient Weighted Fuzzy Possibilistic C-Means Clustering Algorithm
Shuang Xie, Haiyan Yu · 2023
The fuzzy possibilistic c-means clustering (FPCM) algorithm is a hybrid clustering algorithm by incorporating fuzzy memberships and possibilistic memberships into the objective function, which can improve the anti-noise ability of the fuzzy c-means clustering (FCM) and the coincident clustering phenomenon of the possibilistic c-means clustering (PCM) algorithm to some extent. However, the FPCM algorithm often obtains poor results for datasets injected with strong long-distance noise, resulting from that these noise points are assigned with too large values for fuzzy memberships. Therefore, this paper introduces a weight parameter into the objective function of the FPCM and proposes an efficient weighted fuzzy possibilistic c-means clustering (EWFPCM) algorithm. The experimental results on several synthetic datasets and color image segmentation show that the proposed EWFPCM algorithm can greatly improve the anti-noise ability of the FPCM, and performs best among compared several clustering algorithms.