A Self-Tuning Version for the Possibilistic Fuzzy $c-\text{means}$ Clustering Algorithm
Mirtill-Boglárka Naghi, Levente Kovács, László Sz. Szilágyi · 2023
This paper presents an alternative version of the possibilistic fuzzy$c-\mathbf{means}$(PFCM) algorithm, which can tune automatically its possibilistic penalty terms and uses less parameters as the original PFCM. The proposed method incorporates some cluster size controlling variables into the objective function of PFCM, and with their help it can dynamically modify the penalty terms during the alternating optimization of the objective function. The proposed method is evaluated in comparison with the original PFCM using the IRIS data set and synthetic data. Numerical experiments show that the proposed method can produce fine partitions, and is stable in a wide range of its parameters.