Weight-oriented possibilistic MKFC algorithm

Multi-kernel fuzzy clustering (MKFC) is a hot topic in the field of clustering.However, it is difficult to manually determine the weight of kernel function in the combination, and difficult to allocate the weight of kernel function ideally. Focusing on such problem, a weight-oriented possibilistic MKFC algorithm was put forward, which is based on multi-kernel mechanism, possibilistic clustering and automatic weight calculation.Firstly, the core formula of the objective function and its inner idea were given.Secondly, the calculation methods ofthe Mercer kernel function and multi-kernel distance were given.Then the automatic calculation mode of kernel function weight was given and the flow of the proposed algorithm was formed.Finally, the proposed algorithm is tested in 8 UCI databases, and compared with the 5 algorithms.It is found that the proposed algorithm is an ideal clustering method.