Multistage random sampling genetic-algorithm-based fuzzy c-means clustering algorithm

This work presents a multistage random sampling genetic-algorithm-based fuzzy c-means clustering algorithm (called GMRFCM), which can significantly reduce the iterative times required to converge, the sensitivity to the initialization, and can obtain a better partition of a data set into c classes. At the first of this algorithm, it uses the multistage random sampling fuzzy c-means clustering algorithm (MRFCM) to produce an initial population; then applies this population on the improved fuzzy genetic cluster algorithm (GFGA) to perform genetic operations. In this way, the proposed algorithm in this paper can have strong global and local searching capability and it is especially significant for high-dimensional and large data sets. Experiments are given in the last of this paper. It is observed that the proposed algorithm in this paper searches better than MRFCM in the iterative times required to converge and the final objective function value.