Research In Fuzzy Cluster Technology Based On Genetic Algorithms

Fuzzy c-means cluster algorithm is the most widespread and sensitive in fuzzy cluster analysis. However its vital shortcoming is the sensibility to initial value, it is easy to run into a local optimum. Theoretically, a global optimum can be found by Genetic Algorithm (GA). The main merits of GA are its simplicity, currency, robustness and implicit parallelism. So combining GA with FCM, we will get a hybrid algorithm which has good global and local search capability. It can enhance convergence speed and solve clustering problems.This paper is engaged in the hybrid algorithm of GA and FCM. we will present an improved fuzzy cluster algorithm based on genetic algorithm (GFGA). Firstly, we introduce a kind of cluster-center-based floating point encoding mode; secondly, we use the hybrid select operator of proportional model and elitist model; thirdly, a kind of arithmetic crossover operator based on nearest -distance gene matching is used in the crossover operation; finally, we optimize the population with FCM before it go down to the next generation. Using MATLAB, the experiments show that this improved algorithm's validity and stability are better. This algorithm is appropriate in the use of large data sets and image edge detection.Furthermore, this paper presents a fast fuzzy cluster algorithm based on GFGA (GMRFCM). It is a multistage random sampling genetic-algorithm-based fuzzy c-means clustering algorithm which can significantly reduce the iterative times required by converge, solve the sensitivity to the initialization and obtain a better partition of a data set into c classes. In this way our algorithm has many superior properties and it is especially significant in multi-dimensional spaces and large data sets.