A Faster Genetic Clustering Algorithm

This paper presents a novel genetic clustering algorithm combining a genetic algorithm (GA) with the classical hard c-means clustering algorithm (HCMCA). It processes partition matrices rather than sets of center points and thus provides a new implementation scheme for the genetic operator - recombination. For comparison of performance with other existing clustering algorithms, a gray-level image quantization problem is considered. Experimental results show that the proposed algorithm converges more quickly to the global optimum and thus provides a better way out of the dilemma in which the traditional clustering algorithms are easily trapped in local optima and the genetic approach is time consuming.