A K-means Optimized Clustering Algorithm Based on Improved Genetic Algorithm

The K-means algorithm is highly sensitive to the initial clustering centers and easily get trapped in a local optimum. To avoid such problems, this paper proposes an improved crossover operator of chromosomes in the genetic algorithm, redefines the calculation method of genetic probability and the natural selection rules, introduces different individual selection mechanisms for the two adjacent generations of chromosomes, and integrates the K-means algorithm into the improved genetic algorithm. Experiment results demonstrate that the improved K-Means algorithm is better than the original genetic algorithm and K-Means algorithm in clustering performance, far better than the bisecting K-means algorithm.