Improvement of the Genetic Algorithm and its Application on Clustering

This paper proposes an improved genetic algorithm, it keeps the population diversity by similarity checks on the population before selection, and the algorithm solves the early-maturing problem of the population evolution, and proposes a formula for mutation probability related with similarity rate and iteration times. The algorithm not only maintains a good diversity of population, but also guarantees the algorithm convergence. Compared to c-means clustering algorithm, the improved genetic algorithm proposed in this paper has been proved its improvement effect by the result of clustering experiments using the UCI datasets of WINE and IRIS.