Current Proceedings on Technology
Yazarlar: Wei Wu, Huanan Wang
Konular:-
Anahtar Kelimeler:Genetic algorithm,Improved exploratory K-Means,Global optimization,Gene clustering,Bioinformatics
Özet: Gene clustering is useful for discovering the function of gene since co-expressed genes are likely to share the same biological function. K-means is one of the well-known clustering methods. However, it is sensitive to the initial cluster formation and is easily trapped in local optimal solutions. In this paper, a novel hybird genetic algorithm named as genetic IXK-Means algorithm (GXKA) is proposed, which handles gene clustering problem and finds a globally optimal partition of gene expression data into a specified number of clusters. We define an improve exploratory K-Means (IXK-Means) operator in which empty clusters are filled, and use it in GXKA as a search operator instead of crossover. Our results indicate that GXKA is superior to K-Means, XK-Means and IXK-Means in terms of error, homogeneity and separation. In addition, GXKA is faster than Genetic K-Means algorithm (GKA) to achieve the global optimum.