F-FDRPSO: A Novel Approach Based on Hybridization of Fuzzy C-means and FDRPSO for Gene Clustering

Abstract

Gene Clustering is one among the most popular issues involve in the field of Bioinformatics and is defined as the process of grouping related genes in the same cluster. Among the various algorithm proposed for clustering, the fuzzy c-means and there hybridization with some other methods has been used by most of the researchers to deal with the problem of premature convergence in fuzzy c-means clustering algorithm, but the results obtained were not satisfactory because the gene expression has huge amounts of ambiguous and uncertain biological data which requires advanced computing tools for processing such data. Particle Swarm Optimization (PSO) one of the variant of Swarm Intelligence (SI) has recently emerged as a nature inspired algorithms, especially known for their ability to produce low cost, fast and reasonably accurate solutions to complex search problems. PSO based Fuzzy C-Means algorithm were proposed but they all uses the traditional PSO algorithm. In traditional PSO algorithm each particle is attracted toward the best ever position discovered by any particle in the swarm, that limits the exploration capability. Instead if particle learn from the experience of the neighbouring that has better fitness than itself, the swarm can be more effectively and efficiently explored. So a method based on hybridization of fuzzy c-means and Fitness Distance Ratio based PSO is proposed. Initially this approach distributes the membership on the basis of the distance between sample and cluster centre making membership meet the constraints of FCM then the ratio of relative fitness and the distance of other particle is used to determine the direction in which each component of the particle position needs to be changed. The experiments were conducted on four real data sets and results shows that F-FDRPSO performs significantly better than FPSO and FCM algorithm.