K-Means Clustering Technique Based On Swarm Intelligence Algorithms

In recent years,with the fast development of the computer information technology,the database and its applications are continuously expanding.the information acquisition technolo-gy is also being updated constantly.Nowadays,the world is in the age of information explosion.Therefore,it has become the difficulty key issue that how to effectively obtain the required in-formation and data from the mass of data sets.Clustering analysis,as an unsupervised machine learning method,is an important research field in pattern recognition and data mining.And it has been widely used in statistical analysis?medical hygiene?biological information pro-cessing?image processing?social science and other fields.Clustering analysis groups the data according to similar characteristics such as similar structure or similar expression,so that the data in the same clusters have the greatest similarity,while the data in different clusters have the greatest dissimilarity.In this paper,some shortcomings of some clustering algorithms are found and corresponding improvement strategies are proposed.In Chapter 1,we review the relevant background knowledge of clustering analysis,and we describe the research purpose and the main content of this paper.In Chapter 2,an EXK-Means(empty-cluster-reassignment technique)is proposed for re-covering the empty clusters that appear during the iteration of XK-Means(eXploratory K-Means).XK-Means has been introduced in the literature by adding an exploratory disturbance onto the vector of cluster centers,so as to jump out of the local optimum and reduce the sensitivity to the initial centers.However,empty clusters may appear during the iteration of XK-Means,caus-ing damage to the efficiency of the algorithm.Furthermore,we combine the EXK-Means with genetic mechanism to form a genetic XK-Means algorithm with empty-cluster-reassignment,re-ferred to as GEXK-Means clustering algorithm.The convergence of GEXK-Means to the global optimum is theoretically proved.In Chapter 3,an improved,immune clone K-Means algorithm is proposed.In which the immune genetic algorithm combined with clonal selection algorithm.In the proposed algorithm,the concept of the immune vaccine of the immune algorithm is introduced in the basic clonal selection algorithm.This algorithm remedies the traditional cloning algorithm's insufficient,and enhances the diversity of antibodies,and improves the global search ability of the K-Means algorithm.In Chapter 4,an improved particle swarm optimization clustering algorithm is proposed,in which the clonal selection algorithm is combined with particle swarm optimization algorithm.It remedies the shortcomings of traditional cloning algorithm and particle swarm optimization algorithm,enhances the diversity of particles and improves the global optimization ability of particle swarm K-Means algorithm.In Chapter 5,the empty-cluster-reassignment technique is introduced into the particle swar-m optimization K-Means algorithm,so as to remedy the similar empty cluster phenomenon as described in Chapter 2.Then we combined the clone algorithm with the particle swarm opti-mization K-Means,referred to as ECPSOKM clustering algorithm.This enhances the global search ability of the particle swarm optimization K-Means algorithm.