The Research On Fuzzy Clustering And Fuzzy Cluster Validation

Fuzzy clustering has been widely studied and applied in a variety of substantive areas.It becomes one of the important research topics,and has received the attention of scholars and experts.Fuzzy C-means(FCM)algorithm is one of the most popular fuzzy clustering techniques because it is efficient,straightforward,and easy to implement.However,FCM is sensitive to initialization,easily trapped in local optima,and requires the user to predefine the number of clusters.For the above mentioned problems,the thesis combination with particle swarm optimization and hierarchical clustering analysis technology,proposes some new methods and results.Main topics include:1.In the aspect of fuzzy clustering,a hybrid fuzzy clustering method based on FCM and fuzzy adaptive particle swarm optimization and the improved fuzzy C-means algorithm based on initial center optimization method are proposed.The hybrid fuzzy clustering method based on FCM and fuzzy adaptive particle swarm optimization makes use of the merits of both algorithms,overcome the shortcomings of fuzzy C-means which sensitive to initial value and usually leads to local minimum.In order to verify the validity and superiority of the algorithm,the algorithm is applied to the six classic data sets and the experimental results show that the proposed method is efficient and can reveal encouraging results.The improved fuzzy C-means algorithm based on initial center optimization method combines the density-based and grid-based method with fuzzy C-means algorithm.The proposed algorithm avoids the sensitivity of FCM to initial centers and can tolerate noise.The performance and effectiveness of the proposed clustering algorithm is evaluated by 4 San Francisco taxi GPS cab mobility traces data sets,and the experimental results show that the proposed algorithm has better clustering results.2.In the aspect of cluster validation,a cluster validity index based on fuzzy hybrid hierarchical clustering is proposed.The proposed cluster validity only needs to perform one time clustering algorithm,and determine the optimal number of clusters according to the clustering results.The proposed index overcomes the shortcoming of traditional clustering validity indices which need to run all possible values.Testing of the proposed index and four previously formulated indices on well-known data sets shows the superior effectiveness and reliability of the proposed index in comparison with other indices.