Open Access
Subscription Access
Message Passing in Clusters using Fuzzy Density based Clustering
The Communication between objects in the cluster takes place through message passing techniques. Affinity Propagation (AP) clustering has been used successfully in a lot of clustering problems, which deals with static data. By applying Fuzzy DENCLUE, similarity between objects and its exemplar is improved. A dynamic variant of AP clustering called Incremental Affinity Propagation with K-Medoid (IAPKM) is used along with the Fuzzy Density based clustering (DENCLUE) method. In Fuzzy DENCLUE, number of clusters is reduced and randomness in the form of noise is removed. The experiment is performed in Net Beans 8.0.2 using jdk1.7 as a language. The evaluation result defines that the effectiveness and efficiency of IAPKM and Fuzzy DENCLUE can achieve comparable performance. Compared with IAPKM and Fuzzy DENCLUE, Sum of similarity and accuracy is increased in Fuzzy DENCLUE which is based on the parameters. The current approach achieves higher accuracy rate which is 10% greater than the former approach.
Keywords
Exemplar, Fuzzy DENCLUE, Incremental Affinity Propagation based on K-Medoid, Message Passing, Sum of Similarities
User
Information
Abstract Views: 271
PDF Views: 0