International Journal of Emerging Trends & Technology in Computer Science
A Motivation for Recent Innovation & Research
ISSN 2278-6856
www.ijettcs.org

Call for Paper, Published Articles, Indexing Infromation Improved Membership Function for Multiclass Clustering with Fuzzy Rule Based Clustering Approach, Authors : Archana N. Mahajan, Prof. Dr. Girish Kumar Patanaik and SandipS. Patil, International Journal of Emerging Trends & Technology in Computer Science (IJETTCS), www.ijettcs.org
Volume & Issue no: Volume 3, Issue 5, September - October 2014

Title:
Improved Membership Function for Multiclass Clustering with Fuzzy Rule Based Clustering Approach
Author Name:
Archana N. Mahajan, Prof. Dr. Girish Kumar Patanaik and SandipS. Patil
Abstract:
Abstract Fuzzy clustering is the combination of clustering and fuzzy set theory. It is useful to handle the problem of determining the vague boundaries of clusters. Fuzzy clustering is better than Crisp clustering when the boundaries between the clusters are vague and ambiguous. In both fuzzy and crisp clustering algorithms there is need and requirement to know the number of potential clusters and/or their initial positions in advance. The existing system identifies the potential clusters in given dataset by itself. It uses the fuzzy rules for identifying the potential clusters. When the multiclass dataset is given as input to the existing system, the number of clusters discovered in the multiclass dataset are less than the classes in it because it forms the two class problem of the given data and applies it as input to genetic algorithm for fuzzy rule generation. The genetic algorithm generates the more general rules. Since each class represent some aspects of that particular class and it can be used to generate the taxonomy, it becomes essential to find optimized clusters. The proposed system identifies the number of clusters which are equal to the number of implicit classes in multi class data. There are three main phases in the proposed system. First, it preprocesses the multi class data. In the second phase, it generates the fuzzy rules with subtractive clustering. The best membership function for multi class data is searched in third phase with the minimum Euclidean norm. The proposed system, then finds the number of clusters in the given dataset with the best membership function and the fuzzy rules generated. The number of clusters is also optimized using the adaptive network based fuzzy inference system. These clusters are discovered without prior knowledge about the number of classes available in the given data. The proposed system can be used to generate the class labels and identify the sub-homogeneous patterns in the given data. Keywords: Fuzzy Clustering, membership function, Fuzzy Rules, Fuzzy Rule Based Clustering
Cite this article:
Archana N. Mahajan, Prof. Dr. Girish Kumar Patanaik and SandipS. Patil , " Improved Membership Function for Multiclass Clustering with Fuzzy Rule Based Clustering Approach " , International Journal of Emerging Trends & Technology in Computer Science (IJETTCS), Volume 3, Issue 5, September - October 2014 , pp. 061-068 , ISSN 2278-6856.
Full Text [PDF]                          Home