Enhanced K-Means Clustering Algorithm using A Heuristic Approach

Abstract

K-means algorithm is one of the most popular clustering algorithms that has been survived for more than 4 decades. Despite its inherent flaw of not knowing the number of clusters in advance, very few methods have been proposed in the literature to overcome it. The paper contains a fast heuristic algorithm for guessing the number of clusters as well as cluster center initialization without actually performing K-means, under the assumption that the clusters are well separated in a certain way. The proposed algorithm is experimented on various synthetic data. The experimental results show the effectiveness of the proposed approach over the existing.