MKF-Firefly: Hybridization of Firefly and Multiple Kernel-Based Fuzzy C-Means Algorithm

Abstract: Clustering find various application in the variety fields like, telecommunication, medical image processing, bioinformatics and so on. This high demand poses a challenge to the researchers to develop an effective and efficient clustering algorithm for grouping the data objects. Accordingly, literature presents various algorithms for data clustering using partitional-based approaches. This paper presents a new clustering algorithm, namely, MKF-Firefly which is developed by combining the multiple kernel-based objective function and firefly algorithm. The firefly algorithm finds the optimal cluster centroids using the multiple kernel-based objective function. The centroids obtained from the firefly algorithm are then utilized for clustering process. The proposed clustering process is evaluated using Rand coefficient, Jaccard coefficient and Clustering Accuracy on the two different datasets like iris and wine. The proposed MKF-firefly achieved the clustering accuracy of 97% on the iris dataset.

Keywords: Clustering, Firefly, optimization, Multiple Kernel-Based clustering, Rand coefficient.