A novel harmony search-based approach for clustering problems
J. Senthilnath, Sushant Kulkarni, D.R. Raghuram, Meghana Sudhindra, S.N. Omkar, Vipul Das, Vinidhra Mani · International Journal of Swarm Intelligence · 2016
Harmony search (HS) is a music-based metaheuristic optimisation algorithm. The harmony in music is analogous to finding the optimality in an optimisation process. In this paper, HS is used to obtain the cluster centres and then classify the data. Three typical benchmark datasets from the UCI machine learning repository and a real time multispectral satellite image are used to demonstrate the results of the technique. The performance of HS is compared with a conventional clustering technique - K-means, and three metaheuristic methods, namely, genetic algorithm (GA), particle swarm optimisation (PSO) and cuckoo search (CS). The results are evaluated using four performance measures, namely, classification error percentage, receiver operating characteristics, time complexity and statistical significance test. From the obtained results, we conclude that HS can be efficiently used to solve data clustering problems.