Hybrid Clustering Approach Using Multidimensional Heuristic Optimization

This paper presents a novel hybrid clustering algorithm that integrates the Firefly Swarm Optimization (FSO) method with a multidimensional heuristic optimization framework. The proposed approach leverages the global exploration capabilities of FSO and combines them with the local refinement efficiency of the classical K-means algorithm, thereby overcoming the limitations associated with random centroid initialization and local optima. Notably, the algorithm performs exceptionally well on highdimensional datasets. It is evaluated on several widely used benchmark datasets, demonstrating significant improvements in clustering accuracy, convergence speed, and computational efficiency compared to traditional K-means, Genetic Algorithm (GA)-based clustering, and standalone FSObased methods.