Clustering Enhancement Using Particle Swarm Optimization Algorithm
Boulemnadjel Amel, Bendaoud Amira, Fella Hachouf · 2023
Clustering aims to find useful groups of objects (clusters), based on the closest distance to clusters centers. Soft Subspace Clustering algorithms are generally formulated as an optimization problem that minimizes an objective function, which makes a great improvement in classification. However, each time several iterations to determine the best cluster center are required. In addition, objective function optimization does not always find the global minimum but some local ones. In this paper, We have introduced the algorithm of Particle Swarm Optimization (PSO), which is combined with two subspace clustering algorithms to enhance the clusters centers positions. the proposed classification method has been tested on a synthetic and aerial data set. Two metrics have been used to compare and validate the final results: Normalized Mutual Information (NMI) and classification accuracy. The clustering results obtained are good and improved.