Hybrid Clustering With Quantum Particle Swarm Optimization Initialization for Fuzzy C-Means and DBSCAN

This research delves into the domain of clustering algorithms, with a specific focus on evaluating the performance of Fuzzy C-Means with Quantum-behaved Particle Swarm Optimization initialization and Density-Based Spatial Clustering of Applications with Noise. The overarching goal is to understand and compare the effectiveness of these clustering techniques in the context of a well-known dataset, namely the Digits dataset. The importance of this investigation lies in the need for robust clustering techniques, crucial in various fields such as pattern recognition and data analysis. By probing the efficacy of FCM with QPSO initialization and contrasting it with DBSCAN, we seek to contribute valuable insights to the ongoing discourse surrounding clustering methods. The methodology involves implementing FCM with Q PSO initialization and DBSCAN on the Digits dataset. FCM, known for its soft clustering approach, is enhanced by QPSO initialization to potentially overcome local optima challenges. DBSCAN, a density-based method, is applied for comparison. Silhouette scores and Adjusted Rand Index are employed as metrics to evaluate the clustering performance. Our findings reveal that FCM with QPSO initialization demonstrates superior clustering quality compared to DBSCAN, as evidenced by higher silhouette scores. The visualization of FCM clusters showcases distinct, well-defined groupings of data points, emphasizing the efficacy of the algorithm. Conversely, the DBSCAN clusters exhibit a lower silhouette score, indicative of a less cohesive clustering structure.