A Hybrid Particle Swarm Optimization K-Means Algorithm with Quantum Behavior

The K-means algorithm is a partition-based clustering approach with two notable drawbacks, sensitivity to the initial center and a tendency to converge to suboptimal solutions. To address these limitations, a hybrid particle swarm optimization k-means algorithm with quantum be-havior (HQPSO-K) is proposed. The improved algorithm uses a nonlinear contraction-expansion coefficient to balance local optimization and global search. Moreover, Levy flight strategy is incorporated to further improve the particles' capability to evade local optima. Finally, the improved QPSO algorithm combined with K-means explores the search space more efficiently. The algorithm is tested with UCI data set. Experimental results demonstrate that HQPSO-K achieves superior clustering outcomes and is less sensitive to initial conditions.