Quantum-Inspired Moth-Flame Optimizer With Enhanced Local Search Strategy for Cluster Analysis.

Area of Science:

• Data Mining
• Machine Learning
• Computational Intelligence

Background:

• Clustering is a key unsupervised learning technique in data mining.
• K-means is popular but suffers from sensitivity to initializations and local optima.
• Existing swarm intelligence algorithms have limitations in clustering efficiency.

Purpose of the Study:

• To propose a quantum-inspired moth-flame optimizer with an enhanced local search strategy (QLSMFO) to address K-means limitations.
• To improve population diversity, exploration, and exploitation capabilities in clustering algorithms.
• To enhance the precision, convergence speed, and stability of clustering.

Main Methods:

• Quantum double-chain encoding and quantum revolving gates for initial population enrichment.
• An enhanced local search strategy based on the Shuffled Frog Leaping Algorithm (SFLA) for improved exploitation.
• Levy flight for updating poor solutions to accelerate convergence.

Main Results:

• QLSMFO demonstrated superior performance on ten UCI benchmark clustering datasets.
• Comparative analysis showed QLSMFO outperformed K-means and ten other swarm intelligence algorithms.
• Statistical tests (Wilcoxon rank-sum and Friedman) confirmed QLSMFO's significant effectiveness.

Conclusions:

• QLSMFO offers significant improvements in clustering precision, convergence speed, and stability.
• The proposed quantum-inspired approach effectively overcomes the limitations of traditional K-means.
• QLSMFO represents a promising advancement in unsupervised learning for data mining applications.