An Extended Membrane System with Active Membranes to Solve Automatic Fuzzy Clustering Problems

Area of Science:

• Computational intelligence
• Data mining
• Machine learning

Background:

• Automatic fuzzy clustering is a challenging problem in data analysis.
• Existing methods often struggle to determine the optimal number of clusters and cluster centers dynamically.

Purpose of the Study:

• To propose a novel automatic fuzzy clustering method.
• To leverage an extended membrane system with active membranes as a computational framework.
• To enhance the determination of cluster numbers and centers.

Main Methods:

• Utilized an extended membrane system with a dynamic and evolving membrane structure.
• Developed a modified differential evolution (DE) mechanism for object evolution rules.
• Implemented object communication and membrane evolution mechanisms for clustering.

Main Results:

• The proposed method effectively determined the optimal number of clusters and corresponding cluster centers.
• Evaluated on 13 benchmark problems, demonstrating superior performance.
• Outperformed four state-of-the-art automatic clustering methods and other comparative techniques.

Conclusions:

• The extended membrane system provides a robust and effective framework for automatic fuzzy clustering.
• The novel method shows significant improvements in effectiveness and robustness compared to existing approaches.