A Delayed Spiking Neural Membrane System for Adaptive Nearest Neighbor-Based Density Peak Clustering

Area of Science:

• Computer Science
• Artificial Intelligence
• Computational Biology

Background:

• Density Peak Clustering (DPC) struggles with adaptability and local data structure.
• Traditional clustering algorithms often exhibit high time complexity.
• P systems, particularly spiking neural P systems (SN P systems), offer potential for reduced complexity and enhanced parallelism.

Purpose of the Study:

• To enhance the Density Peak Clustering (DPC) algorithm's adaptability and performance.
• To address the limitations of existing clustering algorithms by incorporating P systems.
• To propose a novel algorithm, DSNP-ANDPC, for improved clustering.

Main Methods:

• Improved DPC by integrating maximum nearest neighbor distance and K-nearest neighbors (KNN).
• Developed a novel approach using delayed spiking neural P systems (DSN P systems).
• Proposed the DSNP-ANDPC algorithm, combining DPC enhancements with DSN P systems.

Main Results:

• The DSNP-ANDPC algorithm demonstrated superior performance compared to other methods.
• Evaluations were conducted on four synthetic and 10 real-world datasets.
• The proposed method showed significant improvements in clustering effectiveness.

Conclusions:

• DSNP-ANDPC effectively addresses DPC's limitations in adaptability and local structure awareness.
• The integration of DSN P systems offers a promising direction for efficient and effective clustering.
• The proposed algorithm represents a significant advancement in clustering techniques.