Optimization of Density Peak Clustering Algorithm Based on Improved Black Widow Algorithm.

Area of Science:

• Data Science
• Machine Learning
• Artificial Intelligence

Background:

• Clustering is a key unsupervised learning technique.
• Density Peak Clustering (DPC) identifies arbitrary-shaped clusters but requires manual parameter tuning.
• The cutoff distance (d) parameter in DPC is critical and often necessitates readjustment for different datasets.

Purpose of the Study:

• To address the limitations of manual parameter setting in DPC.
• To develop an automated method for optimizing the cutoff distance (d) in DPC.
• To improve the accuracy and efficiency of density-based clustering algorithms.

Main Methods:

• Integration of the Black Widow Optimization Algorithm (BWOA) with Density Peak Clustering (DPC).
• Development of the Black Widow Density Peaks Clustering (BWDPC) algorithm for automatic cutoff distance (d) optimization.
• Comparative analysis of BWDPC against other clustering algorithms using synthetic and UCI datasets.

Main Results:

• BWDPC effectively automates the determination of the cutoff distance (d).
• The proposed BWDPC algorithm demonstrates higher accuracy in identifying cluster centers (density peak points).
• BWDPC achieves superior overall clustering performance compared to traditional DPC and other methods.

Conclusions:

• BWDPC offers a significant improvement over the standard DPC algorithm.
• The automated parameter optimization enhances the robustness and applicability of density-based clustering.
• BWDPC provides a more accurate and efficient solution for clustering diverse datasets.