An improved elastic net clustering algorithm with dynamic parameter strategy.

Area of Science:

• Data Science
• Machine Learning
• Artificial Intelligence

Background:

• Clustering is crucial for data mining, but challenges persist in achieving high-quality solutions for diverse and large-scale datasets.
• High computational complexity often hinders effective data mining and knowledge discovery.

Purpose of the Study:

• To propose an improved elastic net clustering algorithm (IENDP) addressing limitations in current clustering methods.
• To enhance the ability to discover structures and knowledge from complex datasets.

Main Methods:

• Developed a novel energy function to better distinguish data point distributions within clusters.
• Integrated a dynamic parameter strategy to improve search ability and convergence speed.
• The IENDP algorithm is self-organizing and self-learning, requiring no manual guidance.

Main Results:

• The IENDP algorithm achieves higher clustering quality, effectively identifying clusters of varying sizes, shapes, and densities.
• Demonstrated superior performance compared to classical and state-of-the-art clustering algorithms.
• Showed low computational and time complexity on synthetic and real-world datasets.

Conclusions:

• The proposed IENDP algorithm offers an effective solution for high-quality clustering, particularly for high-dimensional and large-scale data.
• The dynamic parameter strategy significantly improves clustering performance and reduces parameter sensitivity.
• IENDP provides a robust and efficient method for data mining and knowledge discovery.