Grey wolf optimizer with self-repulsion strategy for feature selection.

Area of Science:

• Artificial Intelligence
• Machine Learning
• Data Science

Background:

• Feature selection is crucial for big data analysis accuracy.
• Traditional grey wolf optimizer (GWO) algorithms face challenges like slow convergence and local optima in high-dimensional tasks.

Purpose of the Study:

• To introduce a novel feature selection algorithm, grey wolf optimizer with self-repulsion strategy (GWO-SRS).
• To enhance the performance of GWO by addressing its limitations in convergence speed and exploration capability.

Main Methods:

• Flattening the hierarchical structure of the wolf pack for faster command transmission.
• Implementing a self-repulsion learning strategy for the alpha wolf.
• Utilizing a pack learning strategy based on alpha wolf predatory behavior for enhanced self-learning.

Main Results:

• GWO-SRS demonstrated accelerated convergence compared to traditional GWO.
• Experimental analysis on the UCI dataset showed a 15% reduction in average classification error.
• The algorithm achieved this improvement while using 20% fewer features.

Conclusions:

• GWO-SRS effectively overcomes the limitations of traditional GWO, including premature convergence and limited exploration.
• The proposed algorithm offers a robust solution for complex feature selection problems in big data analysis.
• This work underscores the importance of refining optimization algorithms for advanced data processing tasks.