Opposition-based learning techniques in metaheuristics: classification, comparison, and convergence analysis

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Metaheuristic Computing

Background:

• Opposition-based learning (OBL) is a key strategy for improving metaheuristic algorithms (MAs).
• A structured analysis of OBL variants' impact on MA performance is lacking.
• Existing MAs often suffer from slow convergence, limited exploration, and exploration-exploitation imbalance.

Purpose of the Study:

• To categorize and analyze nine distinct OBL techniques.
• To systematically assess the effectiveness of five OBL variants integrated into five MAs.
• To demonstrate OBL's capability to enhance MAs facing common optimization challenges.

Main Methods:

• Implemented five OBL variants within differential evolution, genetic algorithm, particle swarm optimization, artificial bee colony, and harmony search.
• Evaluated hybridized algorithms across initialization and generation update phases.
• Tested algorithms on 12 benchmark functions from the CEC2022 suite, analyzing key performance metrics and using a Friedman test for statistical validation.

Main Results:

• Quasi-reflection opposition-based learning consistently outperformed other OBL variants.
• The enhanced MAs showed improved convergence speed and solution quality across most benchmark functions.
• Statistical validation confirmed significant performance differences among OBL-enhanced MA variants.

Conclusions:

• Opposition-based learning significantly enhances metaheuristic algorithm performance.
• Quasi-reflection OBL is a highly effective variant for improving convergence and solution quality.
• This study provides a structured framework for understanding and applying OBL in metaheuristic optimization.