APPROACH-AVOIDANCE OPTIMIZATION ALGORITHM: A NOVEL METAPHOR-FREE OPTIMIZER WITH APPLICATIONS IN DATA CLUSTERING

This paper proposes a novel metaphor-free optimization algorithm called Approach-Avoidance Optimization Algorithm (AAOA) to solve real-world optimization problems. AAOA is built on the principle that candidate solutions should move towards better solutions and away from worse ones. The algorithm utilizes an effective collective search strategy by enabling randomized variable-level information exchange among candidate solutions, thereby facilitating a comprehensive and efficient exploration of the search space. Due to its simplicity, ease of implementation, and parameter-free nature, AAOA is particularly suitable for users without deep expertise in optimization. The performance of AAOA was evaluated through extensive experiments on the CEC 2017 benchmark functions across various dimensions, as well as on 30 real-world data clustering problems. Its performance was compared against eight metaphor-free algorithms using different evaluation criteria. Experimental results demonstrated that AAOA achieved competitive or superior performance in most cases, highlighting its potential as a robust and practical tool for high-dimensional and complex optimization tasks. This study contributes to the growing body of research on simple yet effective optimization algorithms that do not rely on metaphorical analogies.