Artificial Bee Colony Algorithm with Adaptive Explorations and Exploitations: A Novel Approach for Continuous Optimization

Abstract

A proper balance between global explorations and local exploitations is often considered necessary for complex, high dimensional optimization problems to avoid local optima and to find a good near optimum solution with sufficient convergence speed. This paper introduces Artificial Bee Colony algorithm with Adaptive eXplorations and eXploitations (ABC-AX), a novel algorithm that improves over the basic Artificial Bee Colony (ABC) algorithm. ABC-AX augments each candidate solution with three control parameters that control the perturbation rate, magnitude of perturbations and proportion of explorative and exploitative perturbations. Together, all the control parameters try to adapt the degree of global explorations and local exploitations around each candidate solution by affecting how new trial solutions are produced from the existing ones. The control parameters are automatically adapted at the individual solution level, separately for each candidate solution. ABC-AX is tested on a number of benchmark problems of continuous optimization and compared with the basic ABC algorithm and several other recent variants of ABC algorithm. Results show that the performance of ABC-AX is often better than most other algorithms in comparison, in terms of both convergence speed and final solution quality.