CARO: Chaotic Artificial Rabbit Optimization Algorithm for Feature Selection

Abstract A modified version of Artificial Rabbit Optimization Algorithm (ARO) for feature selection is proposed in this paper. The ARO method is one of the swarm intelligence algorithms that mimic survival strategies of rabbits in nature. This research introduces an enhanced Artificial Rabbit Optimization (ARO) method, termed CARO, which incorporates ten chaotic maps to improve feature selection performance. Utilizing a wrapper-based approach, CARO aims to identify optimal features that maximize classification accuracy. By integrating chaotic maps to modify key elements of the optimization process, the method seeks to accelerate convergence and enhance algorithmic effectiveness. The proposed method's performance was rigorously evaluated using ten diverse UCI datasets and compared against three prominent meta-heuristic techniques: Ant Colony Optimization (ACO), Whale Optimization Algorithm (WOA), and Particle Swarm Optimization (PSO). Experimental results demonstrate that CARO consistently outperform the comparative meta-heuristic methods in feature selection. Besides, the proposed method is evaluated using 12 functions from CEC’2022 benchmarks where it outperformed the existing methods.