The Bombus-terrestris bee optimization algorithm for feature selection

Meta-heuristic algorithms are one of the well-known methods to solve optimization problems, especially NP-hard problems. These algorithms are mainly developed based on the behavior of the organisms in nature or human behavior. One type of the meta-heuristic algorithm in solving optimization problems is swarm intelligence algorithms which are modeled based on swarm behaviors. In this paper, we focus on the behavior of a sort of bees, bombus-terrestris bee. These bees show several intelligent behaviors, such as finding food, encouraging other cloned bees to find food, learning from other bees to find food, and caring for the queen. Inspired by bombus-terrestris bee behaviors, we introduce an algorithm to solve different kinds of optimization problems, unimodal and multimodal. We mainly focus on solving the feature selection problem based on the binary version of the proposed algorithm. Experimental results show that our proposed method performs better than other meta-heuristic algorithms, such as gray wolf optimization algorithm, Grasshopper optimization algorithm, spotted hyena optimization algorithm, and Harris Hawks Optimizer (HHO), Black Widow Optimization Algorithm (BWO), Artificial bee colony (ABC) algorithm, and Water Strider Algorithm (WSA). We further apply the proposed algorithm on different problems to show the efficiency of the algorithm.