An Enhanced Binary Cuckoo Search Algorithm Using Crossover Operators for Features Selection

One of the most important preprocessing steps is the determination of the most relevant subset of features in any dataset. This step is called “Feature Selection”, which is considered an NP-Hard optimization problem. Cuckoo Search Algorithm (CSA) is a popular Nature-Inspired Meta-heuristic algorithm, which is used for handling continuous and discrete optimization problems. Although CSA has a great performance on many optimization problems, however, it lacks the balancing between the exploration and exploitation abilities. In this research, a binary cuckoo search algorithm with different types of crossover operators to improve is proposed. The crossover operators help the nest to discover different unexplored regions and avoid the issue of trapping in the local optima. All of the proposed versions have been applied on several datasets, and the results in terms of the classification accuracy and number of selected features were recorded. The experimental results demonstrated the effectiveness of the proposed algorithms in comparison to the original binary cuckoo search algorithm and different classifications.