Feature Extraction Using Levy Distribution-Based Salp Swarm Algorithm

Feature selection is a major pre-processing stage in machine learning to identify the optimal features through diminishing the redundant features to enhance the classification accuracy. The swarm intelligence algorithm provides superior optimal solution through distributive search over complete search in the feature space. Nevertheless, the SSA algorithm falls with local optima and poor convergence rate. In this paper, the levy distribution-based Salp Swarm Algorithm (LDSSA) framework is proposed to strengthen the balance among explorative and exploitative samples. The goal of this work is to minimize the higher dimensionality for larger datasets by feature extraction. The experimental results were examined on 10 Benchmark datasets from UCI repository to legitimize the supremacy of proposed algorithm with other existing methodologies. The comparative outcome shows the distribution improves the performance of SSA method and surpasses over other optimization algorithm.