A Novel Feature Selection Method Based on Salp Swarm Algorithm

Biomedical and clinical data usually contain some redundant and irrelevant features, which may lead to misleading and over-fitting problems in the process of modeling algorithms. In order to effectively remove irrelevant or redundant features, the use of feature selection methods can reduce the number of features, improve the accuracy of the model, and reduce the running time. In recent years, Wrapper-based feature selection algorithms have received widespread attention because they can obtain better accuracy. This paper uses a wrapper feature selection algorithm FS_SSA based on Salp swarm. In the FS_SSA algorithm, the position of the follower salps is updated by the relative position of the Salp. The followers gradually moves to the leading Salp. The gradual movement of the follower salps can make the Salp Swarm Algorithm not easy to fall into a local optimal state. The two behaviors of exploration and development in subset searches are balanced, and the search process of feature subsets is prevented from falling into the local optimum. Experimental results based on public medical data sets show that the FS_SSA has better classification performance than other methods.