Boosting capuchin search with stochastic learning strategy for feature selection

Abstract The technological revolution has made available a large amount of data with many irrelevant and noisy features that alter the analysis process increase time processing. Therefore, feature selection (FS) approaches are used to select smallest subset relevant features. Feature is viewed as an optimization for which meta-heuristics have been successfully applied. Thus, in this paper, new approach proposed based on enhanced version Capuchin search algorithm (CapSA). In developed FS approach, named ECapSA, three modifications introduced avoid lack diversity, premature convergence basic CapSA: (1) inertia weight adjusted using logistic map, (2) sine cosine acceleration coefficients added improve convergence, (3) stochastic learning strategy add more diversity movement levy random walk. To demonstrate performance different datasets used, it compared other well-known methods. results provide evidence superiority ECapSA among tested competitive methods terms metrics.