Two-stage improved Grey Wolf optimization algorithm for feature selection on high-dimensional classification

Abstract In recent years, evolutionary algorithms have shown great advantages in the field of feature selection because their simplicity and potential global search capability. However, most existing based on computation are wrapper methods, which computationally expensive, especially for high-dimensional biomedical data. To significantly reduce computational cost, it is essential to study an effective evaluation method. this paper, a two-stage improved gray wolf optimization (IGWO) algorithm data proposed. first stage, multilayer perceptron (MLP) network with group lasso regularization terms trained construct integer problem using proposed pre-selection features hidden layer structure. The dataset compressed subset obtained stage. second retrained dataset, employed discrete selection. Meanwhile, rapid strategy constructed mitigate cost improve efficiency process. effectiveness was analyzed ten gene expression datasets. experimental results show that not only removes almost more than 95.7% all datasets, but also has better classification accuracy test set. addition, time consumption, size become prominent as dimensionality increases. This indicates particularly suitable solving problems.