AIEOU: Automata-based improved equilibrium optimizer with U-shaped transfer function for feature selection

High dimensional data has become an essential challenge to address in Data Science and Machine Learning. Reducing the number of dimensions by excluding noisy, irrelevant, or correlated dimensions is often referred to as the feature selection (FS). The ultimate goal in FS is to identify an optimal set of dimensions (features) to realize better classification accuracy, decrease the computational time and optimize the memory requirements with the help of some methods. Recently, optimization algorithms have gained popularity in different fields because of their flexibility and effectiveness. Equilibrium optimizer (EO) is a physics-based meta-heuristic algorithm, which is inspired from a well-mixed dynamic mass balance on a control volume that has good exploration and exploitation capabilities. In this work, an improved version of EO is proposed with the inclusion of learning based automata to find proper values of its parameters and Adaptive β Hill Climbing (AβHC) to find a better equilibrium pool. The method is used as a feature selector, evaluated on 18 standard UCI datasets with the help of K-nearest neighbors (KNN) classifier, and compared with eight state-of-the-art methods including classical and hybrid meta-heuristic algorithms. Moreover, the proposed methods is applied on high dimensional Microarray datasets which generally contain a few samples but large number of features, and often lead to ‘curse of dimensionality’. The obtained results illustrate the supremacy of the proposed method over the other state-of-the-art methods mentioned in literature. The source code of this work is publicly available at https://github.com/ahmed-shameem/Feature_selection.