Hybridization Harris Hawk Optimization and β-Hill Climbing for Voxel Selection in Brain Decoding

Decoding visual stimuli is a growingly significant neuroscience task. The objective is to categorize the brain's activity during the perceptron of visual items. The outputs are the names of the stimuli, while the inputs are functional magnetic resonance imaging (fMRI) data. Selecting informative features (voxels) is one of the key challenges facing in the brain decoder. We propose the hybrid Harris hawk voxel selection (HHHVS) algorithm to precisely decode the brain activity induced by visual stimuli. This inspired us to develop a better version of Harris hawk optimization algorithm that includes (β-hill climbing to maximize the exploration procedure. An effective final voxel is evaluated using support vector machine classifier and six (6) well-known metaheuristic algorithms. The obtained outcomes are beyond the other metaheuristic algorithms in terms of number of selected voxels, classification accuracy, standard deviation, and convergence speed. The experimental findings of the proposed technique are highly encouraging and can be employed successfully in the brain computer interface system.