Fast channel selection method using crow search algorithm

Abstract

In Brain Computer Interface (BCI), the brain signals are used to perform some commands or actions in a computer. Brain signals are recorded using many methods. Electroencephalogram (EEG) is one of the non-invasive methods. EEG signals are recorded using multiple channels. Selection methods are used to choose the most relevant and powerful signals. Usually Meta-heuristic algorithms are used for selection. Meta-heuristic algorithms depends on random generated population of solutions for the objective function. Because of the randomness, there is always a chance to select zero as a solution. Zero in EEG channels selection means no channel is chosen to extract its signal features. This situation is not practical, the selection process should be repeated whenever a zero solution appears. The repetition will reduce the algorithm speed. This paper introduces a fast channel selection algorithm using Crow Search Algorithm (CSA). CSA is used to select the best channels offline. Using no-zero channel condition to fasten the algorithm. Our results show that CSA with no-zero channels condition is better than Genetic algorithm (GA). Although CSA and GA results are almost have the same accuracy, but CSA with no-zero condition is faster.