EEG channel selection in CIT: a binary Harris Hawks optimization using a time-varying transfer function

Abstract

Channel selection is difficult for electroencephalographic (EEG) channel data. The test for confidential information is performed using a 16-channel electrode and recorded EEG data. The system’s performance suffers when optimal channel selection data are not available. Removing noise and irrelevant channels improves the outcomes and implementation of the EEG deception identification system. Swarm optimization techniques are frequently used to overcome feature selection problems. Transfer functions need swarm algorithms because they change the search space from one that is continuous to one that is discrete. The currently used sigmoid transfer function cannot balance exploration and exploitation stages in the search space, including optimal local problems. The novel binary Harris Hawks optimization with a time-varying transfer function (TVk-BHHO) is used in this work to select channels from EEG channel data optimally. Controlling the time-varying parameter is essential for balancing the search space’s exploration and exploitation stages and avoiding the optimum local problem. In this case, a specific classifier from a random forest improves the system’s performance and accuracy. The results show that TVk-BHHO has consistent channel selection, resulting in higher classification accuracy and more harmonious convergence. Existing meta-heuristic algorithms are against the simulation results of the new TVk-BHHO approach. The best accuracy achieved here is 98.50%.