Classification of intelligence quotient using EEG sub-band power ratio and ANN during mental task

It has been a long debate on conventional psychometric test as benchmark of individual´s intelligence quotient (IQ). However, recent studies in a variety of neurophysiological researches have been done to link intelligence with individual´s brainwave pattern. Hence this paper proposes an intelligent approach to classify IQ via brainwave sub-band power ratio and artificial neural network (ANN). Fifty samples of electroencephalogram (EEG) dataset have been collected during IQ test session. Three IQ levels have been categorized based on the IQ scores from Raven´s Progressive Matrices as the control group. Left hemispheric brainwave focusing on theta, alpha and beta sub-bands are the key discussion of this paper. The features are used as input to train the ANN. Formerly, synthetic data have also been generated with white Gaussian noise to increase the performance of the classifier. Subsequently, the network model have been developed using an ANN that is trained with optimized parameters which are learning rate, momentum constant and hidden neurons. The network model trained with back-propagation algorithm has yielded low mean squared error (MSE). Findings also indicate that the distinct intelligence quotient levels can be classified with 97.62% training and 94.44% testing accuracies via brainwave sub-band power ratio.