EEG-based IQ and learning style classification model using artificial neural network / Muhammad Marwan Anoor

Muhammad Marwan , Anoor (2021) EEG-based IQ and learning style classification model using artificial neural network / Muhammad Marwan Anoor. Masters thesis, Universiti Malaya.

	PDF (The Candidate's Agreement) Restricted to Repository staff only Download (179Kb)
	PDF (Thesis M.A.) Download (3391Kb)

Abstract

Intelligence and learning styles are among most widely studied traits in cognitive psychology. Currently, both aspects of cognition can only be assessed using paper-based psychometric tests. The methods, however, are exposed to inconsistency issues due to the variation of examination format and language barriers. Hence, this study proposes an intelligent system for assessing intelligence quotient (IQ) level and learning style from the resting brainwaves using artificial neural network (ANN). Eighty-five individuals from varying highest educational backgrounds have participated in this study. Resting electroencephalogram (EEG) is recorded from the left prefrontal cortex using NeuroSky. Control groups are established using Kolb’s Learning Style Inventory (LSI) and a model developed based on Raven’s Progressive Matrices (RPM). Subsequently, theta, alpha and beta power ratio is extracted from the pre-processed EEG. Distribution and pattern of features show a correlation with the Neural Efficiency Hypothesis of intelligence and Alpha Suppression Theory. The power ratio features are then used to train, validate and test the ANN model. The system has demonstrated satisfactory performance for IQ classification with accuracies of 98.3% for training and 94.7% for testing. The proposed model is also able to classify learning style with accuracies of 96.9% for training and 80.0% for testing.

Actions (For repository staff only : Login required)