文章基本信息

标题：Learning Style Classification via EEG Sub-band Spectral Centroid Frequency Features
本地全文：下载
作者：Megat Syahirul Amin Megat Ali ; Aisyah Hartini Jahidin ; Nooritawati Md Tahir 等
期刊名称：International Journal of Electrical and Computer Engineering
电子版ISSN：2088-8708
出版年度：2014
卷号：4
期号：6
页码：931-938
DOI：10.11591/ijece.v4i6.6833
语种：English
出版社：Institute of Advanced Engineering and Science (IAES)
摘要：Kolb’s Experiential Learning Theory postulates that in learning, knowledge is created by the learners’ ability to absorb and transform experience. Many studies have previously suggested that at rest, the brain emits signatures that can be associated with cognitive and behavioural patterns. Hence, the study attempts to characterise and classify learning styles from EEG using the spectral centroid frequency features. Initially, learning style of 68 university students has been assessed using Kolb’s Learning Style Inventory. Resting EEG is then recorded from the prefrontal cortex. Next, the EEG is pre-processed and filtered into alpha and theta sub-bands in which the spectral centroid frequencies are computed from the corresponding power spectral densities. The dataset is further enhanced to 160 samples via synthetic EEG. The obtained features are then used as input to the k-nearest neighbour classifier that is incorporated with k-fold cross-validation. Feature classification via k-nearest neighbour has attained five-fold mean training and testing accuracies of 100% and 97.5%, respectively. Hence, results show that the alpha and theta spectral centroid frequencies represent distinct and stable EEG signature to distinguish learning styles from the resting brain.
关键词：Computer and Informatics;EEG; learning style; spectral centroid frequency; k-nearest neighbour; k-fold cross-validation