File(s) under permanent embargo
Evaluation of classification techniques for identifying cognitive load levels using EEG signals
conference contribution
posted on 2020-08-24, 00:00 authored by Syed Moshfeq Salaken, Imali Hettiarachchi, Luke Crameri, Samer HanounSamer Hanoun, Saeid Nahavandi, Thanh Thi NguyenThanh Thi NguyenWearable technology is gaining enormous attention among researchers due to their low cost and ease to transfer from laboratory environment to real world applications. In this paper we evaluate the detection of cognitive load using an off the shelf low cost electroencephalography (EEG) device, namely the EMOTIV EPOC+, by utilising four classifiers including random forest, neural network, linear discriminant analysis (LDA) and logistic regression. We relied on automatic power spectral features calculated from the EmotivPro software for evaluation of classifiers. Using power spectral features automatically calculated from the EMOTIVE headset, we show that the cognitive load levels can be efficiently distinguished (reaching upto 95% accuracy) using the Random forest classification method in near real-time at 8 Hz frequency.