EEG-based mental workload level estimation using deep learning models

Kunjira Kingphai, Yashar Moshfeghi

Research output: Chapter in Book/Report/Conference proceedingConference contribution book


Deep learning-based approaches have recently received much attention and managed to accurately capture variance characteristics in the Electroencephalography (EEG) signals. In this paper, we aim to classify the subject’s mental workload (MWL) level from EEG signal by using deep learning models named Stacked Gated Recurrent Unit (GRU), Bidirectional GRU (BGRU), BGRU-GRU, Stacked Long-Short Term Memory (LSTM), Bidirectional LSTM (BLSTM), BLSTM-LSTM and Convolutional Neural Network (CNN). The classification was performed on a publicly available mental workload dataset, STEW. Our encouraging results show the potential of deep learning models for MWL level detection.
Original languageEnglish
Title of host publicationContemporary Ergonomics & Human Factors 2022
Subtitle of host publicationProceedings for the Annual Conference of the Chartered Institute of Ergonomics and Human Factors
EditorsNora Balfe, David Golightly
Place of PublicationBirmingham
Number of pages3
Publication statusPublished - 24 Jun 2022


  • EEG
  • deep learning
  • mental workload classification


Dive into the research topics of 'EEG-based mental workload level estimation using deep learning models'. Together they form a unique fingerprint.

Cite this