EEG-based mental workload estimation using deep BLSTM-LSTM network and evolutionary algorithm

•A deep model-based mental workload estimation from EEG has been implemented.•The proposed model uses GWO & deep BLSTM-LSTM for producing effective results.•Evaluation is done based on two tasks: No task & SIMKAP-based multitasking activity.•The proposed model achieves 86.33% and 82.57% accu...

Full description

Saved in:
Bibliographic Details
Published in:Biomedical signal processing and control Vol. 60; p. 101989
Main Authors: Das Chakladar, Debashis, Dey, Shubhashis, Roy, Partha Pratim, Dogra, Debi Prosad
Format: Journal Article
Language:English
Published: Elsevier Ltd 01.07.2020
Subjects:
BCI
Electroencephalogram
GWO
LSTM
Mental workload
BCI
LSTM
Electroencephalogram
Mental workload
GWO
ISSN:1746-8094, 1746-8108, 1746-8108
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•A deep model-based mental workload estimation from EEG has been implemented.•The proposed model uses GWO & deep BLSTM-LSTM for producing effective results.•Evaluation is done based on two tasks: No task & SIMKAP-based multitasking activity.•The proposed model achieves 86.33% and 82.57% accuracy for the two tasks. The mental workload can be estimated by monitoring different mental states from neural activity. The spectral power of EEG and Event-Related Potentials (ERPs) are the two mediums for monitoring the mental states. In this paper, we estimate the workload during the multitasking mental activities of human subjects. The estimation of mental workload is done using the “STEW” dataset [16]. The dataset consists of two tasks, namely “No task” and “simultaneous capacity (SIMKAP)-based multitasking activity”. Different workload levels of two tasks have been estimated using the composite framework consists of Grey Wolf Optimizer (GWO) and deep neural network. GWO has been used to select optimized features related to mental activities. Other optimization techniques such as Genetic Algorithm (GA), Particle Swarm Optimization (PSO) are generally slower compared to the convergence rate of GWO. A deep hybrid model based on Bidirectional Long Short-Term Memory (BLSTM) and Long Short-Term Memory (LSTM) has been proposed for the classification of workload levels. The proposed deep model achieves 86.33% and 82.57% classification accuracy for “No task” and “SIMKAP-based multitasking activity,” respectively. A judicious distinction between different workload levels at higher accuracy will essentially increase the performance of an operator, which effectively improves the efficiency of the Brain-Computer Interface (BCI) systems.
ISSN:1746-8094
1746-8108
1746-8108
DOI:10.1016/j.bspc.2020.101989