Innovative deep learning models for EEG-based vigilance detection

Khessiba, Souhir; Blaiech, Ahmed Ghazi; Khalifa, Khaled Ben; Abdallah, Asma Ben; Bedoui, Mohamed Hédi

doi:10.1007/s00521-020-05467-5

Cited by 32 publications

(16 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A large body of prior literature has indicated that neurophysiological signals contain information related to physiological state changes, which were acquired via EEG in our experiment ( Lin et al, 2010 ; Xia et al, 2018 ; Asif et al, 2019 ; Monteiro et al, 2019 ; Xu et al, 2019 ; Bajaj et al, 2020 ). Some of the research in recent years has applied machine learning or deep learning methods to identify different stages of physiological states ( Jebelli et al, 2019 ; Ma et al, 2019 ; Khessiba et al, 2021 ). However, the high computational power that deep learning demands does not necessarily provide better classification performance due to inter-subject variabilities.…”

Section: Discussionmentioning

confidence: 99%

Multi-Parameter Physiological State Monitoring in Target Detection Under Real-World Settings

Chang

Tsai

et al. 2021

Front. Hum. Neurosci.

View full text Add to dashboard Cite

Mental state changes induced by stimuli under experimental settings or by daily events in real life affect task performance and are entwined with physical and mental health. In this study, we developed a physiological state indicator with five parameters that reflect the subject’s real-time physiological states based on online EEG signal processing. These five parameters are attention, fatigue, stress, and the brain activity shifts of the left and right hemispheres. We designed a target detection experiment modified by a cognitive attention network test for validating the effectiveness of the proposed indicator, as such conditions would better approximate a real chaotic environment. Results demonstrated that attention levels while performing the target detection task were significantly higher than during rest periods, but also exhibited a decay over time. In contrast, the fatigue level increased gradually and plateaued by the third rest period. Similar to attention levels, the stress level decreased as the experiment proceeded. These parameters are therefore shown to be highly correlated to different stages of the experiment, suggesting their usage as primary factors in passive brain-computer interfaces (BCI). In addition, the left and right brain activity indexes reveal the EEG neural modulations of the corresponding hemispheres, which set a feasible reference of activation for an active BCI control system, such as one executing motor imagery tasks. The proposed indicator is applicable to potential passive and active BCI applications for monitoring the subject’s physiological state change in real-time, along with providing a means of evaluating the associated signal quality to enhance the BCI performance.

show abstract

Section: Discussionmentioning

confidence: 99%

Multi-Parameter Physiological State Monitoring in Target Detection Under Real-World Settings

Chang

Tsai

et al. 2021

Front. Hum. Neurosci.

View full text Add to dashboard Cite

show abstract

“…They performed experiments with the physionet sleep EEG dataset and reported 86.30% accuracy for the wake and sleep state classification. In [35], the authors identified drowsiness using DL classifiers such as a convolutional neural network (CNN) and long short-term memory (LSTM) on spectral band energy features captured with FFT. Zeng et al [36], unlike the hand-engineered features, classification have been performed with automated features using deep learning CNN classification process to identify drowsiness as shown in Path-A in Figure 1.…”

Section: Drowsiness Detection Using Eeg Signalsmentioning

confidence: 99%

Section: Related Work or Backgroundmentioning

confidence: 99%

Automated classification system for drowsiness detection using convolutional neural network and electroencephalogram

Balam

Sameer

Chinara

2021

IET Intelligent Trans Sys

View full text Add to dashboard Cite

Detecting drowsiness in drivers while driving is extremely important to avoid possible accidents and reduce the fatality rate due to drivers sleeping at the wheel. A real‐time alert generation when the driver might possibly go into sleepy state is essential to safeguard any unwarranted incidents. Wearable sensors to monitor vehicle movement and camera‐based systems to monitor driver behaviour are commonly used to detect driver drowsiness. Due to the fact that electroencephalogram (EEG) signals have the ability to monitor the mood of humans and are easily obtainable, many different EEG‐based drowsiness detection systems have been proposed to date. In this study, a novel deep learning architecture based on a convolutional neural network (CNN) is proposed for automated drowsiness detection using a single‐channel EEG signal. To improve the generalization performance of the proposed method, subject‐wise, cross‐subject‐wise, and combined‐subjects‐wise validations have been employed. The whole of the work is carried over pre‐recorded sleep state EEG data obtained from benchmarked dataset. The experimental results show a superior detection capability compared to the existing state–of–the–art drowsiness detection methods using single‐channel EEG signals.

show abstract

“…In [12], three types of deep covariance learning models were suggested to predict drivers' drowsy and alert states using EEG signals: the CNN, the Symmetric Positive Definite Network (SPDNet), and the Deep Neural Network (DNN).The experimental results indicated that all the three models of deep covariance-learning reported a very good classification performance compared with shallow learning methods. In [14], the authors proposed two DL models to predict individuals' vigilance states based on the study of one derivation of EEG signals: a 1D-UNet model and 1D-UNet-Long Short-Term Memory (1D-UNet-LSTM). Experimental results showed that the suggested models can stabilize the training process and well recognize the subject vigilance.…”

Section: Introductionmentioning

confidence: 99%

“…Specifically, the per-class average of precision and recall could be respectively up to 86% with 1D-UNet and 85% with 1D-UNet-LSTM. All these studies have used several DL approaches to analyze EEG signals [12] [2] [14], but the choice of the architecture has been done empirically by the human expert through a slow trial and error process, guided mainly by intuition.…”

Section: Introductionmentioning

confidence: 99%

Hyperparameter Optimization of Deep Learning Models for EEG-Based Vigilance Detection

Khessiba

Blaiech

Manzanera

et al. 2022

Advances in Computational Collective Intelligence

Self Cite

View full text Add to dashboard Cite

ElectroEncephaloGraphy (EEG) signals have a nonlinear and complex nature and require the design of sophisticated methods for their analysis. Thus, Deep Learning (DL) models, which have enabled the automatic extraction of complex data features at high levels of abstraction, play a growing role in the field of medical science to help diagnose various diseases, and have been successfully used to predict the vigilance states of individuals. However,the performance of these models is highly sensitive to the choice of the hyper-parameters that define the structure of the network and the learning process. When targeting an application, tuning the hyper-parameters of deep neural networks is a tedious and time-consuming process.This explains the necessity of automating the calibration of these hyper-parameters. In this paper, we perform hyperparameters optimization using two popular methods: Tree Parzen Estimator (TPE) and Bayesian optimisation (BO) to predict vigilance states of individuals based on their EEG signal. The performance of the methods is evaluated on the vigilance states classification. Compared with empirical optimization,the accuracy is improved from 0.84 to 0.93 with TPE and from 0.84 to 0.97 with Bayesian optimization using the 1D-UNet-LSTM deep learning model. Obtained results show that the combination of the 1D-UNet encoder and LSTM offers an excellent compromise between the performance and network size (thus training duration), which allows a more efficient hyper-parameter optimization.

show abstract

Innovative deep learning models for EEG-based vigilance detection

Cited by 32 publications

References 40 publications

Multi-Parameter Physiological State Monitoring in Target Detection Under Real-World Settings

Multi-Parameter Physiological State Monitoring in Target Detection Under Real-World Settings

Automated classification system for drowsiness detection using convolutional neural network and electroencephalogram

Hyperparameter Optimization of Deep Learning Models for EEG-Based Vigilance Detection

Contact Info

Product

Resources

About