Figure 4
From: Driver drowsiness estimation using EEG signals with a dynamical encoder–decoder modeling framework
Flowchart of the proposed neural encdoer-decoder modeling framework for PERCLOS. This framework has two main steps: encoder model, and decoder model. In the encoder model, state transition process of PERCLOS and the relationship between every EEG feature and PERCLOS are defined. Biomarkers are obtained using t-test and they are used in the decoder step to predict the PERCLOS in real-time using Bayes filtering.
