The Potential of 1D-CNN for EEG Mental Attention State Detection

In cognitive neuroscience, attention detection via electroencephalogram (EEG) signals is a crucial task, especially when looking into brain-computer interfaces (BCIs). However, the current biggest challenge in BCI research is the accurate classification of EEG signals. The old methods, such as visual assessment, were neither standardised or statistically analysed. Because of these limitations, many techniques have been created to quantify and extract meaningful data from EEG signals. It has been shown that machine learning algorithms can reliably classify attention states using EEG data. Support Vector Machine (SVM) and k-Nearest Neighbors (KNN) are two examples of such methods. However, more study is required to ensure these algorithms are as precise and reliable as possible. In light of the challenges presented by EEG signal classification, we have employed state-of-the-art deep learning algorithms to enhance the identification of mental attention states. Specifically, we have used a sophisticated deep learning model that integrates spatial dependency analysis, namely 1D Convolutional Neural Networks (CNN). By leveraging the power of 1D CNN, our approach is able to capture intricate spatial patterns within EEG signals, allowing for robust feature extraction and improved classification accuracy. Our model has undergone rigorous training on a diverse dataset comprising 34 EEG signals from 5 participants across different conditions. The model successfully differentiates between distinct mental states (focused, unfocused, drowsy) by analysing EEG data gathered in the experimental setting, with an impressive accuracy of 98.47% and a loss of 0.0144. This study underscores the importance of leveraging advanced computational approaches to unlock the valuable information contained within EEG signals.