Published Paper Details:

Electroencephalography (EEG) is a vital tool for monitoring brain activity, but its utility is often compromised by the presence of non-neuronal artifacts, such as eye blinks, muscle movements, and environmental noise. Traditional methods for artifact detection typically rely on manual inspection or supervised machine learning models, which require extensive labeled datasets. These approaches are time-consuming, subject to human error, and difficult to scale. Self-supervised learning (SSL) presents a promising alternative by enabling models to learn from large amounts of unlabeled data through proxy tasks. This paper explores the application of SSL in EEG artifact detection, proposing a novel approach to automate the identification and removal of artifacts without the need for manual labeling. By leveraging the inherent structural properties of EEG signals, SSL models can effectively distinguish between artifacts and meaningful neural activity. We evaluate the performance of SSL-based models against conventional supervised approaches, demonstrating that SSL can achieve comparable or superior accuracy while requiring less human intervention. Our findings suggest that SSL has the potential to enhance the efficiency and scalability of EEG analysis, making it a valuable tool for both clinical and research applications. This work contributes to the ongoing development of automated EEG processing methods, with the goal of improving signal quality and ensuring more reliable interpretations of brain activity.