From Video to EEG: Adapting Joint Embedding Predictive Architecture to Uncover Visual Concepts in Brain Signal Analysis

EEG signals capture brain activity with high temporal but low spatial resolution, supporting applications such as neurological diagnosis, cognitive monitoring, and brain-computer interfaces. However, effective analysis remains challenging due to limited labeled data, high dimensionality, and the lack of scalable models that fully capture spatiotemporal dependencies. Existing self-supervised learning (SSL) methods often focus on either spatial or temporal features in isolation, leading to suboptimal representations. To this end, we propose EEG-VJEPA, a novel adaptation of the Video Joint Embedding Predictive Architecture (V-JEPA) for EEG classification. By treating EEG as video-like sequences, EEG-VJEPA learns semantically meaningful spatiotemporal representations using joint embeddings and adaptive masking. To our knowledge, this is the first work that exploits V-JEPA for EEG classification and explores the visual concepts learned by the model. EEG-VJEPA achieves state-of-the-art performance on the publicly available Temple University Hospital (TUH) Abnormal EEG dataset, outperforming both self-supervised and fully supervised baselines. Likewise, we demonstrate the model's good generalization ability on an independent, smaller clinical dataset from the General Hospital of Thessaloniki, involving dementia classification. Keywords: Electroencephalography (EEG), Joint Embedding Predictive Architecture (JEPA), Vision Transformer (ViT), Self-Supervised Learning, Foundation Model 1. Introduction Electroencephalography (EEG) is a non-invasive and cost-effective technique for capturing rhythmic brain activity, widely used in clinical neurology to monitor conditions such as epilepsy, encephalopathy, and cognitive disorders [1, 2].