Laya: A LeJEPA Approach to EEG via Latent Prediction over Reconstruction
This work addresses the challenge of learning transferable representations for EEG data, which is important for applications like brain-computer interfaces, though it appears incremental as it adapts an existing SSL paradigm to a new domain.
The paper tackled the problem of limited effectiveness in EEG foundation models by introducing Laya, a model based on LeJEPA that uses latent prediction instead of signal reconstruction, resulting in improved performance under linear probing on EEG benchmarks.
Electroencephalography (EEG) is a widely used tool for studying brain function, with applications in clinical neuroscience, diagnosis, and brain-computer interfaces (BCIs). Recent EEG foundation models trained on large unlabeled corpora aim to learn transferable representations, but their effectiveness remains unclear; reported improvements over smaller task-specific models are often modest, sensitive to downstream adaptation and fine-tuning strategies, and limited under linear probing. We hypothesize that one contributing factor is the reliance on signal reconstruction as the primary self-supervised learning (SSL) objective, which biases representations toward high-variance artifacts rather than task-relevant neural structure. To address this limitation, we explore an SSL paradigm based on Joint Embedding Predictive Architectures (JEPA), which learn by predicting latent representations instead of reconstructing raw signals. While earlier JEPA-style methods often rely on additional heuristics to ensure training stability, recent advances such as LeJEPA provide a more principled and stable formulation. We introduce Laya, the first EEG foundation model based on LeJEPA. Across a range of EEG benchmarks, Laya demonstrates improved performance under linear probing compared to reconstruction-based baselines, suggesting that latent predictive objectives offer a promising direction for learning transferable, high-level EEG representations.