{"ID":6537484,"CreatedAt":"2026-07-14T02:54:43.516908796Z","UpdatedAt":"2026-07-15T03:28:55.185153975Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.11578","arxiv_id":"2607.11578","title":"DiffEEG: A Self-Supervised Denoising Diffusion Model for Learning EEG Generic Representations","abstract":"Deep learning for EEG-based seizure detection faces critical challenges: severe annotation scarcity and extreme class imbalance, where ictal events comprise less than 10\\% of clinical recordings. We present DiffEEG, a 9.6M-parameter self-supervised foundation model that addresses both limitations through denoising diffusion pre-training and reinforcement learning (RL)-based fine-tuning. Pre-trained on 1.3M unlabeled segments from the Temple University Hospital Seizure Corpus (TUHSZ), DiffEEG learns generic neural representations via a 1D U-Net with multi-head self-attention. For downstream adaptation, a reinforced decision layer employs policy gradient optimization to directly maximize F1-score, prioritizing sensitivity to rare seizure events over overall accuracy. Under strict patient-wise evaluation (279 patients, Leave-One-Fold-Out), DiffEEG achieves 61\\% accuracy and 59\\% F1 for 4-class seizure subtyping, and 81\\% accuracy with 85\\% weighted F1 for binary detection, maintaining clinically viable seizure recall (59\\%) despite extreme imbalance (6.7\\% prevalence). Segment-level evaluation establishes an upper bound of 97.6\\% accuracy, confirming strong architectural capacity. DiffEEG demonstrates that diffusion-based pre-training combined with metric-aware reinforcement learning enables clinically deployable seizure monitoring with minimal labeled data requirements.","short_abstract":"Deep learning for EEG-based seizure detection faces critical challenges: severe annotation scarcity and extreme class imbalance, where ictal events comprise less than 10\\% of clinical recordings. We present DiffEEG, a 9.6M-parameter self-supervised foundation model that addresses both limitations through denoising diff...","url_abs":"https://arxiv.org/abs/2607.11578","url_pdf":"https://arxiv.org/pdf/2607.11578v1","authors":"[\"Abdulkader Helwan\",\"Lina Abou-Abbas\",\"Hussein El Amouri\",\"Belkacem Chikhaoui\",\"Khadidja Henni\"]","published":"2026-07-13T13:59:45Z","proceeding":"cs.LG","tasks":"[\"cs.LG\",\"cs.AI\",\"cs.CV\",\"eess.SP\"]","methods":"[\"Reinforcement Learning\",\"Diffusion Model\"]","has_code":false}