DiffEEG: A Self-Supervised Denoising Diffusion Model for Learning EEG seizure-related Representations

Name: DiffEEG: A Self-Supervised Denoising Diffusion Model for Learning EEG seizure-related Representations
Uploaded: 2026-10-22T11:14:10.241830+00:00
Duration: 368 s
Description: DiffEEG is a self-supervised foundation model for EEG seizure detection that uses denoising diffusion pre-training and reinforcement learning to overcome data scarcity and severe class imbalance. It learns robust neural representations from unlabeled EEG data, allowing it to achieve clinically viable seizure monitoring with minimal labeled examples.

DiffEEG is a self-supervised foundation model for EEG seizure detection that uses denoising diffusion pre-training and reinforcement learning to overcome data scarcity and severe class imbalance. It learns robust neural representations from unlabeled EEG data, allowing it to achieve clinically viable seizure monitoring with minimal labeled examples.

Open the full episode