Data Augmentation for Seizure Prediction with Generative Diffusion Model
Abstract
Task: EEG-based seizure prediction
Existing DA methodos generate samples by overrlapping / recombining prerictal data for each seizure
\(\rightarrow\) limited by original data
Solution: propose DiffEEG
Previous Methods
DiffEEG
Diffusion network
- conditined on the initial STFT spectogram to provide the guiding time-frequency features
