Leveraging Generic Time Series Foundation Models for EEG Classification