Core Decoding Workflow

Estimated reading time:2 minutes

Preprocess real nm000118 SSVEP data, compare trial and subject splits, fit a spectral baseline, and persist prepared windows. Each script runs independently on one to three participants (about 7–21 MB) and uses EEGDASH_CACHE_DIR for persistent storage. Subject-disjoint evaluation uses scikit-learn group splitting.

Preprocess and window recorded EEG

Preprocess and window recorded EEG

Split real windows without subject leakage

Split real windows without subject leakage

Train a baseline on real SSVEP trials

Train a baseline on real SSVEP trials

Save and reload prepared EEG windows

Save and reload prepared EEG windows