Identified types of objects seen in a video by subjects from corresponding ECoG data.
The most useful feature- and label-related tools used include common spatial patterns, LDA, SHAP values, and CLIP.
The most useful ML models were k-NN, gradient boosting trees, and random forest.