Multimodal emotion recognition from ECG and GSR signals. 88 handcrafted features + gradient boosting ensembles beat the published CNN-LSTM baseline by over 14 points (80.97% vs 66.64%).
python machine-learning signal-processing ecg feature-extraction supervised-learning lightgbm ensemble-learning heart-rate-variability physiological-signals affective-computing gradient-boosting emotion-recognition gsr catboost multimodal valence-arousal histogram-gradient-boosting yaad-dataset empathic-computing
-
Updated
Sep 18, 2026 - Python