Empathy Detection on HRI (LOSO)
74.6AccuracyTransformer
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| TransformerPrivacy regime=Partial privacy, Learning features=temporal features2025.04 | 74.6 | 82.7 | 74.2 | 75.4 | 74.8 | |
| TFMPathyFT(TabPFN; frozen X and Y encoders)Privacy regime=Strong privacy, Learning features=summary features, Paradigm=Fine-Tuning, Backbone=TabPFN, Trainable layers=frozen X and Y encoders2025.04 | 73 | 66.9 | 81.8 | 59 | 68.6 | |
| TCNPrivacy regime=Partial privacy, Learning features=temporal features2025.04 | 70.5 | 78.8 | 70.5 | 70.5 | 70.5 | |
| TFMPathyFT(TabPFN; all layers trainable)Privacy regime=Strong privacy, Learning features=summary features, Paradigm=Fine-Tuning, Backbone=TabPFN, Trainable layers=all layers trainable2025.04 | 68.9 | 64.2 | 78 | 52.5 | 62.7 | |
| LSTMPrivacy regime=Partial privacy, Learning features=temporal features2025.04 | 64.8 | 78.7 | 68 | 55.7 | 61.3 | |
| SVMPrivacy regime=Strong privacy, Learning features=summary features2025.04 | 59 | 56.4 | 59 | 59 | 59 | |
| TFMPathyICL(TabICL)Privacy regime=Strong privacy, Learning features=summary features, Paradigm=In-Context Learning, Backbone=TabICL2025.04 | 54.1 | 51.9 | 54.7 | 47.5 | 50.9 | |
| RFPrivacy regime=Strong privacy, Learning features=summary features2025.04 | 53.3 | 52.3 | 53.6 | 49.2 | 51.3 | |
| XGBoostPrivacy regime=Strong privacy, Learning features=summary features2025.04 | 51.6 | 51.1 | 51.9 | 44.3 | 47.8 | |
| TFMPathyICL(TabPFN)Privacy regime=Strong privacy, Learning features=summary features, Paradigm=In-Context Learning, Backbone=TabPFN2025.04 | 45.9 | 47.2 | 45.1 | 37.7 | 41.1 |