Anger Detection on CREMA-D
0.83AUC-ROCMP-IB+MSTF
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MP-IB+MSTFtransfer_mode=Zero-shot, head_type=state head only, domain_shift=clinical to acted, fine-tuning=false2026.05 | 0.83 | 0.805 | 81 | |
| MP-IBtransfer_mode=Zero-shot, head_type=state head only, domain_shift=clinical to acted, fine-tuning=false2026.05 | 0.817 | 0.791 | 79 | |
| MI-Minimizationtransfer_mode=Zero-shot, head_type=state head only, domain_shift=clinical to acted, fine-tuning=false2026.05 | 0.76 | 0.73 | 73 | |
| β-VAEtransfer_mode=Zero-shot, head_type=state head only, domain_shift=clinical to acted, fine-tuning=false2026.05 | 0.74 | 0.71 | 71 | |
| B2: Adversarialtransfer_mode=Zero-shot, head_type=state head only, domain_shift=clinical to acted, fine-tuning=false2026.05 | 0.72 | 0.69 | 68 | |
| B3: VQ-Disent.transfer_mode=Zero-shot, head_type=state head only, domain_shift=clinical to acted, fine-tuning=false2026.05 | 0.69 | 0.66 | 65 | |
| B1: SingleHeadtransfer_mode=Zero-shot, head_type=state head only, domain_shift=clinical to acted, fine-tuning=false2026.05 | 0.65 | 0.62 | 61 |