Table 7 Ablation results for multimodal vs. unimodal performance.
From: MULTICAUSENET temporal attention for multimodal emotion cause pair extraction
Model | IEMOCAP (WF1) | MELD (WF1) |
|---|---|---|
Text-only | 62.45 | 43.34 |
Audio-only | 61.43 | 42.11 |
Video-only | 60.11 | 40.54 |
Full model (T+V+A) | 73.02 | 53.67 |
