Table 6.
Comparison between CMC-HF and other state-of-the-art methods for the CMU-MOSI dataset.
| Model | Acc (7 class h) | Acc (2 class h) | F1 score | MAE | Corr |
|---|---|---|---|---|---|
| MFN | 34.1 | 77.4 | 77.3 | 0.965 | 0.632 |
| MCTN | 35.6 | 79.3 | 79.1 | 0.909 | 0.676 |
| RAVEN | 33.2 | 78.0 | 76.6 | 0.915 | 0.691 |
| HFusion | 35.3 | 77.9 | 79.8 | 0.890 | 0.703 |
| MulT | 39.1 | 81.1 | 81.0 | 0.889 | 0.686 |
| SSE-FT | 55.7 | 87.3 | 87 | 0.529 | 0.792 |
| CMC-HF (ours) | 58.2 | 88.2 | 87.3 | 0.581 | 0.798 |
The bold values represent the best values of the attributes in this column. The larger the values of Acc (7 class-h), Acc (2 class-h), F1 score, and Corr, the better the corresponding models. The smaller the value of MAE, the better the corresponding model effect.