Table 1 Comparison on MOSI and MOSEI.
From: Adaptive multimodal transformer based on exchanging for multimodal sentiment analysis
Model | MOSI | MOSEI | ||||||
|---|---|---|---|---|---|---|---|---|
Acc2 | F1 | MAE | Corr | Acc2 | F1 | MAE | Corr | |
MFN | 78.87 | 78.90 | 0.927 | 0.670 | 82.86 | 82.85 | 0.573 | 0.718 |
EF-LSTM | 78.48 | 78.51 | 0.948 | 0.699 | 80.79 | 80.67 | 0.605 | 0.682 |
TFN | 79.08 | 79.11 | 0.947 | 0.673 | 81.89 | 81.74 | 0.573 | 0.714 |
LMF | 79.18 | 79.15 | 0.950 | 0.651 | 83.48 | 83.36 | 0.576 | 0.717 |
MulT | 80.98 | 80.95 | 0.880 | 0.702 | 84.63 | 84.52 | 0.559 | 0.733 |
Self-MM | 84.65 | 84.69 | 0.718 | 0.793 | 85.15 | 84.93 | 0.531 | 0.765 |
MISA | 83.54 | 83.58 | 0.777 | 0.776 | 84.67 | 84.66 | 0.558 | 0.752 |
BBFN | 84.31 | 84.32 | 0.776 | 0.755 | 86.22 | 86.21 | 0.529 | 0.767 |
EMT-DLFR | 85.01 | 85.12 | 0.705 | 0.798 | 86.13 | 86.04 | 0.527 | 0.774 |
CENet | 85.20 | 85.22 | 0.725 | 0.795 | 86.38 | 86.32 | 0.526 | 0.778 |
UniMSE | 86.90 | 86.42 | 0.691 | 0.809 | 87.50 | 86.46 | 0.523 | 0.773 |
TETFN | 86.10 | 86.07 | 0.717 | 0.800 | 85.18 | 85.27 | 0.551 | 0.748 |
HCIL | 86.07 | 86.01 | 0.703 | 0.810 | 85.97 | 85.29 | 0.532 | 0.768 |
MTAMW | 86.59 | 86.46 | 0.712 | 0.794 | 86.49 | 86.45 | 0.525 | 0.782 |
AMTE | 89.18 | 89.25 | 0.742 | 0.823 | 88.28 | 88.28 | 0.517 | 0.794 |
