Table 5 Performance comparison between our method and baselines on SIMS Dataset.
From: Multimodal sentiment analysis based on multi-layer feature fusion and multi-task learning
Models | SIMS | Â | ||||
|---|---|---|---|---|---|---|
Acc2 | F1-score | MAE | Corr | Acc3 | Acc5 | |
EF-\(\hbox {LSTM}^\textrm{a}\) | 69.37 | 56.82 | 0.599 | 0.521 | 54.27 | 21.23 |
LF-\(\hbox {DNN}^\textrm{a}\) | 78.99 | 79.72 | 0.419 | 0.589 | 64.99 | 41.36 |
\(\hbox {TFN}^\textrm{a}\) | 80.31 | 80.66 | 0.451 | 0.581 | 64.33 | 36.76 |
\(\hbox {LMF}^\textrm{a}\) | 78.99 | 78.99 | 0.442 | 0.574 | 66.74 | 37.86 |
\(\hbox {MFN}^\textrm{a}\) | 79.21 | 79.15 | 0.434 | 0.581 | 66.08 | 39.39 |
Graph-\(\hbox {MFN}^\textrm{a}\) | 79.65 | 80.40 | 0.477 | 0.581 | 67.4 | 39.17 |
\(\hbox {MulT}^\textrm{a}\) | 79.65 | 79.94 | 0.439 | 0.582 | 65.86 | 38.95 |
\(\hbox {MISA}^\textrm{b}\) | 69.37 | 56.82 | 0.587 | 0.113 | 51.42 | 20.79 |
Self-\(\hbox {MM}^*\) | 78.77 | 78.88 | 0.425 | 0.591 | 63.68 | 39.82 |
\(\hbox {MMIM}^*\) | 73.96 | 74.59 | 0.441 | 0.531 | 60.61 | 43.54 |
Our method | 81.56 | 81.27 | 0.423 | 0.583 | 67.77 | 45.20 |
