Aspect-based Sentiment Classification on 19 ASC tasks averaged (test)
88.29AccuracyB-CL
Evaluation results
| Method | Links | ||
|---|---|---|---|
| B-CLScenario=Continual Learning, evaluation_protocol=Final Result (after all tasks learned)2021.12 | 88.29 | 81.4 | |
| B-CLScenario=Continual Learning, evaluation_protocol=Forward Transfer (results when first learned)2021.12 | 88.09 | 79.93 | |
| OWMScenario=Continual Learning, Backbone=BERT (Frozen)2021.12 | 87.02 | 79.31 | |
| HATScenario=Continual Learning, Backbone=BERT (Frozen)2021.12 | 86.74 | 78.16 | |
| EWCScenario=Continual Learning, Backbone=BERT (Frozen)2021.12 | 86.37 | 74.52 | |
| Adapter-BERTScenario=Non-continual Learning (NL), Backbone=Adapter-BERT, Learning Type=NL2021.12 | 85.96 | 78.07 | |
| BERTScenario=Non-continual Learning (NL), Backbone=BERT, Learning Type=NL2021.12 | 85.84 | 76.35 | |
| KANScenario=Continual Learning, Backbone=BERT (Frozen)2021.12 | 85.49 | 77.38 | |
| SRKScenario=Continual Learning, Backbone=BERT (Frozen)2021.12 | 84.76 | 78.52 | |
| UCLScenario=Continual Learning, Backbone=W2V2021.12 | 84.41 | 75.99 | |
| EWCScenario=Continual Learning, Backbone=W2V2021.12 | 84.16 | 72.29 | |
| UCLScenario=Continual Learning, Backbone=BERT (Frozen)2021.12 | 83.89 | 74.82 | |
| OWMScenario=Continual Learning, Backbone=W2V2021.12 | 82.7 | 71.18 | |
| W2VScenario=Continual Learning (WDF), Backbone=W2V, Learning Type=WDF2021.12 | 82.69 | 73.56 | |
| HATScenario=Continual Learning, Backbone=W2V2021.12 | 80.83 | 63.63 | |
| W2VScenario=Non-continual Learning (NL), Backbone=W2V, Learning Type=NL2021.12 | 77.01 | 51.89 | |
| KANScenario=Continual Learning, Backbone=W2V2021.12 | 72.06 | 40.01 | |
| SRKScenario=Continual Learning, Backbone=W2V2021.12 | 71.01 | 39.63 | |
| Adapter-BERTScenario=Continual Learning (WDF), Backbone=Adapter-BERT, Learning Type=WDF2021.12 | 54.03 | 44.81 | |
| BERTScenario=Continual Learning (WDF), Backbone=BERT, Learning Type=WDF2021.12 | 49.6 | 43.08 |