Text Classification on Yelp Review (1000 labels)
37.58Top-1 Error RateFlexMatch + SemiReward
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FlexMatch + SemiRewardBackbone=Bert [Devlin et al., 2018], SemiReward=true2023.10 | 37.58 | 2.26 | |
| SoftMatch/FreeMatch + SemiRewardBackbone=Bert [Devlin et al., 2018], SemiReward=true2023.10 | 38.43 | 2.26 | |
| ILL# Labels=1000, Backbone=Bert-Base2023.05 | 39.67 | — | |
| SoftMatch# Labels=1000, Backbone=Bert-Base2023.05 | 39.76 | — | |
| Dash# Labels=1000, Backbone=Bert-Base2023.05 | 40.14 | — | |
| AdaMatch# Labels=1000, Backbone=Bert-Base2023.05 | 40.16 | — | |
| SimMatch# Labels=1000, Backbone=Bert-Base2023.05 | 40.26 | — | |
| CoMatch# Labels=1000, Backbone=Bert-Base2023.05 | 40.27 | — | |
| FreeMatch# Labels=1000, Backbone=Bert-Base2023.05 | 40.37 | — | |
| FlexMatchBackbone=Bert [Devlin et al., 2018], SemiReward=false2023.10 | 40.38 | — | |
| SoftMatch/FreeMatchBackbone=Bert [Devlin et al., 2018], SemiReward=false2023.10 | 40.43 | — | |
| FlexMatch# Labels=1000, Backbone=Bert-Base2023.05 | 40.51 | — | |
| FixMatch# Labels=1000, Backbone=Bert-Base2023.05 | 40.65 | — | |
| Pseudo Label + SemiRewardBackbone=Bert [Devlin et al., 2018], SemiReward=true2023.10 | 42.08 | 2.26 | |
| Pseudo LabelBackbone=Bert [Devlin et al., 2018], SemiReward=false2023.10 | 44.06 | — | |
| VAT# Labels=1000, Backbone=Bert-Base2023.05 | 45.3 | — | |
| Mean-Teacher# Labels=1000, Backbone=Bert-Base2023.05 | 47.21 | — | |
| Pseudo-Label# Labels=1000, Backbone=Bert-Base2023.05 | 47.33 | — | |
| MixMatch# Labels=1000, Backbone=Bert-Base2023.05 | 51.7 | — |