Image Classification on ImageNet novel and all classes
55.9Novel Top-5 AccuracyPN w/ G
Evaluation results
| Method | Links | ||
|---|---|---|---|
| PN w/ Gn=2, Backbone=ResNet-102019.04 | 55.9 | 63.2 | |
| PNn=2, Backbone=ResNet-102019.04 | 54.4 | 61 | |
| MNn=2, Backbone=ResNet-102019.04 | 54 | 61 | |
| TAFE-Netn=2, Backbone=ResNet-102019.04 | 53.9 | 61.9 | |
| LogRegn=2, Backbone=ResNet-102019.04 | 51.1 | 49.9 | |
| LogReg w/ Analogiesn=2, Backbone=ResNet-102019.04 | 50.8 | 59.4 | |
| PN w/ Gn=1, Backbone=ResNet-102019.04 | 45 | 56.9 | |
| MNn=1, Backbone=ResNet-102019.04 | 43.6 | 54.4 | |
| TAFE-Netn=1, Backbone=ResNet-102019.04 | 43 | 55.7 | |
| LogReg w/ Analogiesn=1, Backbone=ResNet-102019.04 | 40.7 | 52.2 | |
| PNn=1, Backbone=ResNet-102019.04 | 39.3 | 49.5 | |
| LogRegn=1, Backbone=ResNet-102019.04 | 38.4 | 40.8 |