Multi-Label Classification on NUS-WIDE 81 labels
53.5Precision @ K=3Our Approach
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Our Approach2021.01 | 53.5 | 66.5 | 59.3 | 39.4 | 81.6 | 53.1 | 46.7 | |
| LESA2021.01 | 52.3 | 65.1 | 58 | 38.6 | 80 | 52 | 31.5 | |
| One Attention per Label2021.01 | 51.3 | 63.7 | 56.8 | 38 | 78.8 | 51.3 | 32.6 | |
| One Attention per Cluster2021.01 | 51.1 | 63.5 | 56.6 | 37.6 | 77.9 | 50.7 | 31.7 | |
| CNN-RNN2021.01 | 49.9 | 61.7 | 55.2 | 37.7 | 78.1 | 50.8 | 28.3 | |
| WARP2021.01 | 49.1 | 61 | 54.4 | 36.6 | 75.9 | 49.4 | 3.1 | |
| Fast0Tag2021.01 | 48.6 | 60.4 | 53.8 | 36 | 74.6 | 48.6 | 22.4 | |
| WSABIE2021.01 | 48.5 | 60.4 | 53.8 | 36.5 | 75.6 | 49.2 | 3.1 | |
| Logistic2021.01 | 46.1 | 57.3 | 51.1 | 34.2 | 70.8 | 46.1 | 21.6 |