Image Classification on FLW
91.27AccuracyOurs (T*csgcn)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Ours (T*csgcn)Backbone=ResNet50, Training Scenario=FT2025.04 | 91.27 | — | |
| Insert LNBackbone=ResNet50, Training Scenario=FT2025.04 | 90.32 | — | |
| Ours (T*csgcn)Backbone=SwinT, Training Scenario=FT2025.04 | 90.21 | — | |
| T = 1Backbone=ResNet50, Training Scenario=FT2025.04 | 90.18 | — | |
| Ours (T*cn)Backbone=ResNet50, Training Scenario=FT2025.04 | 89.17 | — | |
| Ours (T*csg)Backbone=ResNet50, Training Scenario=FT2025.04 | 84.45 | — | |
| T = 1Backbone=SwinT, Training Scenario=FT2025.04 | 80.29 | — | |
| Ours (T*)Backbone=ResNet50, Training Scenario=FT2025.04 | 79.1 | — | |
| Insert LNBackbone=SwinT, Training Scenario=FT2025.04 | 77.18 | — | |
| Ours (T*cn)Backbone=SwinT, Training Scenario=FT2025.04 | 76.57 | — | |
| Ours (T*)Backbone=SwinT, Training Scenario=FT2025.04 | 66.47 | — | |
| O-TPTBackbone=CLIP-RN502025.03 | 65.61 | 2.5 | |
| C-TPTBackbone=CLIP-RN502025.03 | 65.2 | 4.14 | |
| Robust-adapt-SaLs-CTPTBackbone=CLIP-RN502025.03 | 65 | 2.29 | |
| Robust-adapt-ZS-CTPTBackbone=CLIP-RN502025.03 | 64.92 | 3.39 | |
| Robust-adapt-Penalty-CTPTBackbone=CLIP-RN502025.03 | 64.88 | 4.03 | |
| TPTBackbone=CLIP-RN502025.03 | 62.5 | 13.4 | |
| Zero ShotBackbone=CLIP-RN502025.03 | 61 | 3.19 | |
| Ours (T*csg)Backbone=SwinT, Training Scenario=FT2025.04 | 60.81 | — | |
| Ours (T*csg)Backbone=ResNet50, Training Scenario=init2025.04 | 54.81 | — | |
| Ours (T*cn)Backbone=ResNet50, Training Scenario=init2025.04 | 54.76 | — | |
| Ours (T*)Backbone=ResNet50, Training Scenario=init2025.04 | 53.54 | — | |
| Ours (T*csgcn)Backbone=ResNet50, Training Scenario=init2025.04 | 52.98 | — | |
| T = sqrt(M)Backbone=SwinT, Training Scenario=FT2025.04 | 51.02 | — | |
| Insert LNBackbone=ResNet50, Training Scenario=init2025.04 | 48.84 | — | |
| T = 1Backbone=ResNet50, Training Scenario=init2025.04 | 39.93 | — | |
| T = sqrt(M)Backbone=ResNet50, Training Scenario=init2025.04 | 39.47 | — | |
| T = sqrt(M)Backbone=ResNet50, Training Scenario=FT2025.04 | 17.3 | — |