Class-incremental learning on TinyImageNet 100 base classes (5-step)
81.14Average AccuracyiGSP
Evaluation results
| Method | Links | ||
|---|---|---|---|
| iGSP2026.05 | 81.14 | 77.34 | |
| MoE-Adapter(++)2026.05 | 81.12 | 76.81 | |
| ZSCL2026.05 | 80.27 | 73.57 | |
| LwF-VR2026.05 | 77.56 | 70.89 | |
| iCaRL2026.05 | 77.02 | 70.39 | |
| CLIPEvaluation protocol=zero-shot2026.05 | 69.62 | 65.3 | |
| Fine-tune2026.05 | 61.54 | 46.66 | |
| LwF2026.05 | 60.97 | 48.77 | |
| DyTox+2026.05 | 55.58 | 47.23 | |
| UCIR2026.05 | 50.3 | 39.42 | |
| PASS2026.05 | 49.54 | 41.64 | |
| EEIL2026.05 | 47.17 | 35.12 | |
| MUC2026.05 | 32.23 | 19.2 | |
| EWC2026.05 | 19.01 | 6 |