Class-incremental learning on 5-datasets (test)
93.93Average AccuracyUpper bound
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Upper bound2021.12 | 93.93 | — | — | |
| L2P-RBuffer size=10/class2021.12 | 88.95 | 4.92 | — | |
| Co2LBuffer size=10/class2021.12 | 86.05 | 12.28 | — | |
| L2P-RBuffer size=5/class2021.12 | 85.56 | 4.22 | — | |
| BiCBuffer size=10/class2021.12 | 85.53 | 10.27 | — | |
| DER++Buffer size=10/class2021.12 | 84.88 | 10.46 | — | |
| ERBuffer size=10/class2021.12 | 84.26 | 12.85 | — | |
| Co2LBuffer size=5/class2021.12 | 82.25 | 17.52 | — | |
| L2PBuffer size=02021.12 | 81.14 | 4.64 | — | |
| DER++Buffer size=5/class2021.12 | 80.81 | 14.38 | — | |
| ERBuffer size=5/class2021.12 | 80.32 | 15.69 | — | |
| BiCBuffer size=5/class2021.12 | 78.74 | 21.15 | — | |
| GDumbBuffer size=10/class2021.12 | 70.76 | — | — | |
| GDumbBuffer size=5/class2021.12 | 56.99 | — | — | |
| EWCBuffer size=02021.12 | 50.93 | 34.94 | — | |
| LwFBuffer size=02021.12 | 47.91 | 38.01 | — | |
| FT-seq-frozenBuffer size=02021.12 | 39.49 | 42.62 | — | |
| FT-seqBuffer size=02021.12 | 20.12 | 94.63 | — | |
| CODA-PromptPre-training=Sup-21K2024.05 | — | — | 64.18 | |
| CODA-PromptPre-training=iBOT-21K2024.05 | — | — | 51.65 | |
| DualPromptPre-training=Sup-21K2024.05 | — | — | 77.91 | |
| DualPromptPre-training=iBOT-21K2024.05 | — | — | 68.03 | |
| HiDe-PromptPre-training=Sup-21K2024.05 | — | — | 93.83 | |
| HiDe-PromptPre-training=iBOT-21K2024.05 | — | — | 94.88 | |
| L2PPre-training=Sup-21K2024.05 | — | — | 81.84 | |
| L2PPre-training=iBOT-21K2024.05 | — | — | 82.25 | |
| NoRGaPre-training=Sup-21K2024.05 | — | — | 94.16 | |
| NoRGaPre-training=iBOT-21K2024.05 | — | — | 94.92 | |
| S-PromptPre-training=Sup-21K2024.05 | — | — | 86.06 | |
| S-PromptPre-training=iBOT-21K2024.05 | — | — | 77.2 |