Continual Learning on ImageNet tiny (BWT, Accuracy)
70BWTEWC
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| EWCepsilon=0.1, Scenario=Uniform2023.11 | 70 | 52.6 | — | — | — | |
| ER-ACEAttack Type=TV2026.06 | 52.4 | 10.5 | 5 | 4 | 1.9 | |
| ERAttack Type=TV2026.06 | 45.9 | 6.5 | 11 | 12 | 1.2 | |
| ER-ACEAttack Type=KL2026.06 | 45.1 | 13.1 | 2.8 | 1 | 0.7 | |
| SCRAttack Type=TV2026.06 | 42.8 | 9.9 | 5 | 4 | 2 | |
| ERAttack Type=KL2026.06 | 39.3 | 8.2 | 4 | 3 | 0.6 | |
| DER++Attack Type=TV2026.06 | 38.4 | 11.8 | 34 | 36 | 5.1 | |
| SCRAttack Type=KL2026.06 | 33.7 | 11.4 | 12 | 2 | 1.1 | |
| DER++Attack Type=KL2026.06 | 30.1 | 10.2 | 11 | 2 | 1.2 | |
| ERAttack Type=None2026.06 | 21.3 | 17 | — | — | — | |
| SCRAttack Type=None2026.06 | 20.2 | 22.5 | — | — | — | |
| DER++Attack Type=None2026.06 | 19.5 | 21.8 | — | — | — | |
| ER-ACEAttack Type=None2026.06 | 15.7 | 23.5 | — | — | — | |
| EWCScenario=Clean2023.11 | -40 | 53 | — | — | — | |
| MASScenario=Clean2023.11 | -110 | 59.4 | — | — | — | |
| AFECScenario=Clean2023.11 | -130 | 51.6 | — | — | — | |
| ANCLepsilon=0.1, Scenario=Uniform2023.11 | -150 | 72.2 | — | — | — | |
| ANCLScenario=Clean2023.11 | -170 | 73.4 | — | — | — | |
| MASepsilon=0.1, Scenario=Uniform2023.11 | -180 | 60 | — | — | — | |
| AFECepsilon=0.1, Scenario=Uniform2023.11 | -280 | 51 | — | — | — | |
| ANCLepsilon=0.3, Scenario=Uniform2023.11 | -340 | 51.8 | — | — | — | |
| ANCLepsilon=0.1, Scenario=Cautious (BrainWash)2023.11 | -430 | 46 | — | — | — | |
| MASepsilon=0.1, Scenario=Cautious (BrainWash)2023.11 | -500 | 54.2 | — | — | — | |
| EWCepsilon=0.1, Scenario=Cautious (BrainWash)2023.11 | -580 | 39.2 | — | — | — | |
| ANCLepsilon=0.1, Scenario=Reckless (BrainWash)2023.11 | -670 | 32.6 | — | — | — | |
| EWCepsilon=0.1, Scenario=Reckless (BrainWash)2023.11 | -710 | 35.6 | — | — | — | |
| EWCepsilon=0.3, Scenario=Uniform2023.11 | -730 | 44.6 | — | — | — | |
| MASepsilon=0.3, Scenario=Uniform2023.11 | -800 | 46.8 | — | — | — | |
| AFECepsilon=0.1, Scenario=Cautious (BrainWash)2023.11 | -1,030 | 34.2 | — | — | — | |
| ANCLepsilon=0.3, Scenario=Cautious (BrainWash)2023.11 | -1,070 | 37.2 | — | — | — | |
| MASepsilon=0.1, Scenario=Reckless (BrainWash)2023.11 | -1,290 | 31.2 | — | — | — | |
| AFECepsilon=0.3, Scenario=Uniform2023.11 | -1,350 | 34 | — | — | — | |
| RWALKScenario=Clean2023.11 | -1,410 | 40.6 | — | — | — | |
| RWALKepsilon=0.1, Scenario=Uniform2023.11 | -1,460 | 39.6 | — | — | — | |
| AFECepsilon=0.1, Scenario=Reckless (BrainWash)2023.11 | -1,540 | 30.4 | — | — | — | |
| ANCLepsilon=0.3, Scenario=Reckless (BrainWash)2023.11 | -1,600 | 22.8 | — | — | — | |
| MASepsilon=0.3, Scenario=Cautious (BrainWash)2023.11 | -2,550 | 37.8 | — | — | — | |
| RWALKepsilon=0.1, Scenario=Reckless (BrainWash)2023.11 | -2,570 | 44.2 | — | — | — | |
| EWCepsilon=0.3, Scenario=Reckless (BrainWash)2023.11 | -2,590 | 16.2 | — | — | — | |
| RWALKepsilon=0.1, Scenario=Cautious (BrainWash)2023.11 | -2,590 | 38.8 | — | — | — | |
| AFECepsilon=0.3, Scenario=Cautious (BrainWash)2023.11 | -2,620 | 15 | — | — | — | |
| AFECepsilon=0.3, Scenario=Reckless (BrainWash)2023.11 | -2,780 | 14.8 | — | — | — | |
| MASepsilon=0.3, Scenario=Reckless (BrainWash)2023.11 | -2,800 | 22.6 | — | — | — | |
| EWCepsilon=0.3, Scenario=Cautious (BrainWash)2023.11 | -2,840 | 11.6 | — | — | — | |
| RWALKepsilon=0.3, Scenario=Uniform2023.11 | -2,980 | 21.2 | — | — | — | |
| RWALKepsilon=0.3, Scenario=Reckless (BrainWash)2023.11 | -3,330 | 25.8 | — | — | — | |
| RWALKepsilon=0.3, Scenario=Cautious (BrainWash)2023.11 | -3,350 | 26.6 | — | — | — |