Exemplar-Free Class-Incremental Learning on Imagenet Subset T=20 (test)
42.6AlastAdaGauss
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| AdaGaussFeature Extractor=trained from scratch2024.09 | 42.6 | 57.4 | — | |
| DS-ALFeature Extractor=trained from scratch2024.09 | 36.7 | 48.5 | — | |
| EFCFeature Extractor=trained from scratch2024.09 | 35.8 | 49.9 | — | |
| FeCAMFeature Extractor=trained from scratch2024.09 | 29 | 44.6 | — | |
| FeTrILFeature Extractor=trained from scratch2024.09 | 26.6 | 42.4 | — | |
| LwFFeature Extractor=trained from scratch2024.09 | 18.6 | 40.2 | — | |
| IL2AFeature Extractor=trained from scratch2024.09 | 17.5 | 34.9 | — | |
| SSREFeature Extractor=trained from scratch2024.09 | 16.3 | 31.2 | — | |
| PASSFeature Extractor=trained from scratch2024.09 | 14.4 | 31.7 | — | |
| EWCFeature Extractor=trained from scratch2024.09 | 12.8 | 27 | — | |
| ACILTotal parameters (M)=21.20M, Learnable parameters (M)=12.81M, Number of tasks (T)=202026.06 | — | — | 87.26 | |
| AdaGaussTotal parameters (M)=11.21M, Learnable parameters (M)=11.21M, Number of tasks (T)=202026.06 | — | — | 858.5 | |
| CIRCLETotal parameters (M)=11.05M, Learnable parameters (M)=6.48M, Number of tasks (T)=202026.06 | — | — | 39.22 | |
| EFC++Total parameters (M)=11.23M, Learnable parameters (M)=11.23M, Number of tasks (T)=202026.06 | — | — | 390.31 | |
| FeCAMTotal parameters (M)=11.22M, Learnable parameters (M)=11.22M, Number of tasks (T)=202026.06 | — | — | 107.07 | |
| FeTrILTotal parameters (M)=11.32M, Learnable parameters (M)=11.32M, Number of tasks (T)=202026.06 | — | — | 288.7 |