Exemplar-Free Class-Incremental Learning on TinyImageNet T=20 (test)
31.3AlastAdaGauss
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| AdaGaussFeature Extractor=trained from scratch2024.09 | 31.3 | 45.1 | — | |
| EFCFeature Extractor=trained from scratch2024.09 | 28.7 | 42.1 | — | |
| DS-ALFeature Extractor=trained from scratch2024.09 | 26.5 | 41.6 | — | |
| FeTrILFeature Extractor=trained from scratch2024.09 | 25.7 | 39.5 | — | |
| FeCAMFeature Extractor=trained from scratch2024.09 | 25.2 | 38.3 | — | |
| IL2AFeature Extractor=trained from scratch2024.09 | 19.8 | 35.5 | — | |
| PASSFeature Extractor=trained from scratch2024.09 | 18.7 | 32 | — | |
| SSREFeature Extractor=trained from scratch2024.09 | 17.3 | 30.6 | — | |
| LwFFeature Extractor=trained from scratch2024.09 | 15 | 32.9 | — | |
| EWCFeature Extractor=trained from scratch2024.09 | 11.3 | 26.8 | — | |
| ACILTotal parameters (M)=17.10M, Learnable parameters (M)=12.90M, Number of tasks (T)=202026.06 | — | — | 52.89 | |
| AdaGaussTotal parameters (M)=11.20M, Learnable parameters (M)=11.20M, Number of tasks (T)=202026.06 | — | — | 200.17 | |
| CIRCLETotal parameters (M)=11.46M, Learnable parameters (M)=10.04M, Number of tasks (T)=202026.06 | — | — | 11.51 | |
| EFC++Total parameters (M)=11.27M, Learnable parameters (M)=11.27M, Number of tasks (T)=202026.06 | — | — | 310.04 | |
| FeCAMTotal parameters (M)=11.27M, Learnable parameters (M)=11.27M, Number of tasks (T)=202026.06 | — | — | 26.11 | |
| FeTrILTotal parameters (M)=11.37M, Learnable parameters (M)=11.37M, Number of tasks (T)=202026.06 | — | — | 560.09 |