Incremental Learning 벤치마크
Incremental Learning on ImageNet - 500 classes + 5 steps of 100 classes
Average Incremental Accuracy
- 2020-04-28 — PODNet: Average Incremental Accuracy 66.95
- 2021-02-18 — CCIL-SD: Average Incremental Accuracy 68.04
- 2023-01-14 — RMM (ResNet-18): Average Incremental Accuracy 69.21
| Rank | Model | Average Incremental Accuracy | Final Accuracy | Extra Training Data | Paper | Code | Year |
|---|---|---|---|---|---|---|---|
| 1 | RMM (ResNet-18) | 69.21 | – | RMM: Reinforced Memory Management for Class-Incremental Learning | g-u-n/pycil · aimagelab/mammoth · yaoyaoliu/rmm · +1 | 2023 | |
| 2 | CCIL-SD | 68.04 | – | Essentials for Class Incremental Learning | sud0301/essentials_for_CIL | 2021 | |
| 3 | PODNet | 66.95 | – | PODNet: Pooled Outputs Distillation for Small-Tasks Incremental Learning | g-u-n/pycil · arthurdouillard/incremental_learning.pytorch | 2020 | |
| 4 | PPCA-CLIP | – | 71.25 | ✓ | Scalable Learning with Incremental Probabilistic PCA | barbua/PPCA | 2022 |