Research map / Continual learning / Class-incremental
RMM: Reinforced Memory Management for Class-Incremental Learning
Yaoyao Liu, Bernt Schiele, Qianru Sun
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Plug RMM into POD-AANets | POD-AANets · CIFAR-100, N=5 | average accuracy (%): 66.61 -> 68.86 | Improved |
| Plug RMM into POD-AANets | POD-AANets · CIFAR-100, N=25 | average accuracy (%): 62.63 -> 66.21 | Improved |
| Plug RMM into POD-AANets | POD-AANets · ImageNet-Subset, N=25 | average accuracy (%): 72.18 -> 76.54 | Improved |
| Plug RMM into POD-AANets | POD-AANets · ImageNet-Full, N=25 | average accuracy (%): 62.03 -> 63.93 | Improved |
| Plug RMM into LUCIR-AANets | LUCIR-AANets · ImageNet-Subset, N=25 | average accuracy (%): 68.07 -> 72.30 | Improved |
| Plug RMM into LUCIR-AANets | LUCIR-AANets · CIFAR-100, N=5 | average accuracy (%): 66.88 -> 68.42 | Improved |
| Increase number of incremental phases N from 5 to 25 (RMM gain grows) | POD-AANets w/ RMM · CIFAR-100 and ImageNet-Subset | gain over baseline: 2.3% -> 3.6% (CIFAR-100), 2.1% -> 4.4% (ImageNet-Subset) | Improved |
| Add Level-1 RL policy for old/new memory split (ablation) | POD-AANets · CIFAR-100, N=5 | average accuracy (%): 66.61 -> 67.92 | Improved |
| Add Level-2 class-specific policy (two-level instead of one-level) (ablation) | RMM · CIFAR-100, N=5 | average accuracy (%): 67.92 -> 68.86 | Improved |
| Add Level-2 policy (two-level vs one-level) (ablation) | RMM · ImageNet-Subset, N=25 | average accuracy (%): 75.47 -> 76.54 | Improved |
| Transfer policy from another dataset instead of learning on target (ablation) | RMM · CIFAR-100, N=5 | average accuracy (%): 68.86 -> 68.62 | No clear effect |
| Train RMM on target task with all data (UpperBound RL) instead of pseudo tasks (ablation) | RMM · CIFAR-100, N=5 | average accuracy (%): 68.86 -> 70.00 | No clear effect |
Closest papers by meaning
- A Model or 603 Exemplars: Towards Memory-Efficient Class-Incremental Learning (ICLR 2023 · similarity 0.91)
- Autoencoder-Based Hybrid Replay for Class-Incremental Learning (ICML 2025 · similarity 0.89)
- Class Incremental Learning via Likelihood Ratio Based Task Prediction (ICLR 2024 · similarity 0.87)
- Long-Tail Class Incremental Learning via Independent Sub-prototype Construction (CVPR 2024 · similarity 0.87)
- Knowledge Memorization and Rumination for Pre-trained Model-based Class-Incremental Learning (CVPR 2025 · similarity 0.87)
- Tensor Decomposition Based Memory-Efficient Incremental Learning (ICML 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
