Research map / Continual learning / Online and replay-based
Learning to Learn without Forgetting by Maximizing Transfer and Minimizing Interference
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Add Reptile-style meta-learning to experience replay | Experience replay (ER) · MNIST Rotations and MNIST Permutations | RA: 87.58 -> 89.56; 83.02 -> 85.50 | Helped |
| Use MER instead of GEM | Continual learning system · MNIST Rotations, buffer 500 and 200 | RA: 74.88 -> 82.08; 67.38 -> 77.42 | Helped |
| Use MER instead of GEM | Continual learning system · MNIST Permutations, buffer 500 and 200 | RA: 69.26 -> 77.50; 55.42 -> 73.46 | Helped |
| Use MER instead of GEM | Continual learning system · Many Permutations | RA: 56.76 -> 62.52; 32.14 -> 47.40 | Helped |
| Use MER instead of GEM | Continual learning system · Omniglot | RA: 18.03 -> 75.23 | Helped |
| Add meta-learning to ER (ablation) | Experience replay (ER) · MNIST lifelong learning benchmarks | RA | Helped |
| Use a smaller replay buffer (ablation) | MER · MNIST Rotations and MNIST Permutations | RA: 89.56 -> 82.08 -> 77.42; 85.50 -> 77.50 -> 73.46 | Hurt |
| Increase the importance of the current example in MER (ablation) | MER variants · MNIST lifelong learning benchmarks | RA | Helped |
| Use either alternative MER sampling/update variant (ablation) | MER · MNIST lifelong learning benchmarks | RA | Helped |
| Use Adam or RMSProp instead of SGD (ablation) | MER and experience replay · Continual learning, especially smaller buffers | Generalization | Hurt |
| Use ER with reservoir sampling rather than GEM's buffer strategy (ablation) | GEM-style episodic replay · Omniglot | RA | Helped |
Closest papers by meaning
- Online Continual Learning with Maximal Interfered Retrieval (NeurIPS 2019 · similarity 0.91)
- Layerwise Proximal Replay: A Proximal Point Method for Online Continual Learning (ICML 2024 · similarity 0.90)
- Learning to Continually Learn with the Bayesian Principle (ICML 2024 · similarity 0.89)
- Trust Region Continual Learning as an Implicit Meta-Learner (NeurIPS 2026 · similarity 0.89)
- Mitigating Catastrophic Forgetting in Online Continual Learning by Modeling Previous Task Interrelations via Pareto Optimization (ICML 2024 · similarity 0.89)
- Pareto Continual Learning: Preference-Conditioned Learning and Adaption for Dynamic Stability-Plasticity Trade-off (AAAI 2025 · similarity 0.88)
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.