Research map / Continual learning / Online and replay-based
New Insights for the Stability-Plasticity Dilemma in Online Continual Learning
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 |
|---|---|---|---|
| Use MuFAN rather than DualNet | Online continual learning · CORe50 | ACC: 57.64±1.36 -> 67.30±1.57 | Helped |
| Use top-down aggregated multi-scale features rather than the standard feature map (ablation) | ER-Ring · Split CIFAR100 | ACC: 62.8±0.9 -> 69.3±1.6 | Helped |
| Use four encoder layers, including a shallow layer, rather than only the first layer (ablation) | MuFAN · CIFAR100 | ACC: 73.89±1.94 -> 75.86±0.35 | Helped |
| Add cross-channel mixing to cross-scale mixing (ablation) | MuFAN · CORe50 | ACC: 64.78±1.31 -> 67.30±1.57 | Helped |
| Use cosine similarity rather than L2 distance for structure-wise distillation (ablation) | ER-Ring · Split CIFAR100 | ACC: 64.5±0.8 -> 64.8±0.5 | Helped |
| Use consecutive-task relations rather than first-task relations in structure-wise distillation (ablation) | MuFAN · CORe50 | ACC: 66.54±1.22 -> 67.30±1.57 | Helped |
| Add multi-scale features (ablation) | MuFAN · Split miniIMN | ACC: 61.56±0.70 -> 72.08±0.42 | Helped |
| Add multi-scale features (ablation) | MuFAN · CORe50 | ACC: 48.71±0.91 -> 56.00±1.63 | Helped |
| Add structure-wise distillation (ablation) | MuFAN · Split miniIMN | FM: 3.95±0.67 -> 3.47±0.41 | Helped |
| Add structure-wise distillation (ablation) | MuFAN · CORe50 | FM: 7.12±0.81 -> 4.60±0.37 | Helped |
| Add SPN (ablation) | MuFAN · Split miniIMN | LA: 61.44±1.31 -> 65.15±0.99 | Helped |
| Add SPN (ablation) | MuFAN · CORe50 | LA: 52.55±0.98 -> 57.31±1.62 | Helped |
Closest papers by meaning
- Improving Plasticity in Online Continual Learning via Collaborative Learning (CVPR 2024 · similarity 0.90)
- Rethinking the Stability-Plasticity Trade-off in Continual Learning from an Architectural Perspective (ICML 2025 · similarity 0.90)
- Pareto Continual Learning: Preference-Conditioned Learning and Adaption for Dynamic Stability-Plasticity Trade-off (AAAI 2025 · similarity 0.88)
- Re-evaluating Continual Learning with Few-Shot Adaptation (NeurIPS 2026 · similarity 0.87)
- STAR: Stability-Inducing Weight Perturbation for Continual Learning (ICLR 2025 · similarity 0.87)
- Enhancing Online Continual Learning with Plug-and-Play State Space Model and Class-Conditional Mixture of Discretization (CVPR 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.