Research map / Continual learning / Online and replay-based
Mitigating Forgetting in Online Continual Learning with Neuron Calibration
Haiyan Yin, Peng Yang, Ping Li
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 |
|---|---|---|---|
| use NCCL neuron calibration with EWC and interleaved optimization instead of CTN | online continual learning · Split CIFAR | ACC: 68.62 -> 74.39 | Improved |
| NCCL instead of CTN | online continual learning · CORe50 | FM: 5.71 -> 2.22 | Improved |
| NCCL instead of CTN | online continual learning · Split miniImageNet | ACC: 65.25 -> 69.49 | Improved |
| NCCL instead of Independent per-task models | online continual learning · pMNIST | ACC: 83.50 -> 83.47 | No clear effect |
| train calibrator with memory rehearsal and update base parameters (NCCL) instead of CCLL with frozen base (ablation) | calibration-based continual learning · Split CIFAR | ACC: 64.77 -> 74.39 | Improved |
| update base parameters in addition to calibrator in CCLL (ablation) | CCLL · Split miniImageNet | ACC: 60.56 -> 68.80 | Improved |
| increase replay memory per task from 50 to 250 | NCCL, CTN, MIR, GEM · Split CIFAR/miniImageNet | ACC | Improved |
Closest papers by meaning
- An Effective Dynamic Gradient Calibration Method for Continual Learning (ICML 2024 · similarity 0.90)
- When to Learn and When to Stop: Quitting at the Optimal Time (Student Abstract) (AAAI 2025 · similarity 0.88)
- STAR: Stability-Inducing Weight Perturbation for Continual Learning (ICLR 2025 · similarity 0.87)
- Do Your Best and Get Enough Rest for Continual Learning (CVPR 2025 · similarity 0.87)
- Learning to Continually Learn with the Bayesian Principle (ICML 2024 · similarity 0.87)
- Predicting the Susceptibility of Examples to Catastrophic Forgetting (ICML 2025 · similarity 0.86)
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.
