Research map / Continual learning / Online and replay-based
Flattening Sharpness for Dynamic Gradient Projection Memory Benefits Continual Learning
Danruo DENG, Guangyong Chen, Jianye HAO, Qiong Wang, Pheng-Ann Heng
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 |
|---|---|---|---|
| add flattening sharpness (adversarial weight perturbation) to GPM with dynamic bases (ablation) | DGPM · CIFAR-100 Split | ACC (%): 73.78 -> 74.33 | Improved |
| add flattening sharpness to DGPM (ablation) | DGPM · CIFAR-100 Superclass | ACC (%): 56.78 -> 58.81 | Improved |
| add dynamic soft weights on bases to FS-GPM (ablation) | FS-GPM · CIFAR-100 Split | ACC (%): 73.96 -> 74.33 | Improved |
| add dynamic soft weights on bases to FS-GPM (ablation) | FS-GPM · CIFAR-100 Superclass | ACC (%): 58.61 -> 58.81 | Improved |
| use adversarial weight perturbation instead of look-ahead descent perturbation (ablation) | FS-DGPM · CIFAR-100 Split | ACC (%): 73.74 -> 74.33 | Improved |
| add flattening sharpness to DGPM to reduce forgetting (ablation) | DGPM · CIFAR-100 Split | BWT (%): -3.67 -> -2.71 | Improved |
| add flattening sharpness to vanilla ER | ER · PMNIST | ACC (%): 86.16 -> 90.44 | Improved |
| use FS-DGPM instead of GPM | GPM · CIFAR-100 Split | BWT (%): -1.17 -> -2.71 | Hurt |
| use FS-DGPM instead of GPM | GPM · TinyImageNet | ACC (%): 67.39 -> 70.41 | Improved |
| use FS-DGPM instead of GPM | GPM · PMNIST | ACC (%): 74.54 -> 76.96 | Improved |
Closest papers by meaning
- Data Augmented Flatness-aware Gradient Projection for Continual Learning (ICCV 2023 · similarity 0.92)
- Make Continual Learning Stronger via C-Flat (NeurIPS 2024 · similarity 0.89)
- The Importance of Being Lazy: Scaling Limits of Continual Learning (ICML 2025 · similarity 0.88)
- Forget Sharpness: Perturbed Forgetting of Model Biases Within SAM Dynamics (ICML 2024 · similarity 0.88)
- PaLoRA: Paced Low-Rank Adaptation for Continual Learning (NeurIPS 2026 · similarity 0.88)
- STAR: Stability-Inducing Weight Perturbation for Continual Learning (ICLR 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.
