Research map / Continual learning / Online and replay-based
Beyond Sharpness: A Flatness Decomposition Framework for Efficient Continual Learning
Yanan Chen, Tieliang Gong, Yunjiao Zhang, Wen Wen
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 in FLAD optimizer | Replay · CIFAR-100 (10 tasks) | Acc: 29.07 -> 31.74 | Improved |
| plug in FLAD optimizer | PODNet · CIFAR-100 (10 tasks) | Acc: 31.87 -> 34.85 | Improved |
| plug in FLAD optimizer | WA · CIFAR-100 (10 tasks) | Acc: 47.29 -> 48.35 | Improved |
| plug in FLAD optimizer | iCaRL · CIFAR-100 (10 tasks) | Acc: 30.05 -> 30.36 | No clear effect |
| plug in FLAD optimizer | FOSTER · CIFAR-100 (10 tasks) | Acc: 39.71 -> 39.13 | Hurt |
| plug in FLAD optimizer | MEMO · CIFAR-100 (10 tasks) | Acc: 53.27 -> 53.92 | Mixed |
| plug in FLAD optimizer | Replay · Tiny-ImageNet (8 tasks) | Acc: 19.09 -> 22.91 | Improved |
| plug in FLAD optimizer | FOSTER · CIFAR-10 (5 tasks) | AAA: 65.15 -> 75.08 | Improved |
| use stochastic-noise perturbation instead of full-gradient perturbation (ablation) | GAM · CIFAR-10/100 ResNet | average accuracy | Improved |
| set gamma = 0 (drop first-order term) (ablation) | FLAD · CIFAR-100 | accuracy | Improved |
| use decomposed (noise) gradient variant instead of original (ablation) | FLAD-0th / FLAD-1st · 6 CL methods | accuracy | Improved |
| apply FLAD in only 10-20% of epochs (ablation) | FLAD · CIFAR-100 Replay | accuracy / training time | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- FOSTER — FOSTER: Feature Boosting and Compression for Class-Incremental Learning (ECCV 2022) · better on 1 of 1 shared result
- iCaRL — iCaRL: Incremental Classifier and Representation Learning (CVPR 2017) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- C-Flat++: Towards a More Efficient and Powerful Framework for Continual Learning (IJCV 2026 · similarity 0.92)
- Make Continual Learning Stronger via C-Flat (NeurIPS 2024 · similarity 0.89)
- Model Sensitivity Aware Continual Learning (NeurIPS 2024 · similarity 0.87)
- A Statistical Theory of Regularization-Based Continual Learning (ICML 2024 · similarity 0.87)
- Optimal Rates in Continual Linear Regression via Increasing Regularization (NeurIPS 2025 · similarity 0.86)
- LoRanPAC: Low-rank Random Features and Pre-trained Models for Bridging Theory and Practice in Continual Learning (ICLR 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.
