Research map / Label-efficient and robust learning / Recognition and long-tail
FaLW: A Forgetting-aware Loss Reweighting for Long-tailed Unlearning
Liheng Yu, Zhe Zhao, Yuxuan Wang, Pengkun Wang, Xiaofeng Cao, Binwu Wang, Yang Wang
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 |
|---|---|---|---|
| apply FaLW reweighting to unlearning | SalUn · CIFAR-100 gamma=2, 30% forget | Avg. Gap: 2.29 -> 1.30 | Improved |
| apply FaLW reweighting | SalUn · CIFAR-100 gamma=0 (balanced), 30% forget | Avg. Gap: 1.55 -> 0.68 | Improved |
| apply FaLW | best baselines · Tiny-ImageNet VGG-16 gamma=1 | Avg. Gap: 2.45 -> 0.35 | Improved |
| apply FaLW | best baselines · CIFAR-10 VGG-16 gamma=1/2 40% | Avg. Gap: 2.14 -> 0.40 | Improved |
| add class-aware balance factor (ablation) | FaLW · CIFAR-100 gamma=2 | tail-class FA gap: -18.76 -> -13.04 | Improved |
| add class-aware balance factor (ablation) | FaLW · CIFAR-100 gamma=1.5 | tail-class FA gap: -12.19 -> -9.46 | Improved |
| add class-aware balance factor (ablation) | FaLW · CIFAR-100 gamma=2 | head-class FA gap: 1.22 -> 1.33 | Hurt |
| increase forget-set imbalance gamma | SalUn · CIFAR-100 30% forget | forget accuracy vs Retrain: FA below Retrain at low gamma (e.g. 56.93 vs 59.11) to above at high gamma (4.79 vs 1.82) | Hurt |
| apply FaLW across forget ratios 10%-40% | SalUn · CIFAR-100 gamma=1/4 | Avg. Gap: 20%: 2.44 -> 0.33 | Improved |
| apply FaLW at 30% forget | SFRon · CIFAR-100 gamma=1/4 30% | Avg. Gap: 1.71 -> 0.91 | Improved |
Closest papers by meaning
- When to Forget? Complexity Trade-offs in Machine Unlearning (ICML 2025 · similarity 0.90)
- From Dormant to Deleted: Tamper-Resistant Unlearning Through Weight-Space Regularization (NeurIPS 2025 · similarity 0.90)
- What makes unlearning hard and what to do about it (NeurIPS 2024 · similarity 0.89)
- The Utility and Complexity of In- and Out-of-Distribution Machine Unlearning (ICLR 2025 · similarity 0.89)
- Unlearning or Obfuscating? Jogging the Memory of Unlearned LLMs via Benign Relearning (ICLR 2025 · similarity 0.89)
- Fast Machine Unlearning without Retraining through Selective Synaptic Dampening (AAAI 2024 · 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.
