Research map / Continual learning / Online and replay-based
Learning to Unlearn for Robust Machine Unlearning
Mark He Huang, Lin Geng Foo, Jun Liu
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 LTU with only 30% of remaining set instead of full-access baselines | LTU · CIFAR-10 random sample unlearning 10% | TA gap to Retrain: SalUn 1.07 -> LTU 0.31 | Improved |
| increase accessible remaining set fraction from rho=0.1 to rho=0.3 (ablation) | LTU · CIFAR-10 random sample unlearning 10% | RA gap to Retrain: 7.67 -> 0.17 | Improved |
| use training-free unlearning (Fisher, SSD) on larger forgetting ratio | Fisher / SSD · CIFAR-10 random sample unlearning 50% | RA gap to Retrain: Fisher 77.32; SSD 10.53 | Hurt |
| add meta optimization (meta-tune, meta-test, meta-update) instead of conventional training feedback (ablation) | LTU · CIFAR-10 ResNet-18 | unlearning gaps | Improved |
| add Gradient Harmonization instead of naive gradient addition (ablation) | LTU · CIFAR-10 ResNet-18 | unlearning gaps | Improved |
| remove either forgetting or remembering feedback (ablation) | LTU · CIFAR-10 ResNet-18 | unlearning gaps | Improved |
| apply Amnesiac unlearning at 50% forget ratio | Amnesiac · CIFAR-10 random sample unlearning 50% | RA gap to Retrain: RA gap 24.79 | Hurt |
Closest papers by meaning
- When to Forget? Complexity Trade-offs in Machine Unlearning (ICML 2025 · similarity 0.92)
- What makes unlearning hard and what to do about it (NeurIPS 2024 · similarity 0.91)
- Efficient Utility-Preserving Machine Unlearning with Implicit Gradient Surgery (NeurIPS 2025 · similarity 0.90)
- Machine Unlearning under Overparameterization (NeurIPS 2025 · similarity 0.90)
- Towards Robust and Parameter-Efficient Knowledge Unlearning for LLMs (ICLR 2025 · similarity 0.90)
- Learning What to Forget: Improving LLM Unlearning via Learned Token-Level Importance (NeurIPS 2026 · similarity 0.90)
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.
