Research map / Label-efficient and robust learning / Recognition and long-tail
Tackling Fake Forgetting through Uncertainty Quantification
Yingdan Shi, Sijia Liu, Kaize Ding, Ren Wang
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 the CPU conformal loss (lambda 0 to 0.5) (ablation) | FT · CIFAR-10 ResNet-18, 10% random forgetting | UA gap to RT: 4.8% -> 0.7% | Improved |
| add the CPU conformal loss (lambda 0 to 0.5) (ablation) | FT · Tiny ImageNet ViT, 10% random forgetting | UA gap to RT: 7.8% -> 0.9% | Improved |
| add the CPU conformal loss (lambda 0 to 0.5) (ablation) | FT · CIFAR-10 ResNet-18, 10% random forgetting | average gap across UA, RA, TA: 2.2 -> 1.47 | Improved |
| add the CPU conformal loss (lambda 0 to 0.5) (ablation) | FT · Tiny ImageNet ViT, 10% random forgetting | average gap across UA, RA, TA: 3.53 -> 1.63 | Improved |
| add the CPU conformal loss (lambda 0 to 1.0) (ablation) | FT · CIFAR-10 ResNet-18, 10% random forgetting | CR on forget data: 0.986 -> 0.835 | Improved |
| add the CPU conformal loss (lambda 0 to 1.0) (ablation) | RL · CIFAR-10 ResNet-18, 10% random forgetting | CR on forget data: 0.788 -> 0.629 | Improved |
| add the CPU conformal loss (lambda 0 to 1.0) (ablation) | RT · CIFAR-10 ResNet-18, 10% random forgetting | UA: 8.6% -> 17.7% | Improved |
| add the CPU conformal loss (lambda 0 to 1.0) (ablation) | RT · Tiny ImageNet ViT, 10% random forgetting | UA: 14.7% -> 35.7% | Improved |
| add the CPU conformal loss (lambda 0 to 1.0) (ablation) | RL · CIFAR-10 ResNet-18, 10% random forgetting | retain accuracy RA: 97.4% -> 95.3% | Hurt |
| add the CPU conformal loss (lambda 0 to 0.5), averaged over methods (ablation) | all methods · CIFAR-10 ResNet-18 and ViT | test accuracy TA: decreases only slightly by 1.0% and 0.57% | Hurt |
| add the CPU conformal loss (lambda 0 to 0.5), averaged over methods (ablation) | all methods · CIFAR-10 ResNet-18 and ViT | CR on test data: 0.03 drop on average | No clear effect |
| add the CPU conformal loss (lambda 0 to 0.5), averaged over methods (ablation) | all methods · CIFAR-10 ResNet-18 and ViT | CR on forget data: 0.08 average reduction | Improved |
Closest papers by meaning
- The BatchNorm Illusion: Diagnosing Normalization Artifacts in Machine Unlearning Evaluation (NeurIPS 2026 · similarity 0.89)
- Not All Wrong is Bad: Using Adversarial Examples for Unlearning (ICML 2025 · similarity 0.89)
- When to Forget? Complexity Trade-offs in Machine Unlearning (ICML 2025 · similarity 0.88)
- Forgetting Has Neighbors: Localized Collateral Forgetting in Machine Unlearning (NeurIPS 2026 · similarity 0.88)
- LoTUS: Large-Scale Machine Unlearning with a Taste of Uncertainty (CVPR 2025 · similarity 0.87)
- From Dormant to Deleted: Tamper-Resistant Unlearning Through Weight-Space Regularization (NeurIPS 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.
