Research map / Adversarial robustness and security / Attacks on text and graphs
Unlearning Evaluation through Subset Statistical Independence
Chenhao Zhang, Muxing Li, Feng Liu, Weitong Chen, Miao Xu
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 SDE (split-half HSIC) instead of MMD | in/out-of-training detection · CIFAR-10 ResNet-18, R=5%, |S|=400 | F1: 0.63 -> 0.87 | Improved |
| use SDE instead of Wasserstein distance | in/out-of-training detection · CIFAR-10 ResNet-18, R=10%, |S|=2000 | F1: 0.98 -> 0.97 | No clear effect |
| increase target subset size from 400 to 2000 (ablation) | SDE · CIFAR-10 ResNet-18, R=5% | F1: 0.87 -> 0.99 | Improved |
| increase forgetting ratio from 5% to 20% (ablation) | SDE · SVHN AllCNN, |S|=400 | F1: SV-AllCNN |S|=400: 0.62 -> 0.50 (R=10%) -> 0.72 (R=20%) | Mixed |
| train retrained model longer (10% -> 40% of epochs) (ablation) | SDE · CIFAR-10 ResNet-18 | F1: 0.69 -> 0.95 | Improved |
| evaluate on deeper layers instead of early layers (ablation) | SDE · layer-wise, R=10%, |S|=1000 | F1 | Improved |
| choose kernel bandwidth sigma = sqrt(dim) vs extreme values (ablation) | SDE · kernel bandwidth | F1 | Improved |
| use SDE OTR to judge Unroll vs ASR/accuracy | Unroll unlearning · CIFAR10-ResNet18, R=10% | OTR (%): Retrain 94.00 vs Unroll 4.40 | Hurt |
| apply RandLabel unlearning | RandLabel · CIFAR10-ResNet18, R=10% | OTR (%): 83.20 vs Retrain 94.00 | Improved |
Closest papers by meaning
- When to Forget? Complexity Trade-offs in Machine Unlearning (ICML 2025 · similarity 0.89)
- What makes unlearning hard and what to do about it (NeurIPS 2024 · similarity 0.88)
- Certified Unlearning for Neural Networks (ICML 2025 · similarity 0.88)
- Efficient Source-free Unlearning via Energy-Guided Data Synthesis and Discrimination-Aware Multitask Optimization (ICML 2025 · similarity 0.88)
- LoTUS: Large-Scale Machine Unlearning with a Taste of Uncertainty (CVPR 2025 · similarity 0.88)
- Machine Unlearning under Overparameterization (NeurIPS 2025 · 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.
