Research map / Continual learning / Continual fine-tuning of LLMs
pH-Strips for Selective Forgetting: A Blunt but Fast Diagnostic Baseline for Machine Unlearning
Chengyao Qian, Jing Wu, Trung Le, Dinh Phung, Mehrtash Harandi
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 |
|---|---|---|---|
| use training-free closed-form subspace projection instead of SalUn | MUpHT · CIFAR-100 class-wise forgetting | Avg. gap: 2.82 -> 0.91 | Improved |
| use MUpHT instead of SSD | MUpHT · CIFAR-100 class-wise forgetting | Avg. gap: 1.12 -> 0.91 | Improved |
| use MUpHT instead of SalUn on subclass forgetting | MUpHT · CIFAR-20 subclass forgetting | Avg. gap: 7.44 -> 14.15 | Hurt |
| add retain-aware Rayleigh variant (MUpHT†) (ablation) | MUpHT · CIFAR-20 unlearn class 0, accuracy of class 83 (same superclass) | accuracy on class 83: 0.00 -> 11.33 | Improved |
| add retain-aware variant for entangled classes (ablation) | MUpHT · CIFAR-20 unlearn class 0, accuracy of class 83 (same superclass) | accuracy on class 1 (dolphin): 98.22 -> 98.22 | No clear effect |
| use MUpHT instead of SalUn for SD class forgetting | MUpHT · Imagenette (SD v1.4) | FID: 1.22 -> 0.98 | Improved |
| use MUpHT instead of SalUn for SD class forgetting | MUpHT · Imagenette (SD v1.4) | UA: 99.82 -> 98.09 | Hurt |
| use fixed hyperparameters across classes with MUpHT | MUpHT · CIFAR-100 class-wise (fixed hyperparameters) | UA: SalUn class 1: 31.33; MUpHT: 99.78 | Improved |
Closest papers by meaning
- Label-Agnostic Forgetting: A Supervision-Free Unlearning in Deep Models (ICLR 2024 · similarity 0.90)
- When to Forget? Complexity Trade-offs in Machine Unlearning (ICML 2025 · similarity 0.90)
- Machine Unlearning for Image-to-Image Generative Models (ICLR 2024 · similarity 0.89)
- Single Image Unlearning: Efficient Machine Unlearning in Multimodal Large Language Models (NeurIPS 2024 · similarity 0.89)
- Fast Machine Unlearning without Retraining through Selective Synaptic Dampening (AAAI 2024 · similarity 0.89)
- The BatchNorm Illusion: Diagnosing Normalization Artifacts in Machine Unlearning Evaluation (NeurIPS 2026 · similarity 0.89)
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.
