Research map / Vision-language and multimodal / CLIP and zero-shot
VL-Eraser: Vacuum Distillation for Machine Unlearning in Vision-Language Models
Yili Wang, Lu Dai, Tairan Huang, Yijie Xu, Hui Xiong
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 |
|---|---|---|---|
| add vacuum-space projection to distillation (ablation) | VL-Eraser · MLLMU-Bench Visual-QA retain set | classification accuracy (%): 36.6 -> 43.4 | Improved |
| add distillation loss to cross-entropy (ablation) | VL-Eraser · MLLMU-Bench Visual-QA forget set | classification accuracy (%): 38.4 -> 26.2 (lower is better) | Improved |
| add cross-entropy loss to distillation (ablation) | VL-Eraser · MLLMU-Bench Visual-QA forget set | classification accuracy (%): 32.6 -> 26.2 (lower is better) | Improved |
| distill-then-delete instead of reverse training (vs GA/NPO) | LLaVA-1.5-7B · MLLMU-Bench Textual-QA forget set | classification accuracy (%): 43.6 (NPO) -> 25.0 | Improved |
| VL-Eraser vs MMUnlearner | Qwen2-VL-7B · MLLMU-Bench Visual-QA forget set | classification accuracy (%): 40.2 -> 33.6 | Improved |
| VL-Eraser utility on real celebrity set | LLaVA-1.5-7B · MLLMU-Bench Visual-QA real set | classification accuracy (%): 47.2 -> 45.6 | Mixed |
Closest papers by meaning
- Single Image Unlearning: Efficient Machine Unlearning in Multimodal Large Language Models (NeurIPS 2024 · similarity 0.91)
- Learning What to Forget: Improving LLM Unlearning via Learned Token-Level Importance (NeurIPS 2026 · similarity 0.89)
- Selective Forgetting: Advancing Machine Unlearning Techniques and Evaluation in Language Models (AAAI 2025 · similarity 0.89)
- AUVIC: Adversarial Unlearning of Visual Concepts for Multi-modal Large Language Models (AAAI 2026 · similarity 0.89)
- Decoupled Distillation to Erase: A General Unlearning Method for Any Class-centric Tasks (CVPR 2025 · similarity 0.89)
- SineProject: Machine Unlearning for Stable Vision-Language Alignment (CVPR 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.
