Research map / Vision-language and multimodal / CLIP and zero-shot
Your Prompts Are Not Safe: Output-Free Membership Inference via Prompt Vectors in Vision-Language Tuning
Yuran Bian, Xiaohan Zhang, Zhiyuan Yu, Changqing Li, Li Pan
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 contrastive learning-based feature projector (ablation) | PIPRA · MIA on 9 datasets (avg) | AUC and accuracy: avg about +7.3% AUC, +6.7% accuracy | Improved |
| add contrastive projector (ablation) | PIPRA · Caltech101 | AUC: 88.0 -> 92.4 | Improved |
| add contrastive projector (ablation) | PIPRA · Food101 | AUC: 74.6 -> 86.0 | Improved |
| add contrastive projector (ablation) | PIPRA · Flowers102 | accuracy: 75.8 -> 76.2 | No clear effect |
| use mismatched surrogate model instead of matching target model (ablation) | PIPRA · Caltech101 | accuracy: ViT-B/32 100.00 -> RN50 surrogate 75.07 | Hurt |
| use a shadow dataset from a different domain (ablation) | PIPRA · Caltech101 | accuracy | Mixed |
| use PIPRA (output-free, prompt only) instead of output-dependent MIAs | PIPRA · nine datasets | AUC: 79.7 -> 87.6 (avg vs Gradient-based) | Improved |
Closest papers by meaning
- VLMGuard: Bootstrapping Malicious Prompt Detectors from Unlabeled Vision-Language Prompts in the Wild (NeurIPS 2026 · similarity 0.88)
- Membership Inference Attacks against Large Vision-Language Models (NeurIPS 2024 · similarity 0.87)
- Private Attribute Inference from Images with Vision-Language Models (NeurIPS 2024 · similarity 0.87)
- MultiPriv: Benchmarking Individual-Level Privacy Reasoning in Vision-Language Models (ICML 2026 · similarity 0.87)
- Watch Out Your Album! On the Inadvertent Privacy Memorization in Multi-Modal Large Language Models (ICML 2025 · similarity 0.87)
- DP-OPT: Make Large Language Model Your Privacy-Preserving Prompt Engineer (ICLR 2024 · similarity 0.86)
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.
