Text-to-Image Membership Inference on CelebA-Dialog
96ASROurs
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| OursBackbone Model=Stable Diffusion v2.1, Knowledge Setting=Partial Knowledge2026.07 | 96 | 99 | 59 | |
| CLiD-MIABackbone Model=Stable Diffusion v2.1, Knowledge Setting=Partial Knowledge2026.07 | 92 | 96 | 48 | |
| unified membership inference frameworkKnowledge Setting=Zero-Knowledge, Base Model=Stable Diffusion v2.1, Embedding Extractor=BLIP2026.07 | 89 | 94 | 53 | |
| Score-based MIABackbone Model=Stable Diffusion v2.1, Knowledge Setting=Partial Knowledge2026.07 | 88 | 95 | 44 | |
| CLiDKnowledge Setting=Zero-Knowledge, Base Model=Stable Diffusion v2.1, Embedding Extractor=BLIP2026.07 | 56 | 64 | 20 | |
| ScoreKnowledge Setting=Zero-Knowledge, Base Model=Stable Diffusion v2.1, Embedding Extractor=BLIP2026.07 | 52 | 52 | 21 |