Research map / Adversarial robustness and security / Backdoor and poisoning
UMD: Unsupervised Model Detection for X2X Backdoor Attacks
Zhen Xiang, Zidi Xiong, Bo Li
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 UMD instead of K-Arm | X2X backdoor detection · CIFAR-10 X2X attacks (Avg) | MIA: 0.74 -> 0.91 | Improved |
| use UMD instead of K-Arm | X2X backdoor detection · GTSRB X2X attacks (Avg) | MIA: 0.87 -> 0.91 | Improved |
| use UMD instead of K-Arm | X2X backdoor detection · Imagenette X2X attacks (Avg) | MIA: 0.77 -> 0.85 | Improved |
| omit TR and apply MAD directly to all class-pair triggers (ablation) | UMD · CIFAR-10 2to2, 5to5, 8to8, A2Ar (Pert trigger) | Overall MIA: 0.93 -> 0.50 | Hurt |
| use TR without the proposed clustering technique (ablation) | UMD · CIFAR-10 2to2, 5to5, 8to8, A2Ar (Pert trigger) | Overall MIA: 0.93 -> 0.70 | Hurt |
| use the trigger reverse-engineering procedure with fewer than the default number of images (ablation) | UMD · CIFAR-10 | MIA | Improved |
| vary the targeted misclassification fraction below 1 (ablation) | UMD trigger reverse-engineering · CIFAR-10 | MIA | No clear effect |
| use UMD with WaNet or Blended triggers | CIFAR-10 X2X backdoor detection · 2to2, 5to5, 8to8, A2Ar | MIA: WaNet: 0.80, 0.80, 1.0, 0.90; Blended: 0.90, 0.90, 0.90, 0.90 | Mixed |
| use UMD for adaptive backdoor detection | CIFAR-10 · ISSBA, CLA, N2N, O2O | MIA: 1.0, 0.60, 1.0, 0.80 | Mixed |
Closest papers by meaning
- UMA: Facilitating Backdoor Scanning via Unlearning-Based Model Ablation (AAAI 2024 · similarity 0.89)
- Lie Detector: Unified Backdoor Detection via Cross-Examination Framework (NeurIPS 2025 · similarity 0.87)
- LOTUS: Evasive and Resilient Backdoor Attacks through Sub-Partitioning (CVPR 2024 · similarity 0.86)
- UFID: A Unified Framework for Black-box Input-level Backdoor Detection on Diffusion Models (AAAI 2025 · similarity 0.85)
- Universal Backdoor Attacks (ICLR 2024 · similarity 0.85)
- Inspecting Prediction Confidence for Detecting Black-Box Backdoor Attacks (AAAI 2024 · similarity 0.85)
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.
