Research map / Adversarial robustness and security / Backdoor and poisoning
CBD: A Certified Backdoor Detector Based on Local Dominant Probability
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 LDP with adjustable conformal prediction for model-level detection | Backdoor detection · GTSRB, SVHN, CIFAR-10; BadNet, CB, Blend | TPR (%): CBD0.2: 90, 95, 95; 95, 100, 90; 75, 100, 60 | Mixed |
| Use CBD rather than the strongest listed baseline for CB detection | Backdoor detection · SVHN CB | TPR (%): 70 -> 100 | Improved |
| Use CBD rather than the strongest listed baseline for CB detection | Backdoor detection · CIFAR-10 CB | TPR (%): 80 -> 100 | Improved |
| Use conformal CBD rather than supervised CBDsup | Backdoor detection · GTSRB, SVHN, CIFAR-10 pooled table average | TPR average (%): 89.4 -> 88.9 | Hurt |
| Increase assumed outlier ratio from 0 to 0.2 (ablation) | CBD · GTSRB BadNet | TPR (%): 75 -> 90 | Improved |
| Increase assumed outlier ratio from 0.1 to 0.2 (ablation) | CBD · GTSRB BadNet | TPR (%): 90 -> 90 | No clear effect |
| Reduce shadow model count from 100 to 10 (ablation) | CBD · GTSRB BadNet | TPR (%): 90 -> 95 | No clear effect |
| Reduce shadow model count from 100 to 25 (ablation) | CBD · GTSRB BadNet | TPR (%): 90 -> 95 | No clear effect |
| Reduce shadow model count from 100 to 50 (ablation) | CBD · GTSRB BadNet | TPR (%): 90 -> 95 | No clear effect |
| Train shadow models on only 100 clean samples per class (ablation) | CBD0.2 · CIFAR-10 BadNet | TPR (%): 75 -> 35 | Hurt |
| With only 100 clean samples per class, increase assumed outlier ratio from 0.2 to 0.4 (ablation) | CBD · CIFAR-10 BadNet | TPR (%): 35 -> 60 | Improved |
| Use small, noise-resilient triggers rather than larger perturbation triggers | CBD certification · Four benchmark datasets | CTPR (%): For ℓ2≤0.75: 98%, 84%, 98%, 40%; for 0.75<ℓ2≤1.5: 68%, 76%, 100%, 28% | Mixed |
Closest papers by meaning
- Inspecting Prediction Confidence for Detecting Black-Box Backdoor Attacks (AAAI 2024 · similarity 0.89)
- IBD-PSC: Input-level Backdoor Detection via Parameter-oriented Scaling Consistency (ICML 2024 · similarity 0.88)
- Lie Detector: Unified Backdoor Detection via Cross-Examination Framework (NeurIPS 2025 · similarity 0.87)
- RPP: A Certified Poisoned-Sample Detection Framework for Backdoor Attacks under Dataset Imbalance (NeurIPS 2026 · similarity 0.87)
- Backdoor Defense via Test-Time Detecting and Repairing (CVPR 2024 · similarity 0.87)
- BAN: Detecting Backdoors Activated by Adversarial Neuron Noise (NeurIPS 2024 · similarity 0.87)
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.
