Research map / Adversarial robustness and security / Backdoor and poisoning
Reliable Poisoned Sample Detection against Backdoor Attacks Enhanced by Sharpness Aware Minimization
Mingda Zhang, Mingli Zhu, Zihao Zhu, Li Shen, Baoyuan Wu
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 |
|---|---|---|---|
| train detector model with SAM | Spectre · CIFAR-10 ResNet18, BadNets | TPR (%): 51.1 -> 88.4 | Improved |
| train detector model with SAM | Beatrix · CIFAR-10 ResNet18, Blended | TPR (%): 5.0 -> 99.8 | Improved |
| train detector model with SAM | AC · CIFAR-10 ResNet18, BadNets | FPR / F1: F1 97.1 -> 42.5 | Hurt |
| train detector model with SAM | SCAn · CIFAR-10 ResNet18, BadNets | TPR (%): 96.0 -> 95.2 | No clear effect |
| train detector model with SAM | SCAn · Tiny-ImageNet, WaNet | TPR (%): 0.0 -> 71.1 | Improved |
| add SAM training alone (ablation) | SS · CIFAR-10 Blended | TPR (%): 32.9 -> 90.6 | Improved |
| add feature scaling alone (ablation) | SS · CIFAR-10 Blended | TPR (%): 32.9 -> 60.4 | Improved |
| add feature scaling on top of SAM (ablation) | SS · CIFAR-10 Blended | TPR (%): 90.6 -> 94.6 | Improved |
| add feature scaling on top of SAM (ablation) | Beatrix · CIFAR-10 Blended | TPR (%): 79.8 -> 99.8 | Improved |
| use SAM at very low poisoning ratio 0.1% | SAM-enhanced PSD · CIFAR-10 ResNet18 | average TPR: TPR below 60% | Mixed |
| vary SAM perturbation budget rho (ablation) | SAM-enhanced PSD · CIFAR-10 ResNet18 | detection performance | No clear effect |
| train with SAM instead of vanilla training (ablation) | BadNets backdoor model · CIFAR-10 ResNet18 | silhouette coefficient: 0.19 -> 0.32 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Spectre — SPECTRE: defending against backdoor attacks using robust statistics (ICML 2021) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Activation Gradient based Poisoned Sample Detection Against Backdoor Attacks (ICLR 2025 · similarity 0.90)
- Demystifying Poisoning Backdoor Attacks from a Statistical Perspective (ICLR 2024 · similarity 0.88)
- Backdoor Defense via Test-Time Detecting and Repairing (CVPR 2024 · similarity 0.87)
- A Set of Generalized Components to Achieve Effective Poison-only Clean-label Backdoor Attacks with Collaborative Sample Selection and Triggers (NeurIPS 2025 · similarity 0.87)
- RPP: A Certified Poisoned-Sample Detection Framework for Backdoor Attacks under Dataset Imbalance (NeurIPS 2026 · similarity 0.86)
- Mitigating Backdoor Attack by Injecting Proactive Defensive Backdoor (NeurIPS 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.
