Research map / Adversarial robustness and security / Backdoor and poisoning
Pre-activation Distributions Expose Backdoor Neurons
Runkai Zheng, Rongjun Tang, Jianze Li, Li Liu
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 BNP instead of ANP to defend BadNets (A2A) | BNP · CIFAR-10 BadNets (A2A) | ASR (%): 86.22 -> 0.72 | Improved |
| Use EP instead of NAD against CLA | EP · CIFAR-10 CLA | ASR (%): 15.31 -> 0.91 | Improved |
| Use EP instead of ANP against CLA | EP · CIFAR-10 CLA | ACC (%): 91.13 -> 94.42 | Improved |
| Prune backdoor neurons with EP against BadNets | ResNet-18 · CIFAR-10 BadNets (A2O) | ACC (%): 93.86 -> 93.88 | No clear effect |
| Use EP instead of ANP on WaNet | EP · Tiny-ImageNet WaNet | ASR (%): 86.98 -> 0.02 | Improved |
| Use BNP instead of ANP on BadNets (A2O) | BNP · Tiny-ImageNet BadNets (A2O) | ACC (%): 53.85 -> 61.60 | Improved |
| Use BNP instead of Refool-case EP on Tiny-ImageNet Refool | EP · Tiny-ImageNet Refool | ASR (%): 0.21 -> 17.07 | Mixed |
| Reduce benign samples for BNP from 500 to 10 (ablation) | BNP · CIFAR-10 BadNets, CLA, Blended | ACC/ASR | No clear effect |
| Increase number of benign samples for BNP (ablation) | BNP · CIFAR-10 Refool | ASR | Improved |
| Use BNP instead of NAD for running time | BNP · 500 CIFAR-10 images | running time (s): 22.08s -> 0.39s | Improved |
| Use EP instead of FT for running time | EP · 500 CIFAR-10 images | running time (s): 12.35s -> 10.69s | Improved |
Closest papers by meaning
- BAN: Detecting Backdoors Activated by Adversarial Neuron Noise (NeurIPS 2024 · similarity 0.90)
- A Theoretical Analysis of Backdoor Poisoning Attacks in Convolutional Neural Networks (ICML 2024 · similarity 0.88)
- Unveiling and Mitigating Backdoor Vulnerabilities based on Unlearning Weight Changes and Backdoor Activeness (NeurIPS 2024 · similarity 0.88)
- Backdoor Defense via Test-Time Detecting and Repairing (CVPR 2024 · similarity 0.88)
- Inspecting Prediction Confidence for Detecting Black-Box Backdoor Attacks (AAAI 2024 · similarity 0.87)
- Demystifying Poisoning Backdoor Attacks from a Statistical Perspective (ICLR 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.
