Research map / Adversarial robustness and security / Certified and empirical robustness
LiBRe: A Practical Bayesian Approach to Adversarial Detection
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Add uniformly perturbed training inputs with uncertainty regularization R (ablation) | LMFVI · ImageNet adversarial detection | AUROC: FGSM: 0.029 -> 0.990; C&W: 0.738 -> 0.921; PGD: 0.943 -> 0.980; DIM: 0.021 -> 0.989 | Helped |
| Add uniformly perturbed training inputs with uncertainty regularization R (ablation) | MFVI · ImageNet adversarial detection | AUROC: FGSM: 0.102 -> 0.986; C&W: 0.780 -> 0.943; PGD: 0.992 -> 0.999; DIM: 0.298 -> 0.992 | Helped |
| Use feature variance instead of softmax variance (ablation) | LiBRe · ImageNet adversarial detection | AUROC: FGSM: 0.708 -> 1.000; C&W: 0.107 -> 0.985; PGD: 0.361 -> 0.994; DIM: 0.650 -> 1.000 | Helped |
| Use instance-wise stochastic L* instead of batch-wise L (ablation) | LiBRe · ImageNet adversarial detection | AUROC: FGSM: 0.433 -> 1.000; C&W: 0.820 -> 0.985; PGD: 0.887 -> 0.994; DIM: 0.247 -> 1.000 | Helped |
| Expose LiBRe to attacks that also minimize predictive uncertainty | LiBRe · ImageNet ideal attacks | AUROC: 0.9996, 0.2374, 0.0363, 0.2211, 0.1627, 0.1990, 0.9998, 0.9627, 0.2537, and 0.2213 | Mixed |
Closest papers by meaning
- Adv-BNN: Improved Adversarial Defense through Robust Bayesian Neural Network (ICLR 2019 · similarity 0.89)
- Feature-Space Bayesian Adversarial Learning Improved Malware Detector Robustness (AAAI 2023 · similarity 0.88)
- Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data (ICLR 2025 · similarity 0.86)
- Robust NAS under adversarial training: benchmark, theory, and beyond (ICLR 2024 · similarity 0.86)
- Statistical Estimation of Adversarial Risk in Large Language Models under Best-of-N Sampling (NeurIPS 2026 · similarity 0.86)
- Revisiting Adversarial Training at Scale (CVPR 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.