Research map / Adversarial robustness and security / Certified and empirical robustness
Feature-Space Bayesian Adversarial Learning Improved Malware Detector Robustness
Read the paperCheck 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 |
|---|---|---|---|
| Use Bayesian adversarial training instead of adversarial FFNN training | Malware detector · SOREL-20M and EMBER, PGD attacks | robustness: SOREL-20M, budget 0.1: 47.74 -> 62.86; EMBER, budget 0.1: 64.00 -> 78.03 | Helped |
| Use Bayesian neural networks instead of FFNNs | Malware detector · SOREL-20M, no attacks, low-FPR regimes | detection rate: up to 20% | Helped |
| Adversarially train the detector on feature-space adversarial examples | Malware detector · Problem-space Sets A and B | robustness: Set A, Adv-FFNN: 69.5% -> 92.6%; Adv-MalBayes: 72.5% -> 99.9%; Set B, Adv-FFNN: 74.9% -> 91.8%; Adv-MalBayes: 83.1% -> 9 | Helped |
| Train Adv-MalBayes with PGD L∞ | Adv-MalBayes · FGSM attacks at budgets 0.03–0.3 | robustness: 95.28; 94.87; 95.24; 93.78; 92.20 | Helped |
| Increase the number of parameter particles (ablation) | Bayesian neural network · Malware detection robustness | robustness | Helped |
Closest papers by meaning
- Adv-BNN: Improved Adversarial Defense through Robust Bayesian Neural Network (ICLR 2019 · similarity 0.88)
- LiBRe: A Practical Bayesian Approach to Adversarial Detection (CVPR 2021 · similarity 0.88)
- Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data (ICLR 2025 · similarity 0.88)
- Robust NAS under adversarial training: benchmark, theory, and beyond (ICLR 2024 · similarity 0.87)
- Statistical Estimation of Adversarial Risk in Large Language Models under Best-of-N Sampling (NeurIPS 2026 · similarity 0.86)
- Mind the Gap: Detecting Black-box Adversarial Attacks in the Making through Query Update Analysis (CVPR 2025 · 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.