Research map / Adversarial robustness and security / Certified and empirical robustness
Adversarial Rademacher Complexity of Deep Neural Networks
Jiancong Xiao, Yanbo Fan, Ruoyu Sun, Zhi-Quan Luo
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 adversarial training instead of standard training | VGG-19 · CIFAR-10 standard generalization gap | generalization gap: 10.45% -> 26.34% | Hurt |
| evaluate robust instead of standard gap for adversarially-trained model | VGG-19 adversarial training · CIFAR-10 | generalization gap: 26.34% -> 58.90% | Hurt |
| increase weight decay (reduce weight norms) (ablation) | VGG adversarial training · CIFAR-10/100 | robust generalization | Mixed |
| use adversarial instead of standard training on weight norm ratio | VGG models · CIFAR-10 and CIFAR-100 | normalized weight norm product | Hurt |
Closest papers by meaning
- Rademacher Complexity for Adversarially Robust Generalization (ICML 2019 · similarity 0.94)
- Adversarial Learning Guarantees for Linear Hypotheses and Neural Networks (ICML 2020 · similarity 0.93)
- Data-dependent Sample Complexity of Deep Neural Networks via Lipschitz Augmentation (NeurIPS 2019 · similarity 0.90)
- On the Generalization Analysis of Adversarial Learning (ICML 2022 · similarity 0.89)
- Adversarial Robust Generalization of Graph Neural Networks (ICML 2025 · similarity 0.89)
- Stability and Generalization of Adversarial Training for Shallow Neural Networks with Smooth Activation (NeurIPS 2024 · similarity 0.88)
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.
