Research map / Adversarial robustness and security / Certified and empirical robustness
Provable Unrestricted Adversarial Training Without Compromise With Generalizability
Lilin Zhang, Ning Yang, Yanchao Sun, Philip S. Yu
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 |
|---|---|---|---|
| add UAE adversarial training (lambda from 0 up to 10) (ablation) | PUAT · CIFAR10 | natural and robust accuracy | Improved |
| raise UAE weight lambda above 10 (ablation) | PUAT · CIFAR10 | natural and robust accuracy | Hurt |
| add unlabeled data during training (ablation) | PUAT · CIFAR10 | natural and robust accuracy | Improved |
| train with UAEs to reduce overfitting on natural samples (ablation) | PUAT · CIFAR10 | test natural accuracy curve | Improved |
| use unlabeled data and generated data (PUAT vs DMAT) | PUAT · CIFAR100 | robust accuracy | Improved |
| PUAT vs RST under GPGD-0.1 and USong on CIFAR100 | PUAT · CIFAR100 | robust accuracy: 64.91 vs 64.99 | Hurt |
| use PUAT instead of RST on SVHN | PUAT · SVHN | natural accuracy: 85.18 -> 92.27 | Improved |
| use PUAT instead of TRADES on Tiny ImageNet | PUAT · Tiny ImageNet | PGD-8/255 robust accuracy: 31.47 -> 55.60 | Improved |
Closest papers by meaning
- On Adversarial Training without Perturbing all Examples (ICLR 2024 · similarity 0.89)
- Adversarial Distributional Training for Robust Deep Learning (NeurIPS 2020 · similarity 0.88)
- Taxonomy Driven Fast Adversarial Training (AAAI 2024 · similarity 0.88)
- Adversarial Robustness without Adversarial Training: A Teacher-Guided Curriculum Learning Approach (NeurIPS 2021 · similarity 0.88)
- Efficient Adversarial Training With Transferable Adversarial Examples (CVPR 2020 · similarity 0.87)
- DiffAdvMAP: Flexible Diffusion-Based Framework for Generating Natural Unrestricted Adversarial Examples (ICML 2025 · 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.
