Research map / Adversarial robustness and security / Certified and empirical robustness
Adversarial Training and Provable Robustness: A Tale of Two Objectives
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 adaptive joint optimization of adversarial and abstract losses instead of fixed equal coefficients | AdvIBP · IBP verified errors across the baseline comparison | IBP verified error: 9.1% to 31.9% | Helped |
| Replace IBP abstract-loss computation with CROWN-IBP during warm-up, then use IBP (ablation) | AdvIBP · MNIST verified error across tested ε and model structures | verified error | Helped |
| Jointly minimize adversarial loss and abstract loss | AdvIBP · CIFAR-10, ε=8/255, DM-Large | IBP verified error (%): 66.94 -> 66.57 | Helped |
| Jointly minimize adversarial loss and abstract loss | AdvIBP · MNIST, ε=0.3, DM-Large | standard error (%): 2.34 -> 2.04 | Helped |
Closest papers by meaning
- Probably Approximately Global Robustness Certification (ICML 2025 · similarity 0.92)
- Probabilistically Robust Learning: Balancing Average and Worst-case Performance (ICML 2022 · similarity 0.91)
- Learning Better Certified Models from Empirically-Robust Teachers (NeurIPS 2026 · similarity 0.90)
- Expressive Losses for Verified Robustness via Convex Combinations (ICLR 2024 · similarity 0.90)
- Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data (ICLR 2025 · similarity 0.89)
- Interpreting Robustness Proofs of Deep Neural Networks (ICLR 2024 · similarity 0.89)
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.