Research map / Adversarial robustness and security / Certified and empirical robustness
Beyond Perturbations: Learning Guarantees with Arbitrary Adversarial Test Examples
Shafi Goldwasser, Adam Tauman Kalai, Yael Kalai, Omar Montasser
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 unlabeled test examples (URejectron) rather than selecting the accepted set without them (SelectiveNet) | selective classification · EMNIST Q_adv | error rate on accepted examples | Improved |
| SelectiveNet on natural shift (upper-case letters) versus adversarial test set | SelectiveNet · EMNIST Q_nat vs Q_adv | error rate on accepted examples: 0.81\% -> 76.78\% | Mixed |
| Threshold distinguisher scores at a suitable value in URejectron (T=1) | URejectron · EMNIST Q_adv and Q_nat | rejection rate on P and error rate on Q | Improved |
Closest papers by meaning
- Adversarially Robust Learning with Uncertain Perturbation Sets (NeurIPS 2023 · similarity 0.88)
- Towards optimally abstaining from prediction with OOD test examples (NeurIPS 2021 · similarity 0.88)
- Wide Two-Layer Networks can Learn from Adversarial Perturbations (NeurIPS 2024 · similarity 0.87)
- Two Heads are Actually Better than One: Towards Better Adversarial Robustness via Transduction and Rejection (ICML 2024 · similarity 0.86)
- Tolerant Algorithms for Learning with Arbitrary Covariate Shift (NeurIPS 2024 · similarity 0.86)
- Theoretical Understanding of Learning from Adversarial Perturbations (ICLR 2024 · similarity 0.85)
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.
