Research map / Adversarial robustness and security / Certified and empirical robustness
On the Role of Randomization in Adversarially Robust Classification
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 |
|---|---|---|---|
| Randomize uniformly over f1 and f2 in the matching-penny configuration | Linear classifiers · Discrete linear-classifier toy example | adversarial risk: \frac{1}{3} -> \frac{1}{4} | Helped |
| Randomize continuously over classifiers whose individual adversarial perturbations differ | Linear classifiers · Maximum matching penny gap example | adversarial risk: 1 -> 0 | Helped |
| Use a binary probabilistic classifier instead of its best threshold classifier | Any probabilistic binary classifier · Binary classification | adversarial risk | No clear effect |
| Use an REC rather than a deterministic classifier when decision regions are closed under union and intersection | Binary classifier family · Binary classification | adversarial risk | No clear effect |
| Use an input-noise probabilistic classifier rather than a deterministic threshold extension | Input-noise injection classifier · Binary classification | adversarial risk | Hurt |
| Use a randomized ensemble rather than a deterministic weighted ensemble | Randomized ensemble classifier · Binary classification | adversarial risk | Hurt |
Closest papers by meaning
- Theoretical evidence for adversarial robustness through randomization (NeurIPS 2019 · similarity 0.91)
- Adversarial Vulnerability of Randomized Ensembles (ICML 2022 · similarity 0.91)
- Treatment of Statistical Estimation Problems in Randomized Smoothing for Adversarial Robustness (NeurIPS 2024 · similarity 0.89)
- Adversarial Robustness via Deformable Convolution with Stochasticity (ICML 2025 · similarity 0.89)
- Learning Randomized Algorithms with Transformers (ICLR 2025 · similarity 0.88)
- Consistent Adversarially Robust Linear Classification: Non-Parametric Setting (ICML 2024 · 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.