Research map / Adversarial robustness and security / Certified and empirical robustness
Adversarially Robust Generalization Requires More Data
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 |
|---|---|---|---|
| Increase training data for robust rather than standard generalization | Robust classifiers · MNIST, CIFAR-10, and SVHN | samples required for a given test accuracy | Helped |
| Train robustly on CIFAR10 | Wide residual network · CIFAR10 test | adversarial accuracy | Hurt |
| Add explicit thresholding filters (ablation) | MNIST convolutional network · MNIST | adversarial test accuracy | Helped |
| Add explicit thresholding filters (ablation) | CIFAR10 convolutional network · CIFAR10 | adversarial accuracy | No clear effect |
| Use thresholding before a linear classifier (ablation) | Bernoulli model classifier · Bernoulli model | robust classification error: at most 1% | Helped |
| Use a linear classifier with limited samples | Bernoulli model · Bernoulli model | robust classification error: at least 1/2 - γ | Hurt |
| Use at most c_2 ε^2 √d / log d samples | Any learning algorithm · Gaussian model | robust classification error: at least (1 - 1/d) 1/2 | Hurt |
Closest papers by meaning
- Robustness May Be at Odds with Accuracy (ICLR 2019 · similarity 0.93)
- Fantastic Robustness Measures: The Secrets of Robust Generalization (NeurIPS 2023 · similarity 0.92)
- More Data Can Expand The Generalization Gap Between Adversarially Robust and Standard Models (ICML 2020 · similarity 0.91)
- Why adversarial training can hurt robust accuracy (ICLR 2023 · similarity 0.91)
- Adversarially Robust Learning: A Generic Minimax Optimal Learner and Characterization (NeurIPS 2022 · similarity 0.91)
- Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data (ICLR 2025 · similarity 0.91)
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.