Research map / Adversarial robustness and security / Certified and empirical robustness
Scalable Differential Privacy with Certified Robustness in Adversarial Learning
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 StoBatch instead of SecureSGD | DP adversarial learning · MNIST, ε=0.2, l∞ attack μa=0.2 | conventional accuracy: 41.64% -> 82.7% | Helped |
| Use StoBatch instead of SecureSGD-AGM | DP adversarial learning · MNIST, ε=0.2, l∞ attack μa=0.2 | conventional accuracy: 11.2% -> 82.7% | Helped |
| Use StoBatch instead of SecureSGD | DP adversarial learning · CIFAR-10, iterative attacks | conventional accuracy: 29.08±11.95% -> 45.25±1.6% | Helped |
| Use StoBatch instead of SecureSGD | DP adversarial learning · CIFAR-10, iterative attacks | certified accuracy: 19.58±5.0% -> 42.59±1.58% | Helped |
| Use distributed StoBatch training instead of iterative batch-by-batch DP adversarial training | CIFAR-10 training with Tμ=200 and Ta=2,000 · CIFAR-10, 600 epochs | training time: over 30 days -> ≈ 3 days | Helped |
| Use StoBatch instead of SecureSGD | DP adversarial learning · Tiny ImageNet, ResNet18, iterative attacks | conventional accuracy: 8.99±5.95% -> 29.78±4.8% | Helped |
| Use StoBatch instead of SecureSGD | DP adversarial learning · Tiny ImageNet, ResNet18, iterative attacks | certified accuracy: 8.72±5.5% -> 28.31±1.58% | Helped |
| Add ensemble adversarial learning, random attack size, tighter sensitivity, and composed input/latent robustness | StoBatch · MNIST, CIFAR-10, and Tiny ImageNet under varied attacks | accuracy and robustness consistency | Helped |
| Randomize both input and latent spaces rather than only one space | StoBatch certified robustness · l_p-norm adversarial attacks | certified robustness bound | Helped |
Closest papers by meaning
- Provable Robustness against Backdoor Attacks via the Primal-Dual Perspective on Differential Privacy (NeurIPS 2026 · similarity 0.90)
- The Limits of Differential Privacy in Online Learning (NeurIPS 2024 · similarity 0.89)
- A Unified Framework for Adversary-Aware Differential Privacy Bounds (NeurIPS 2026 · similarity 0.88)
- 𝑓-Differential Privacy Filters: Validity and Approximate Solutions (NeurIPS 2026 · similarity 0.87)
- Convex Approximation of Two-Layer ReLU Networks for Hidden State Differential Privacy (NeurIPS 2025 · similarity 0.87)
- Towards hyperparameter-free optimization with differential privacy (ICLR 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.