Research map / Adversarial robustness and security / Certified and empirical robustness
Robust Training of Neural Networks against Bias Field Perturbations
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 |
|---|---|---|---|
| Replace IBP bounds with SSIP bounds in robust training | RSIP-IBP · CIFAR10 test, ε=40 | certified robustness (%): 35.2 -> 62.9 | Helped |
| Replace IBP bounds with SSIP bounds in robust training | RSIP-IBP · CIFAR10 test, ε=40 | accuracy (%): 46.3 -> 76.6 | Helped |
| Replace IBP bounds with SSIP bounds in robust training | RSIP-IBP · CIFAR10 test, ε=1 | certified robustness (%): 77.8 -> 81.3 | Helped |
| Replace IBP bounds with SSIP bounds in robust training | RSIP-IBP · CIFAR10 test, ε=1 | accuracy (%): 78.1 -> 81.6 | Helped |
| Increase perturbation radius | RSIP-IBP · CIFAR10 test | certified robustness (%): 77.8 -> 35.2 | Hurt |
| Increase perturbation radius | RSIP-SSIP · CIFAR10 test | certified robustness (%): 81.3 -> 62.9 | Hurt |
| Increase perturbation radius | RSIP-IBP · CIFAR10 test | accuracy (%): 78.1 -> 46.3 | Hurt |
| Increase perturbation radius | RSIP-SSIP · CIFAR10 test | accuracy (%): 81.6 -> 76.6 | Hurt |
| Replace IBP bounds with SSIP bounds | RSIP-IBP · MRI brain tumour classification, ε=1 | accuracy (%): 98.2 -> 97.6 | Hurt |
| Replace IBP bounds with SSIP bounds | RSIP-IBP · MRI brain tumour classification, ε=1 | certified robustness (%): 98.1 -> 97.5 | Hurt |
| Replace IBP bounds with SSIP bounds | RSIP-IBP · MRI brain tumour classification, ε=40 | certified robustness (%): 44.3 -> 96.6 | Helped |
| Replace IBP bounds with SSIP bounds | RSIP-IBP · MRI brain tumour classification, ε=40 | accuracy (%): 87.4 -> 97.4 | Helped |
Closest papers by meaning
- On the Convergence of Certified Robust Training with Interval Bound Propagation (ICLR 2022 · similarity 0.91)
- Understanding Certified Training with Interval Bound Propagation (ICLR 2024 · similarity 0.90)
- Towards Evaluating and Training Verifiably Robust Neural Networks (CVPR 2021 · similarity 0.89)
- Certified Robustness for Deep Equilibrium Models via Interval Bound Propagation (ICLR 2022 · similarity 0.89)
- Towards Stable and Efficient Training of Verifiably Robust Neural Networks (ICLR 2020 · similarity 0.88)
- Learning Better Certified Models from Empirically-Robust Teachers (NeurIPS 2026 · 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.