Research map / Learning theory and algorithms / Combinatorial and search optimization
Lipschitz-Certifiable Training with a Tight Outer Bound
Sungyoon Lee, Jaewook Lee, Saerom Park
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 |
|---|---|---|---|
| add box constraint propagation to global Lipschitz outer bound (ablation) | BCP outer bound · MNIST | outer bound tightness (translation): 55.4% tighter on average | Improved |
| add box constraint propagation to global Lipschitz outer bound (ablation) | BCP outer bound · CIFAR-10 | outer bound tightness (translation): 45.8% tighter on average | Improved |
| add box constraint propagation to global Lipschitz outer bound (ablation) | BCP outer bound · Tiny ImageNet | outer bound tightness (translation): 25.3% tighter on average | Improved |
| BCP instead of dual relaxation CAP for training | certifiable training, MNIST 4C3F · MNIST | verification accuracy (%): 43.95 -> 47.95 | Improved |
| BCP instead of CAP | certifiable training, CIFAR-10 · CIFAR-10 verification accuracy | verification accuracy (%): 50.29 -> 49.58 (4C3F), 50.87 -> 51.30 (6C2F), 51.63 -> 50.33 (WRN) | Mixed |
| BCP instead of CAP | certifiable training, CIFAR-10 4C3F · CIFAR-10 | standard accuracy (%): 60.14 -> 63.88 | Improved |
| BCP instead of CAP | training speed · MNIST 4C3F | training time (s/epoch): 689 -> 57.5 | Improved |
| BCP instead of IBP and LMT | Tiny ImageNet · Tiny ImageNet | verification accuracy (%): 20.08 vs constant models | Improved |
Closest papers by meaning
- A Recipe for Improved Certifiable Robustness (ICLR 2024 · similarity 0.89)
- DP-SGD Without Clipping: The Lipschitz Neural Network Way (ICLR 2024 · similarity 0.86)
- ECLipsE: Efficient Compositional Lipschitz Constant Estimation for Deep Neural Networks (NeurIPS 2024 · similarity 0.86)
- Understanding Certified Training with Interval Bound Propagation (ICLR 2024 · similarity 0.86)
- Enhancing Certified Robustness via Block Reflector Orthogonal Layers and Logit Annealing Loss (ICML 2025 · similarity 0.86)
- Efficient Robust Conformal Prediction via Lipschitz-Bounded Networks (ICML 2025 · 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.
