mlsanityData to 7 Oct 2026

Towards Generalized Certified Robustness with Multi-Norm Training

arXiv 2410.03000 ↗First posted 3 Oct 2024cs.LG cs.CRcited by 1

Not on the map. Why?

Abstract (arXiv)

Existing certified training methods can only train models to be robust against a certain perturbation type (e.g. l_∞ or l2). However, an l_∞ certifiably robust model may not be certifiably robust against l2 perturbation (and vice versa) and also has low robustness against other perturbations (e.g. geometric and patch transformation). By constructing a theoretical framework to analyze and mitigate the tradeoff, we propose the first multi-norm certified training framework CURE, consisting of several multi-norm certified training methods, to attain better union robustness when training from scratch or fine-tuning a pre-trained certified model. Inspired by our theoretical findings, we devise bound alignment and connect natural training with certified training for better union robustness. Compared with SOTA-certified training, CURE improves union robustness to 32.0% on MNIST, 25.8% on CIFAR-10, and 10.6% on TinyImagenet across different epsilon values. It leads to better generalization on a diverse set of challenging unseen geometric and patch perturbations to 6.8% and 16.0% on CIFAR-10. Overall, our contributions pave a path towards textitgeneralized certified robustness.

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 49 of its 65 references that are in the map. First posted on arXiv 3 Oct 2024.

Builds on

49 of 65 references.

References · 49

  1. 2014Adam: A Method for Stochastic Optimizationcited by 36,960
  2. 2014Explaining and Harnessing Adversarial Examplescited by 7,287
  3. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  4. 2016Adversarial examples in the physical worldcited by 1,762
  5. 2019Theoretically Principled Trade-off between Robustness and Accuracycited by 1,510
  6. 2017Ensemble Adversarial Training: Attacks and Defensescited by 833
  7. 2017Reluplex: An Efficient SMT Solver for Verifying Deep Neural Networkscited by 739
  8. 2019Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and Entailmentcited by 663
  9. 2018Efficient Neural Network Robustness Certification with General Activation Functionscited by 439
  10. 2019Unlabeled Data Improves Adversarial Robustnesscited by 415
  11. 2018On the Effectiveness of Interval Bound Propagation for Training Verifiably Robust Modelscited by 322
  12. 2018Adversarially Robust Generalization Requires More Datacited by 318
37 more
  1. 2020Uncovering the Limits of Adversarial Training against Norm-Bounded Adversarial Examplescited by 275
  2. 2021Beta-CROWN: Efficient Bound Propagation with Per-neuron Split Constraints for Complete and Incomplete Neural Network Robustness Verificationcited by 224
  3. 2023Better Diffusion Models Further Improve Adversarial Trainingcited by 207
  4. 2019Towards Stable and Efficient Training of Verifiably Robust Neural Networkscited by 204
  5. 2020Geometry-aware Instance-reweighted Adversarial Trainingcited by 198
  6. 2019Adversarial Training and Robustness for Multiple Perturbationscited by 182
  7. 2020Torchattacks: A PyTorch Repository for Adversarial Attackscited by 156
  8. 2020Understanding and Mitigating the Tradeoff Between Robustness and Accuracycited by 148
  9. 2018Scaling provable adversarial defensescited by 141
  10. 2021Globally-Robust Neural Networkscited by 106
  11. 2020Certified Defenses for Adversarial Patchescited by 100
  12. 2019Adversarial Robustness Against the Union of Multiple Perturbation Modelscited by 88
  13. 2018Adversarial examples from computational constraintscited by 79
  14. 2022Certified Training: Small Boxes are All You Needcited by 51
  15. 2023Robust Principles: Architectural Design Principles for Adversarially Robust CNNscited by 46
  16. 2021Fast Certified Robust Training with Short Warmupcited by 45
  17. 2022LOT: Layer-wise Orthogonal Training on Improving ℓ2 Certified Robustnesscited by 32
  18. 2023Understanding Certified Training with Interval Bound Propagationcited by 26
  19. 2019Transfer of Adversarial Robustness Between Perturbation Typescited by 25
  20. 2021Adversarial Robustness against Multiple and Single lp-Threat Models via Quick Fine-Tuning of Robust Classifierscited by 23
  21. 2023Expressive Losses for Verified Robustness via Convex Combinationscited by 23
  22. 2022IBP Regularization for Verified Adversarial Robustness via Branch-and-Boundcited by 21
  23. 2020Defending Against Multiple and Unforeseen Adversarial Videoscited by 19
  24. 2020Learning to Generate Noise for Multi-Attack Robustnesscited by 17
  25. 2022Provable Defense Against Geometric Transformationscited by 14
  26. 2023Unlocking Deterministic Robustness Certification on ImageNetcited by 13
  27. 2024CTBENCH: A Library and Benchmark for Certified Trainingcited by 13
  28. 2022Adversarial Training for Improving Model Robustness? Look at Both Prediction and Interpretationcited by 12
  29. 2022Adaptive Smoothness-weighted Adversarial Training for Multiple Perturbations with Its Stability Analysiscited by 9
  30. 2020Approximate Manifold Defense Against Multiple Adversarial Perturbationscited by 7
  31. 2021Mixture of Robust Experts (MoRE):A Robust Denoising Method towards multiple perturbationscited by 6
  32. 2023Certified Adversarial Robustness Within Multiple Perturbation Boundscited by 4
  33. 2023A Recipe for Improved Certifiable Robustnesscited by 4
  34. 2024Gaussian Loss Smoothing Enables Certified Training with Tight Convex Relaxationscited by 4
  35. 2023Is Certifying ℓp Robustness Still Worthwhile?cited by 1
  36. 2024CARE: Ensemble Adversarial Robustness Evaluation Against Adaptive Attackers for Security Applicationscited by 1
  37. 2024RAMP: Boosting Adversarial Robustness Against Multiple lp Perturbations for Universal Robustnesscited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Uncovering the Limits of Proof Sharing for Neural Networkscited by 0