Certifying Ensembles: A General Certification Theory with S-Lipschitzness
Not on the map. Why?
Abstract (arXiv)
Improving and guaranteeing the robustness of deep learning models has been a topic of intense research. Ensembling, which combines several classifiers to provide a better model, has shown to be beneficial for generalisation, uncertainty estimation, calibration, and mitigating the effects of concept drift. However, the impact of ensembling on certified robustness is less well understood. In this work, we generalise Lipschitz continuity by introducing S-Lipschitz classifiers, which we use to analyse the theoretical robustness of ensembles. Our results are precise conditions when ensembles of robust classifiers are more robust than any constituent classifier, as well as conditions when they are less robust.
Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 43 of its 56 references that are in the map. First posted on arXiv 25 Apr 2023.
Builds on
43 of 56 references.
References · 43
- 2015Deep Residual Learning for Image Recognitioncited by 51,809
- 2014ImageNet Large Scale Visual Recognition Challengecited by 8,441
- 2014Explaining and Harnessing Adversarial Examplescited by 7,287
- 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
- 2013Intriguing properties of neural networkscited by 5,114
- 2016Simple and Scalable Predictive Uncertainty Estimation using Deep Ensemblescited by 3,888
- 2019Theoretically Principled Trade-off between Robustness and Accuracycited by 1,510
- 2019Certified Adversarial Robustness via Randomized Smoothingcited by 1,324
- 2016Practical Black-Box Attacks against Machine Learningcited by 893
- 2018Robustness May Be at Odds with Accuracycited by 873
- 2017Reluplex: An Efficient SMT Solver for Verifying Deep Neural Networkscited by 739
- 2017Evasion Attacks against Machine Learning at Test Timecited by 686
31 more
- 2017Spectrally-normalized margin bounds for neural networkscited by 681
- 2017Provable defenses against adversarial examples via the convex outer adversarial polytopecited by 564
- 2018Certified Robustness to Adversarial Examples with Differential Privacycited by 462
- 2019Provably Robust Deep Learning via Adversarially Trained Smoothed Classifierscited by 328
- 2018On the Effectiveness of Interval Bound Propagation for Training Verifiably Robust Modelscited by 322
- 2020Towards Understanding Ensemble, Knowledge Distillation and Self-Distillation in Deep Learningcited by 292
- 2017Parseval Networks: Improving Robustness to Adversarial Examplescited by 288
- 2016Safety Verification of Deep Neural Networkscited by 262
- 2017Formal Verification of Piece-Wise Linear Feed-Forward Neural Networkscited by 241
- 2019Improving Adversarial Robustness via Promoting Ensemble Diversitycited by 213
- 2020Mix-n-Match: Ensemble and Compositional Methods for Uncertainty Calibration in Deep Learningcited by 190
- 2017Formal Guarantees on the Robustness of a Classifier against Adversarial Manipulationcited by 182
- 2018Evaluating the Robustness of Neural Networks: An Extreme Value Theory Approachcited by 181
- 2020Randomized Smoothing of All Shapes and Sizescited by 146
- 2017An approach to reachability analysis for feed-forward ReLU neural networkscited by 126
- 2019A Convex Relaxation Barrier to Tight Robustness Verification of Neural Networkscited by 124
- 2019Improving Adversarial Robustness of Ensembles with Diversity Trainingcited by 69
- 2021Training Certifiably Robust Neural Networks with Efficient Local Lipschitz Boundscited by 64
- 2022Rethinking Lipschitz Neural Networks and Certified Robustness: A Boolean Function Perspectivecited by 48
- 2021Boosting Randomized Smoothing with Variance Reduced Classifierscited by 47
- 2022Make Some Noise: Reliable and Efficient Single-Step Adversarial Trainingcited by 42
- 2021Towards Certifying L-infinity Robustness using Neural Networks with L-inf-dist Neuronscited by 40
- 2021On the Certified Robustness for Ensemble Models and Beyondcited by 40
- 2020Data-Dependent Randomized Smoothingcited by 37
- 2022On the Adversarial Robustness of Mixture of Expertscited by 27
- 2021ANCER: Anisotropic Certification via Sample-wise Volume Maximizationcited by 23
- 2021DeformRS: Certifying Input Deformations with Randomized Smoothingcited by 17
- 2018PROVEN: Certifying Robustness of Neural Networks with a Probabilistic Approachcited by 10
- 2018MixTrain: Scalable Training of Verifiably Robust Neural Networkscited by 9
- 2020PaRoT: A Practical Framework for Robust Deep Neural Network Trainingcited by 7
- 2021Mixture of Robust Experts (MoRE):A Robust Denoising Method towards multiple perturbationscited by 6
Built on by
Citing papers per year: 2024: 1; 2025: 1.
2 papers.