SMLE: Safe Machine Learning via Embedded Overapproximation
Not on the map. Why?
Abstract (arXiv)
Despite the extent of recent advances in Machine Learning (ML) and Neural Networks, providing formal guarantees on the behavior of these systems is still an open problem, and a crucial requirement for their adoption in regulated or safety-critical scenarios. We consider the task of training differentiable ML models guaranteed to satisfy designer-chosen properties, stated as input-output implications. This is very challenging, due to the computational complexity of rigorously verifying and enforcing compliance in modern neural models. We provide an innovative approach based on three components: 1) a general, simple architecture enabling efficient verification with a conservative semantic; 2) a rigorous training algorithm based on the Projected Gradient Method; 3) a formulation of the problem of searching for strong counterexamples. The proposed framework, being only marginally affected by model complexity, scales well to practical applications, and produces models that provide full property satisfaction guarantees. We evaluate our approach on properties defined by linear inequalities in regression, and on mutually exclusive classes in multilabel classification. Our approach is competitive with a baseline that includes property enforcement during preprocessing, i.e. on the training data, as well as during postprocessing, i.e. on the model predictions. Finally, our contributions establish a framework that opens up multiple research directions and potential improvements.
Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 24 of its 43 references that are in the map. First posted on arXiv 30 Sep 2024.
Builds on
24 of 43 references.
References · 24
- 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
- 2019Certified Adversarial Robustness via Randomized Smoothingcited by 1,324
- 2017Reluplex: An Efficient SMT Solver for Verifying Deep Neural Networkscited by 739
- 2020Fast is better than free: Revisiting adversarial trainingcited by 723
- 2017Feature Squeezing: Detecting Adversarial Examples in Deep Neural Networkscited by 484
- 2018Defense-GAN: Protecting Classifiers Against Adversarial Attacks Using Generative Modelscited by 362
- 2021Recent Advances in Adversarial Training for Adversarial Robustnesscited by 276
- 2016Safety Verification of Deep Neural Networkscited by 262
- 2020Attacks Which Do Not Kill Training Make Adversarial Learning Strongercited by 242
- 2017Formal Verification of Piece-Wise Linear Feed-Forward Neural Networkscited by 241
- 2017On Detecting Adversarial Perturbationscited by 232
- 2019Algorithms for Verifying Deep Neural Networkscited by 221
12 more
- 2019You Only Propagate Once: Accelerating Adversarial Training via Maximal Principlecited by 181
- 2018Stochastic Activation Pruning for Robust Adversarial Defensecited by 171
- 2018Strong mixed-integer programming formulations for trained neural networkscited by 115
- 2018A predictive safety filter for learning-based control of constrained nonlinear dynamical systemscited by 91
- 2017Output Reachable Set Estimation and Verification for Multi-Layer Neural Networkscited by 79
- 2020Understanding Catastrophic Overfitting in Single-step Adversarial Trainingcited by 75
- 2019ME-Net: Towards Effective Adversarial Robustness with Matrix Estimationcited by 65
- 2021Partition-based formulations for mixed-integer optimization of trained ReLU neural networkscited by 39
- 2023The Fourth International Verification of Neural Networks Competition (VNN-COMP 2023): Summary and Resultscited by 38
- 2019Analyzing Deep Neural Networks with Symbolic Propagation: Towards Higher Precision and Faster Verificationcited by 28
- 2022Towards Safe Reinforcement Learning with a Safety Editor Policycited by 28
- 2024Can Go AIs be adversarially robust?cited by 3
Built on by
Citing papers per year: 2025: 1.
1 paper.