mlsanityData to 14 Oct 2026

Support is All You Need for Certified VAE Training

arXiv 2504.11831 ↗First posted 16 Apr 2025cs.LG stat.MLcited by 0

Not on the map. Why?

Abstract (arXiv)

Variational Autoencoders (VAEs) have become increasingly popular and deployed in safety-critical applications. In such applications, we want to give certified probabilistic guarantees on performance under adversarial attacks. We propose a novel method, CIVET, for certified training of VAEs. CIVET depends on the key insight that we can bound worst-case VAE error by bounding the error on carefully chosen support sets at the latent layer. We show this point mathematically and present a novel training algorithm utilizing this insight. We show in an extensive evaluation across different datasets (in both the wireless and vision application areas), architectures, and perturbation magnitudes that our method outperforms SOTA methods achieving good standard performance with strong robustness guarantees.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 37 of its 61 references that are in the map. First posted on arXiv 16 Apr 2025.

Builds on

37 of 61 references.

References · 37

  1. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,305
  2. 2017Neural Discrete Representation Learningcited by 4,712
  3. 2016End to End Learning for Self-Driving Carscited by 1,009
  4. 2019An Introduction to Variational Autoencoderscited by 1,000
  5. 2016Universal adversarial perturbationscited by 784
  6. 2017Reluplex: An Efficient SMT Solver for Verifying Deep Neural Networkscited by 744
  7. 2017Provable defenses against adversarial examples via the convex outer adversarial polytopecited by 567
  8. 2018On the Effectiveness of Interval Bound Propagation for Training Verifiably Robust Modelscited by 328
  9. 2017Formal Verification of Piece-Wise Linear Feed-Forward Neural Networkscited by 242
  10. 2018Sorting out Lipschitz function approximationcited by 213
  11. 2019Towards Stable and Efficient Training of Verifiably Robust Neural Networkscited by 205
  12. 2020Fast and Complete: Enabling Complete Neural Network Verification with Rapid and Massively Parallel Incomplete Verifierscited by 187
25 more
  1. 2018Efficient Formal Safety Analysis of Neural Networkscited by 165
  2. 2021PRIMA: General and Precise Neural Network Certification via Scalable Convex Hull Approximationscited by 77
  3. 2018Training for Faster Adversarial Robustness Verification via Inducing ReLU Stabilitycited by 67
  4. 2019Uncertainty Quantification with Statistical Guarantees in End-to-End Autonomous Driving Controlcited by 61
  5. 2017Adversarial examples for generative modelscited by 58
  6. 2022Certified Training: Small Boxes are All You Needcited by 52
  7. 2021Fast Certified Robust Training with Short Warmupcited by 46
  8. 2020Probabilistic Safety for Bayesian Neural Networkscited by 42
  9. 2019Statistical Guarantees for the Robustness of Bayesian Neural Networkscited by 35
  10. 2022Lossy Image Compression with Quantized Hierarchical VAEscited by 35
  11. 2021Bayesian Inference with Certifiable Adversarial Robustnesscited by 26
  12. 2023Expressive Losses for Verified Robustness via Convex Combinationscited by 23
  13. 2018Adversarial Attacks on Variational Autoencoderscited by 19
  14. 2021Make Sure You're Unsure: A Framework for Verifying Probabilistic Specificationscited by 18
  15. 2020Towards a Theoretical Understanding of the Robustness of Variational Autoencoderscited by 17
  16. 2021Certifiably Robust Variational Autoencoderscited by 14
  17. 2022Provable Defense Against Geometric Transformationscited by 14
  18. 2023Incremental Verification of Neural Networkscited by 14
  19. 2020Adversarial Training and Provable Robustness: A Tale of Two Objectivescited by 12
  20. 2023Incremental Randomized Smoothing Certificationcited by 12
  21. 2023TAPS: Connecting Certified and Adversarial Trainingcited by 11
  22. 2024Relational DNN Verification With Cross Executional Bound Refinementcited by 5
  23. 2023Interpreting Robustness Proofs of Deep Neural Networkscited by 3
  24. 2022Robust Universal Adversarial Perturbationscited by 2
  25. 2024Cross-Input Certified Training for Universal Perturbationscited by 1

Built on by

Not cited yet.