mlsanityData to 7 Oct 2026

CAPM: Fast and Robust Verification on Maxpool-based CNN via Dual Network

arXiv 2407.09550 ↗First posted 27 Jun 2024cs.CV cs.AI cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

This study uses CAPM (Convex Adversarial Polytope for Maxpool-based CNN) to improve the verified bound for general purpose maxpool-based convolutional neural networks (CNNs) under bounded norm adversarial perturbations. The maxpool function is decomposed as a series of ReLU functions to extend the convex relaxation technique to maxpool functions, by which the verified bound can be efficiently computed through a dual network. The experimental results demonstrate that this technique allows the state-of-the-art verification precision for maxpool-based CNNs and involves a much lower computational cost than current verification methods, such as DeepZ, DeepPoly and PRIMA. This method is also applicable to large-scale CNNs, which previous studies show to be often computationally prohibitively expensive. Under certain circumstances, CAPM is 40-times, 20-times or twice as fast and give a significantly higher verification bound (CAPM 98% vs. PRIMA 76%/DeepPoly 73%/DeepZ 8%) as compared to PRIMA/DeepPoly/DeepZ. Furthermore, we additionally present the time complexity of our algorithm as O(W2NK), where W is the maximum width of the neural network, N is the number of neurons, and K is the size of the maxpool layer's kernel.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 36 of its 48 references that are in the map. First posted on arXiv 27 Jun 2024.

Builds on

36 of 48 references.

References · 36

  1. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,267
  2. 2014Explaining and Harnessing Adversarial Examplescited by 7,287
  3. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  4. 2016Towards Evaluating the Robustness of Neural Networkscited by 2,976
  5. 2021Convex optimizationcited by 2,479
  6. 2015DeepFool: a simple and accurate method to fool deep neural networkscited by 1,315
  7. 2016Universal adversarial perturbationscited by 774
  8. 2017Reluplex: An Efficient SMT Solver for Verifying Deep Neural Networkscited by 739
  9. 2020Fast is better than free: Revisiting adversarial trainingcited by 723
  10. 2019Adversarial Training for Free!cited by 638
  11. 2017One pixel attack for fooling deep neural networkscited by 579
  12. 2017Provable defenses against adversarial examples via the convex outer adversarial polytopecited by 564
24 more
  1. 2020On Adaptive Attacks to Adversarial Example Defensescited by 547
  2. 2018Defense-GAN: Protecting Classifiers Against Adversarial Attacks Using Generative Modelscited by 362
  3. 2017Adversarial Examples: Attacks and Defenses for Deep Learningcited by 350
  4. 2018On the Effectiveness of Interval Bound Propagation for Training Verifiably Robust Modelscited by 322
  5. 2018Feature Denoising for Improving Adversarial Robustnesscited by 312
  6. 2018Certified Defenses against Adversarial Examplescited by 305
  7. 2021Beta-CROWN: Efficient Bound Propagation with Per-neuron Split Constraints for Complete and Incomplete Neural Network Robustness Verificationcited by 224
  8. 2019Towards Stable and Efficient Training of Verifiably Robust Neural Networkscited by 204
  9. 2018Event-based Vision meets Deep Learning on Steering Prediction for Self-driving Carscited by 197
  10. 2020Fast and Complete: Enabling Complete Neural Network Verification with Rapid and Massively Parallel Incomplete Verifierscited by 183
  11. 2017EAD: Elastic-Net Attacks to Deep Neural Networks via Adversarial Examplescited by 154
  12. 2017Generative Adversarial Perturbationscited by 153
  13. 2018Scaling provable adversarial defensescited by 141
  14. 2018A Dual Approach to Scalable Verification of Deep Networkscited by 125
  15. 2019A Convex Relaxation Barrier to Tight Robustness Verification of Neural Networkscited by 124
  16. 2018ComDefend: An Efficient Image Compression Model to Defend Adversarial Examplescited by 104
  17. 2019Wasserstein Adversarial Examples via Projected Sinkhorn Iterationscited by 89
  18. 2021PRIMA: General and Precise Neural Network Certification via Scalable Convex Hull Approximationscited by 77
  19. 2017Defense against Universal Adversarial Perturbationscited by 48
  20. 2021Improved Branch and Bound for Neural Network Verification via Lagrangian Decompositioncited by 44
  21. 2020Lagrangian Decomposition for Neural Network Verificationcited by 35
  22. 2016A Comprehensive Analysis of Deep Learning Based Representation for Face Recognitioncited by 23
  23. 2015When Face Recognition Meets with Deep Learning: an Evaluation of Convolutional Neural Networks for Face Recognitioncited by 20
  24. 2019Mimic and Fool: A Task Agnostic Adversarial Attackcited by 11

Built on by

Not cited yet.