Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
162 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Incremental Verification of Neural Networks (2304.01874v2)

Published 4 Apr 2023 in cs.LG, cs.PL, and cs.SE

Abstract: Complete verification of deep neural networks (DNNs) can exactly determine whether the DNN satisfies a desired trustworthy property (e.g., robustness, fairness) on an infinite set of inputs or not. Despite the tremendous progress to improve the scalability of complete verifiers over the years on individual DNNs, they are inherently inefficient when a deployed DNN is updated to improve its inference speed or accuracy. The inefficiency is because the expensive verifier needs to be run from scratch on the updated DNN. To improve efficiency, we propose a new, general framework for incremental and complete DNN verification based on the design of novel theory, data structure, and algorithms. Our contributions implemented in a tool named IVAN yield an overall geometric mean speedup of 2.4x for verifying challenging MNIST and CIFAR10 classifiers and a geometric mean speedup of 3.8x for the ACAS-XU classifiers over the state-of-the-art baselines.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (70)
  1. Optuna: A Next-generation Hyperparameter Optimization Framework. In Proceedings of the 25rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.
  2. Aws Albarghouthi. 2021. Introduction to Neural Network Verification. verifieddeeplearning.com. arXiv:2109.10317 [cs.LG] http://verifieddeeplearning.com.
  3. Artificial neural networks in medical diagnosis. Journal of Applied Biomedicine 11, 2 (2013).
  4. Optimization and Abstraction: A Synergistic Approach for Analyzing Neural Network Robustness. In Proc. Programming Language Design and Implementation (PLDI).
  5. Strong mixed-integer programming formulations for trained neural networks. Mathematical Programming (2020).
  6. The Second International Verification of Neural Networks Competition (VNN-COMP 2021): Summary and Results. CoRR abs/2109.00498 (2021). arXiv:2109.00498 https://arxiv.org/abs/2109.00498
  7. Improved Geometric Path Enumeration for Verifying ReLU Neural Networks. In Computer Aided Verification - 32nd International Conference, CAV 2020, Los Angeles, CA, USA, July 21-24, 2020, Proceedings, Part I (Lecture Notes in Computer Science, Vol. 12224), Shuvendu K. Lahiri and Chao Wang (Eds.). Springer, 66–96. https://doi.org/10.1007/978-3-030-53288-8_4
  8. Mislav Balunovic and Martin Vechev. 2020. Adversarial Training and Provable Defenses: Bridging the Gap. In International Conference on Learning Representations. https://openreview.net/forum?id=SJxSDxrKDr
  9. cvc5: A Versatile and Industrial-Strength SMT Solver. In Tools and Algorithms for the Construction and Analysis of Systems - 28th International Conference, TACAS 2022, Held as Part of the European Joint Conferences on Theory and Practice of Software, ETAPS 2022, Munich, Germany, April 2-7, 2022, Proceedings, Part I (Lecture Notes in Computer Science, Vol. 13243), Dana Fisman and Grigore Rosu (Eds.). Springer, 415–442. https://doi.org/10.1007/978-3-030-99524-9_24
  10. Precision Reuse for Efficient Regression Verification. In Proceedings of the 2013 9th Joint Meeting on Foundations of Software Engineering (Saint Petersburg, Russia) (ESEC/FSE 2013). Association for Computing Machinery, New York, NY, USA, 389–399. https://doi.org/10.1145/2491411.2491429
  11. What is the State of Neural Network Pruning?. In Proceedings of Machine Learning and Systems 2020, MLSys 2020, Austin, TX, USA, March 2-4, 2020.
  12. End to end learning for self-driving cars. arXiv preprint arXiv:1604.07316 (2016).
  13. Branch and bound for piecewise linear neural network verification. Journal of Machine Learning Research 21, 2020 (2020).
  14. An efficient nonconvex reformulation of stagewise convex optimization problems. Advances in Neural Information Processing Systems 33 (2020).
  15. Robust Out-of-distribution Detection for Neural Networks. In AAAI-22 Workshop on Adversarial Machine Learning and Beyond.
  16. Chih-Hong Cheng and Rongjie Yan. 2020. Continuous Safety Verification of Neural Networks. arXiv:2010.05689 [cs.LG]
  17. IBM ILOG Cplex. 2009. V12. 1: User’s Manual for CPLEX. International Business Machines Corporation 46, 53 (2009), 157.
  18. Leonardo De Moura and Nikolaj Bjørner. 2008. Z3: An Efficient SMT Solver. In Proceedings of the Theory and Practice of Software, 14th International Conference on Tools and Algorithms for the Construction and Analysis of Systems (Budapest, Hungary) (TACAS’08/ETAPS’08). Springer-Verlag, Berlin, Heidelberg, 337–340.
  19. Boosting Adversarial Attacks With Momentum. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR).
  20. Output Range Analysis for Deep Neural Networks. CoRR abs/1709.09130 (2017). arXiv:1709.09130 http://arxiv.org/abs/1709.09130
  21. Ruediger Ehlers. 2017. Formal verification of piece-wise linear feed-forward neural networks. In International Symposium on Automated Technology for Verification and Analysis.
  22. Complete Verification via Multi-Neuron Relaxation Guided Branch-and-Bound. In International Conference on Learning Representations. https://openreview.net/forum?id=l_amHf1oaK
  23. Shared Certificates for Neural Network Verification. In Computer Aided Verification - 34th International Conference, CAV 2022, Haifa, Israel, August 7-10, 2022, Proceedings, Part I (Lecture Notes in Computer Science, Vol. 13371), Sharon Shoham and Yakir Vizel (Eds.). Springer, 127–148. https://doi.org/10.1007/978-3-031-13185-1_7
  24. Fast Geometric Projections for Local Robustness Certification. In International Conference on Learning Representations. https://openreview.net/forum?id=zWy1uxjDdZJ
  25. Feisi Fu and Wenchao Li. 2022. Sound and Complete Neural Network Repair with Minimality and Locality Guarantees. In International Conference on Learning Representations. https://openreview.net/forum?id=xS8AMYiEav3
  26. Ai2: Safety and robustness certification of neural networks with abstract interpretation. In 2018 IEEE Symposium on Security and Privacy (SP).
  27. A Survey of Quantization Methods for Efficient Neural Network Inference. CoRR abs/2103.13630 (2021). arXiv:2103.13630
  28. Attribute-Guided Adversarial Training for Robustness to Natural Perturbations. In AAAI. AAAI Press, 7574–7582.
  29. Gurobi Optimization, LLC. 2018. Gurobi Optimizer Reference Manual.
  30. Patrick Henriksen and Alessio Lomuscio. 2021. DEEPSPLIT: An Efficient Splitting Method for Neural Network Verification via Indirect Effect Analysis. In Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, IJCAI-21, Zhi-Hua Zhou (Ed.). International Joint Conferences on Artificial Intelligence Organization, 2549–2555. https://doi.org/10.24963/ijcai.2021/351 Main Track.
  31. An Incremental Verification Framework for Component-Based Software Systems. In Proceedings of the 16th International ACM Sigsoft Symposium on Component-Based Software Engineering (Vancouver, British Columbia, Canada) (CBSE ’13). Association for Computing Machinery, New York, NY, USA, 33–42. https://doi.org/10.1145/2465449.2465456
  32. Deep Neural Network Compression for Aircraft Collision Avoidance Systems. CoRR abs/1810.04240 (2018).
  33. Deep Neural Network Compression for Aircraft Collision Avoidance Systems. Journal of Guidance, Control, and Dynamics 42, 3 (mar 2019), 598–608. https://doi.org/10.2514/1.g003724
  34. Anan Kabaha and Dana Drachsler-Cohen. 2022. Boosting Robustness Verification of Semantic Feature Neighborhoods. https://doi.org/10.48550/ARXIV.2209.05446
  35. Reluplex: An efficient SMT solver for verifying deep neural networks. In International Conference on Computer Aided Verification.
  36. Reluplex: An Efficient SMT Solver for Verifying Deep Neural Networks. In Computer Aided Verification - 29th International Conference, CAV 2017, Heidelberg, Germany, July 24-28, 2017, Proceedings, Part I (Lecture Notes in Computer Science, Vol. 10426). https://doi.org/10.1007/978-3-319-63387-9_5
  37. Incremental Verification by Abstraction. In Tools and Algorithms for the Construction and Analysis of Systems: 7th International Conference, TACAS 2001, T. Margaria and W. Yi (Eds.), Vol. 2031. Springer-Verlag, Genova, Italy, 98–112.
  38. Statheros: Compiler for Efficient Low-Precision Probabilistic Programming. In Design Automation Conference (DAC). 787–792.
  39. A General Construction for Abstract Interpretation of Higher-Order Automatic Differentiation. Proc. ACM Program. Lang. 6, OOPSLA2, Article 161 (oct 2022), 29 pages. https://doi.org/10.1145/3563324
  40. Towards deep learning models resistant to adversarial attacks. arXiv preprint arXiv:1706.06083 (2017).
  41. Scaling Polyhedral Neural Network Verification on GPUs. Proc. Machine Learning and Systems (MLSys) (2021).
  42. Peter W. O’Hearn. 2018. Continuous Reasoning: Scaling the impact of formal methods. In Proceedings of the 33rd Annual ACM/IEEE Symposium on Logic in Computer Science, LICS 2018, Oxford, UK, July 09-12, 2018, Anuj Dawar and Erich Grädel (Eds.). ACM, 13–25. https://doi.org/10.1145/3209108.3209109
  43. Scaling the Convex Barrier with Active Sets. In 9th International Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021.
  44. ReluDiff: differential verification of deep neural networks. In ICSE ’20: 42nd International Conference on Software Engineering, Seoul, South Korea, 27 June - 19 July, 2020. https://doi.org/10.1145/3377811.3380337
  45. NEURODIFF: Scalable Differential Verification of Neural Networks using Fine-Grained Approximation. In 35th IEEE/ACM International Conference on Automated Software Engineering, ASE 2020, Melbourne, Australia, September 21-25, 2020. https://doi.org/10.1145/3324884.3416560
  46. A Convex Relaxation Barrier to Tight Robustness Verification of Neural Networks. In Advances in Neural Information Processing Systems 32: Annual Conference on Neural Information Processing Systems 2019, NeurIPS 2019, December 8-14, 2019, Vancouver, BC, Canada.
  47. Beyond the single neuron convex barrier for neural network certification. In Advances in Neural Information Processing Systems.
  48. Fast and effective robustness certification. Advances in Neural Information Processing Systems 31 (2018).
  49. An abstract domain for certifying neural networks. Proceedings of the ACM on Programming Languages 3, POPL (2019).
  50. Boosting Robustness Certification of Neural Networks. In International Conference on Learning Representations.
  51. Matthew Sotoudeh and Aditya V. Thakur. 2019. Computing Linear Restrictions of Neural Networks. In Advances in Neural Information Processing Systems 32: Annual Conference on Neural Information Processing Systems 2019, NeurIPS 2019, December 8-14, 2019, Vancouver, BC, Canada.
  52. Demanded abstract interpretation. In PLDI ’21: 42nd ACM SIGPLAN International Conference on Programming Language Design and Implementation, Virtual Event, Canada, June 20-25, 2021, Stephen N. Freund and Eran Yahav (Eds.). ACM, 282–295. https://doi.org/10.1145/3453483.3454044
  53. Intriguing properties of neural networks. In 2nd International Conference on Learning Representations, ICLR 2014, Banff, AB, Canada, April 14-16, 2014, Conference Track Proceedings.
  54. Convolutional neural networks for medical image analysis: Full training or fine tuning? IEEE transactions on medical imaging 35, 5 (2016), 1299–1312.
  55. TFLite. 2017. TF Lite post-training quantization. https://www.tensorflow.org/lite/performance/post_training_quantization.
  56. Evaluating robustness of neural networks with mixed integer programming. arXiv preprint arXiv:1711.07356 (2017).
  57. Proof transfer for fast certification of multiple approximate neural networks. Proc. ACM Program. Lang. 6, OOPSLA (2022), 1–29. https://doi.org/10.1145/3527319
  58. Caterina Urban and Antoine Miné. 2021. A Review of Formal Methods applied to Machine Learning. https://doi.org/10.48550/ARXIV.2104.02466
  59. Green: Reducing, Reusing and Recycling Constraints in Program Analysis. In Proceedings of the ACM SIGSOFT 20th International Symposium on the Foundations of Software Engineering (Cary, North Carolina) (FSE ’12). Association for Computing Machinery, New York, NY, USA, Article 58, 11 pages. https://doi.org/10.1145/2393596.2393665
  60. Efficient formal safety analysis of neural networks. In Advances in Neural Information Processing Systems.
  61. Beta-CROWN: Efficient Bound Propagation with Per-neuron Split Constraints for Complete and Incomplete Neural Network Verification. arXiv preprint arXiv:2103.06624 (2021).
  62. Tianhao Wei and Changliu Liu. 2021. Online Verification of Deep Neural Networks under Domain or Weight Shift. CoRR abs/2106.12732 (2021). arXiv:2106.12732 https://arxiv.org/abs/2106.12732
  63. A survey of transfer learning. Journal of Big data 3, 1 (2016), 1–40.
  64. Eric Wong and Zico Kolter. 2018a. Provable defenses against adversarial examples via the convex outer adversarial polytope. In International Conference on Machine Learning.
  65. Eric Wong and Zico Kolter. 2018b. Provable Defenses against Adversarial Examples via the Convex Outer Adversarial Polytope. In Proceedings of the 35th International Conference on Machine Learning.
  66. Automatic Perturbation Analysis for Scalable Certified Robustness and Beyond. (2020).
  67. Regression model checking. In 2009 IEEE International Conference on Software Maintenance. 115–124. https://doi.org/10.1109/ICSM.2009.5306334
  68. Provable Defense Against Geometric Transformations. arXiv:2207.11177 [cs.LG]
  69. General Cutting Planes for Bound-Propagation-Based Neural Network Verification. In Advances in Neural Information Processing Systems, Alice H. Oh, Alekh Agarwal, Danielle Belgrave, and Kyunghyun Cho (Eds.). https://openreview.net/forum?id=5haAJAcofjc
  70. Efficient neural network robustness certification with general activation functions. In Advances in neural information processing systems.
Citations (6)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now