Abstract Neural Networks

Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 12389)


Deep Neural Networks (DNNs) are rapidly being applied to safety-critical domains such as drone and airplane control, motivating techniques for verifying the safety of their behavior. Unfortunately, DNN verification is NP-hard, with current algorithms slowing exponentially with the number of nodes in the DNN. This paper introduces the notion of Abstract Neural Networks (ANNs), which can be used to soundly overapproximate DNNs while using fewer nodes. An ANN is like a DNN except weight matrices are replaced by values in a given abstract domain. We present a framework parameterized by the abstract domain and activation functions used in the DNN that can be used to construct a corresponding ANN. We present necessary and sufficient conditions on the DNN activation functions for the constructed ANN to soundly over-approximate the given DNN. Prior work on DNN abstraction was restricted to the interval domain and ReLU activation function. Our framework can be instantiated with other abstract domains such as octagons and polyhedra, as well as other activation functions such as Leaky ReLU, Sigmoid, and Hyperbolic Tangent.


Deep Neural Networks Abstraction Soundness 



We thank the anonymous reviewers and Cindy Rubio González for their feedback on this work.


  1. 1.
    Bagnara, R., Hill, P.M., Zaffanella, E.: The parma polyhedra library: toward a complete set of numerical abstractions for the analysis and verification of hardware and software systems. Sci. Comput. Program. 72(1–2), 3–21 (2008). Scholar
  2. 2.
    Beheshti, M., Berrached, A., de Korvin, A., Hu, C., Sirisaengtaksin, O.: On interval weighted three-layer neural networks. In: Proceedings 31st Annual Simulation Symposium (SS 1998), 5–9 April 1998, Boston, MA, USA. pp. 188–194. IEEE Computer Society (1998).
  3. 3.
    Brown, T.B., et al.: Language models are few-shot learners. CoRR abs/2005.14165 (2020).
  4. 4.
    Chen, L., Miné, A., Cousot, P.: A sound floating-point polyhedra abstract domain. In: Ramalingam, G. (ed.) APLAS 2008. LNCS, vol. 5356, pp. 3–18. Springer, Heidelberg (2008). Scholar
  5. 5.
    Cousot, P., Cousot, R.: Abstract interpretation: a unified lattice model for static analysis of programs by construction or approximation of fixpoints. In: Graham, R.M., Harrison, M.A., Sethi, R. (eds.) Conference Record of the Fourth ACM Symposium on Principles of Programming Languages, Los Angeles, California, USA, January 1977, pp. 238–252. ACM (1977).
  6. 6.
    Cousot, P., Halbwachs, N.: Automatic discovery of linear restraints among variables of a program. In: Aho, A.V., Zilles, S.N., Szymanski, T.G. (eds.) Conference Record of the Fifth Annual ACM Symposium on Principles of Programming Languages, Tucson, Arizona, USA, January 1978, pp. 84–96. ACM Press (1978).
  7. 7.
    Deng, L., Li, G., Han, S., Shi, L., Xie, Y.: Model compression and hardware acceleration for neural networks: a comprehensive survey. Proc. IEEE 108(4), 485–532 (2020). Scholar
  8. 8.
    Garczarczyk, Z.A.: Interval neural networks. In: IEEE International Symposium on Circuits and Systems, ISCAS 2000, Emerging Technologies for the 21st Century, Geneva, Switzerland, 28–31 May 2000, Proceedings. pp. 567–570. IEEE (2000).
  9. 9.
    Gehr, T., Mirman, M., Drachsler-Cohen, D., Tsankov, P., Chaudhuri, S., Vechev, M.T.: AI2: safety and robustness certification of neural networks with abstract interpretation. In: 2018 IEEE Symposium on Security and Privacy, SP 2018, Proceedings, 21–23 May 2018, San Francisco, California, USA, pp. 3–18. IEEE Computer Society (2018).
  10. 10.
    Ghorbal, K., Ivančić, F., Balakrishnan, G., Maeda, N., Gupta, A.: Donut domains: efficient non-convex domains for abstract interpretation. In: Kuncak, V., Rybalchenko, A. (eds.) VMCAI 2012. LNCS, vol. 7148, pp. 235–250. Springer, Heidelberg (2012). Scholar
  11. 11.
    Giacobazzi, R., Ranzato, F., Scozzari, F.: Making abstract interpretations complete. J. ACM 47(2), 361–416 (2000). Scholar
  12. 12.
    Gokulanathan, S., Feldsher, A., Malca, A., Barrett, C.W., Katz, G.: Simplifying neural networks with the marabou verification engine. CoRR abs/1910.12396 (2019).
  13. 13.
    Goodfellow, I.J., Bengio, Y., Courville, A.C.: Deep Learning. Adaptive Computation and Machine Learning. MIT Press (2016).
  14. 14.
    Han, S., Mao, H., Dally, W.J.: Deep compression: Compressing deep neural network with pruning, trained quantization and huffman coding. In: Bengio, Y., LeCun, Y. (eds.) 4th International Conference on Learning Representations, ICLR 2016, San Juan, Puerto Rico, May 2–4, 2016, Conference Track Proceedings (2016).
  15. 15.
    Iandola, F.N., Moskewicz, M.W., Ashraf, K., Han, S., Dally, W.J., Keutzer, K.: Squeezenet: Alexnet-level accuracy with 50x fewer parameters and \(<\)1mb model size. CoRR abs/1602.07360 (2016).
  16. 16.
    Jeannet, B., Miné, A.: Apron: a library of numerical abstract domains for static analysis. In: Bouajjani, A., Maler, O. (eds.) CAV 2009. LNCS, vol. 5643, pp. 661–667. Springer, Heidelberg (2009). Scholar
  17. 17.
    Julian, K.D., Kochenderfer, M.J., Owen, M.P.: Deep neural network compression for aircraft collision avoidance systems. CoRR abs/1810.04240 (2018).
  18. 18.
    Katz, G., Barrett, C., Dill, D.L., Julian, K., Kochenderfer, M.J.: Reluplex: an efficient SMT solver for verifying deep neural networks. In: Majumdar, R., Kunčak, V. (eds.) CAV 2017. LNCS, vol. 10426, pp. 97–117. Springer, Cham (2017). Scholar
  19. 19.
    Katz, G., et al.: The marabou framework for verification and analysis of deep neural networks. In: Dillig, I., Tasiran, S. (eds.) CAV 2019. LNCS, vol. 11561, pp. 443–452. Springer, Cham (2019). Scholar
  20. 20.
    Krizhevsky, A., Sutskever, I., Hinton, G.E.: Imagenet classification with deep convolutional neural networks. In: Bartlett, P.L., Pereira, F.C.N., Burges, C.J.C., Bottou, L., Weinberger, K.Q. (eds.) Advances in Neural Information Processing Systems 25: 26th Annual Conference on Neural Information Processing Systems 2012. Proceedings of a meeting held December 3–6, 2012, Lake Tahoe, Nevada, United States, pp. 1106–1114 (2012).
  21. 21.
    Li, Y., Albarghouthi, A., Kincaid, Z., Gurfinkel, A., Chechik, M.: Symbolic optimization with SMT solvers. In: Jagannathan, S., Sewell, P. (eds.) The 41st Annual ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages, POPL 2014, San Diego, CA, USA, January 20–21, 2014. pp. 607–618. ACM (2014).
  22. 22.
    Maas, A., Hannun, A., Ng, A.: Rectifier nonlinearities improve neural network acoustic models. In: Proceedings of the International Conference on Machine Learning (2013)Google Scholar
  23. 23.
    Miné, A.: The octagon abstract domain. High. Order Symb. Comput. 19(1), 31–100 (2006). Scholar
  24. 24.
    Miné, A.: Tutorial on static inference of numeric invariants by abstract interpretation. Found. Trends Program. Lang. 4(3–4), 120–372 (2017). Scholar
  25. 25.
    Nakanishi, T., Joe, K., Polychronopoulos, C.D., Fukuda, A.: The modulo interval: a simple and practical representation for program analysis. In: Proceedings of the 1999 International Conference on Parallel Architectures and Compilation Techniques, Newport Beach, California, USA, October 12–16, 1999, pp. 91–96. IEEE Computer Society (1999).
  26. 26.
    Paszke, A., et al.: Pytorch: An imperative style, high-performance deep learning library. In: Wallach, H.M., Larochelle, H., Beygelzimer, A., d’Alché-Buc, F., Fox, E.B., Garnett, R. (eds.) Advances in Neural Information Processing Systems 32: Annual Conference on Neural Information Processing Systems 2019, NeurIPS 2019, 8–14 December 2019, Vancouver, BC, Canada, pp. 8024–8035 (2019).
  27. 27.
    Patiño-Escarcina, R.E., Callejas Bedregal, B.R., Lyra, A.: Interval computing in neural networks: one layer interval neural networks. In: Das, G., Gulati, V.P. (eds.) CIT 2004. LNCS, vol. 3356, pp. 68–75. Springer, Heidelberg (2004). Scholar
  28. 28.
    Ponsini, O., Michel, C., Rueher, M.: Verifying floating-point programs with constraint programming and abstract interpretation techniques. Autom. Softw. Eng. 23(2), 191–217 (2016). Scholar
  29. 29.
    Prabhakar, P., Afzal, Z.R.: Abstraction based output range analysis for neural networks. In: Wallach, H.M., Larochelle, H., Beygelzimer, A., d’Alché-Buc, F., Fox, E.B., Garnett, R. (eds.) Advances in Neural Information Processing Systems 32: Annual Conference on Neural Information Processing Systems 2019, NeurIPS 2019, 8–14 December 2019, Vancouver, BC, Canada, pp. 15762–15772 (2019).
  30. 30.
    Reps, T., Sagiv, M., Yorsh, G.: Symbolic implementation of the best transformer. In: Steffen, B., Levi, G. (eds.) VMCAI 2004. LNCS, vol. 2937, pp. 252–266. Springer, Heidelberg (2004). Scholar
  31. 31.
    Reps, T., Thakur, A.: Automating abstract interpretation. In: Jobstmann, B., Leino, K.R.M. (eds.) VMCAI 2016. LNCS, vol. 9583, pp. 3–40. Springer, Heidelberg (2016). Scholar
  32. 32.
    Shriver, D., Xu, D., Elbaum, S.G., Dwyer, M.B.: Refactoring neural networks for verification. CoRR abs/1908.08026 (2019).
  33. 33.
    Singh, G., Gehr, T., Püschel, M., Vechev, M.T.: An abstract domain for certifying neural networks. Proc. ACM Program. Lang. 3(POPL), 411–4130 (2019). Scholar
  34. 34.
    Sotoudeh, M., Thakur, A.V.: Abstract neural networks. CoRR abs/2009.05660 (2020).
  35. 35.
    Thakur, A., Elder, M., Reps, T.: Bilateral algorithms for symbolic abstraction. In: Miné, A., Schmidt, D. (eds.) SAS 2012. LNCS, vol. 7460, pp. 111–128. Springer, Heidelberg (2012). Scholar
  36. 36.
    Thakur, A., Reps, T.: A method for symbolic computation of abstract operations. In: Madhusudan, P., Seshia, S.A. (eds.) CAV 2012. LNCS, vol. 7358, pp. 174–192. Springer, Heidelberg (2012). Scholar
  37. 37.
    Wang, S., Pei, K., Whitehouse, J., Yang, J., Jana, S.: Efficient formal safety analysis of neural networks. In: Bengio, S., Wallach, H.M., Larochelle, H., Grauman, K., Cesa-Bianchi, N., Garnett, R. (eds.) Advances in Neural Information Processing Systems 31: Annual Conference on Neural Information Processing Systems 2018, NeurIPS 2018, 3–8 December 2018, Montréal, Canada, pp. 6369–6379 (2018).
  38. 38.
    Wang, S., Pei, K., Whitehouse, J., Yang, J., Jana, S.: Formal security analysis of neural networks using symbolic intervals. In: Enck, W., Felt, A.P. (eds.) 27th USENIX Security Symposium, USENIX Security 2018, Baltimore, MD, USA, August 15–17, 2018, pp. 1599–1614. USENIX Association (2018).

Copyright information

© Springer Nature Switzerland AG 2020

Authors and Affiliations

  1. 1.University of CaliforniaDavisUSA

Personalised recommendations