Advertisement

Regional Homogeneity: Towards Learning Transferable Universal Adversarial Perturbations Against Defenses

Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 12356)

Abstract

This paper focuses on learning transferable adversarial examples specifically against defense models (models to defense adversarial attacks). In particular, we show that a simple universal perturbation can fool a series of state-of-the-art defenses.

Adversarial examples generated by existing attacks are generally hard to transfer to defense models. We observe the property of regional homogeneity in adversarial perturbations and suggest that the defenses are less robust to regionally homogeneous perturbations. Therefore, we propose an effective transforming paradigm and a customized gradient transformer module to transform existing perturbations into regionally homogeneous ones. Without explicitly forcing the perturbations to be universal, we observe that a well-trained gradient transformer module tends to output input-independent gradients (hence universal) benefiting from the under-fitting phenomenon. Thorough experiments demonstrate that our work significantly outperforms the prior art attacking algorithms (either image-dependent or universal ones) by an average improvement of 14.0% when attacking 9 defenses in the transfer-based attack setting. In addition to the cross-model transferability, we also verify that regionally homogeneous perturbations can well transfer across different vision tasks (attacking with the semantic segmentation task and testing on the object detection task). The code is available here: https://github.com/LiYingwei/Regional-Homogeneity.

Keywords

Transferable adversarial example Universal attack 

Notes

Acknowledgements

We thank Yuyin Zhou and Zhishuai Zhang for their insightful comments and suggestions. This work was partially supported by the Johns Hopkins University Institute for Assured Autonomy with grant IAA 80052272.

Supplementary material

504452_1_En_46_MOESM1_ESM.pdf (345 kb)
Supplementary material 1 (pdf 345 KB)

References

  1. 1.
    Akhtar, N., Liu, J., Mian, A.: Defense against universal adversarial perturbations. In: CVPR (2018)Google Scholar
  2. 2.
    Ba, J.L., Kiros, J.R., Hinton, G.E.: Layer normalization. arXiv preprint arXiv:1607.06450 (2016)
  3. 3.
    Bai, S., Li, Y., Zhou, Y., Li, Q., Torr, P.H.: Metric attack and defense for person re-identification. arXiv preprint arXiv:1901.10650 (2019)
  4. 4.
    Baluja, S., Fischer, I.: Learning to attack: adversarial transformation networks. In: AAAI (2018)Google Scholar
  5. 5.
    Bhagoji, A.N., He, W., Li, B., Song, D.: Practical black-box attacks on deep neural networks using efficient query mechanisms. In: Ferrari, V., Hebert, M., Sminchisescu, C., Weiss, Y. (eds.) ECCV 2018. LNCS, vol. 11216, pp. 158–174. Springer, Cham (2018).  https://doi.org/10.1007/978-3-030-01258-8_10CrossRefGoogle Scholar
  6. 6.
    Bishop, C.M.: The bias-variance decomposition. In: Pattern Recognition and Machine Learning, pp. 147–152. Springer, Heidelberg (2006)Google Scholar
  7. 7.
    Borkar, T., Heide, F., Karam, L.: Defending against universal attacks through selective feature regeneration. In: CVPR (2020)Google Scholar
  8. 8.
    Brendel, W., Rauber, J., Bethge, M.: Decision-based adversarial attacks: reliable attacks against black-box machine learning models. In: ICLR (2018)Google Scholar
  9. 9.
    Cao, Y., et al.: Adversarial sensor attack on lidar-based perception in autonomous driving. In: ACM SIGSAC CCS (2019)Google Scholar
  10. 10.
    Chen, L.-C., Zhu, Y., Papandreou, G., Schroff, F., Adam, H.: Encoder-decoder with Atrous separable convolution for semantic image segmentation. In: Ferrari, V., Hebert, M., Sminchisescu, C., Weiss, Y. (eds.) ECCV 2018. LNCS, vol. 11211, pp. 833–851. Springer, Cham (2018).  https://doi.org/10.1007/978-3-030-01234-2_49CrossRefGoogle Scholar
  11. 11.
    Chen, P.Y., Zhang, H., Sharma, Y., Yi, J., Hsieh, C.J.: ZOO: zeroth order optimization based black-box attacks to deep neural networks without training substitute models. In: Proceedings of the 10th ACM Workshop on Artificial Intelligence and Security (2017)Google Scholar
  12. 12.
    Chen, X., Liu, C., Li, B., Lu, K., Song, D.: Targeted backdoor attacks on deep learning systems using data poisoning. arXiv preprint arXiv:1712.05526 (2017)
  13. 13.
    Chollet, F.: Xception: deep learning with depthwise separable convolutions. In: ICCV (2017)Google Scholar
  14. 14.
    Das, N., et al.: SHIELD: fast, practical defense and vaccination for deep learning using JPEG compression. In: KDD. ACM (2018)Google Scholar
  15. 15.
    Deng, J., Dong, W., Socher, R., Li, L.J., Li, K., Fei-Fei, L.: ImageNet: a large-scale hierarchical image database. In: CVPR (2009)Google Scholar
  16. 16.
    Dong, Y., et al.: Boosting adversarial attacks with momentum. In: CVPR (2018)Google Scholar
  17. 17.
    Dong, Y., Pang, T., Su, H., Zhu, J.: Evading defenses to transferable adversarial examples by translation-invariant attacks. In: CVPR (2019)Google Scholar
  18. 18.
    Dziugaite, G.K., Ghahramani, Z., Roy, D.M.: A study of the effect of JPG compression on adversarial images. arXiv preprint arXiv:1608.00853 (2016)
  19. 19.
    Everingham, M., Eslami, S.A., Van Gool, L., Williams, C.K., Winn, J., Zisserman, A.: The pascal visual object classes challenge: a retrospective. IJCV 111(1), 98–136 (2015)CrossRefGoogle Scholar
  20. 20.
    Eykholt, K., et al.: Robust physical-world attacks on deep learning visual classification. In: CVPR (2018)Google Scholar
  21. 21.
    Gao, L., Zhang, Q., Song, J., Liu, X., Shen, H.T.: Patch-wise attack for fooling deep neural network. arXiv preprint arXiv:2007.06765 (2020)
  22. 22.
    Goodfellow, I., Bengio, Y., Courville, A.: Deep Learning. MIT Press, Cambridge (2016)zbMATHGoogle Scholar
  23. 23.
    Goodfellow, I.J., Shlens, J., Szegedy, C.: Explaining and harnessing adversarial examples. In: ICLR (2015)Google Scholar
  24. 24.
    Goyal, P., et al.: Accurate, large minibatch SGD: training ImageNet in 1 hour. arXiv preprint arXiv:1706.02677 (2017)
  25. 25.
    Guo, C., Frank, J.S., Weinberger, K.Q.: Low frequency adversarial perturbation. arXiv preprint arXiv:1809.08758 (2018)
  26. 26.
    Guo, C., Rana, M., Cissé, M., van der Maaten, L.: Countering adversarial images using input transformations. In: ICLR (2018)Google Scholar
  27. 27.
    Haykin, S.S.: Finite sample-size considerations. In: Neural Networks and Learning Machines, vol. 3, pp. 82–86. Pearson Upper Saddle River (2009)Google Scholar
  28. 28.
    He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. In: CVPR (2016)Google Scholar
  29. 29.
    Hendrik Metzen, J., Chaithanya Kumar, M., Brox, T., Fischer, V.: Universal adversarial perturbations against semantic image segmentation. In: ICCV (2017)Google Scholar
  30. 30.
    Huang, L., et al.: Universal physical camouflage attacks on object detectors. In: CVPR (2020)Google Scholar
  31. 31.
    Ioffe, S., Szegedy, C.: Batch normalization: Accelerating deep network training by reducing internal covariate shift. In: ICML (2015)Google Scholar
  32. 32.
    Jia, R., Konstantakopoulos, I.C., Li, B., Spanos, C.: Poisoning attacks on data-driven utility learning in games. In: ACC (2018)Google Scholar
  33. 33.
    Jin, W., Li, Y., Xu, H., Wang, Y., Tang, J.: Adversarial attacks and defenses on graphs: a review and empirical study. arXiv preprint arXiv:2003.00653 (2020)
  34. 34.
    Jin, W., Ma, Y., Liu, X., Tang, X., Wang, S., Tang, J.: Graph structure learning for robust graph neural networks. arXiv preprint arXiv:2005.10203 (2020)
  35. 35.
    Kannan, H., Kurakin, A., Goodfellow, I.: Adversarial logit pairing. arXiv preprint arXiv:1803.06373 (2018)
  36. 36.
    Khrulkov, V., Oseledets, I.: Art of singular vectors and universal adversarial perturbations. In: CVPR (2018)Google Scholar
  37. 37.
    Kingma, D.P., Ba, J.: Adam: a method for stochastic optimization. arXiv preprint arXiv:1412.6980 (2014)
  38. 38.
    Kurakin, A., Goodfellow, I., Bengio, S.: Adversarial examples in the physical world. In: ICLR Workshop (2017)Google Scholar
  39. 39.
    Kurakin, A., et al.: Adversarial attacks and defences competition. arXiv preprint arXiv:1804.00097 (2018)
  40. 40.
    Li, Y., Bai, S., Zhou, Y., Xie, C., Zhang, Z., Yuille, A.: Learning transferable adversarial examples via ghost networks. In: AAAI (2020)Google Scholar
  41. 41.
    Li, Y., et al.: Volumetric medical image segmentation: a 3D deep coarse-to-fine framework and its adversarial examples. In: Lu, L., Wang, X., Carneiro, G., Yang, L. (eds.) Deep Learning and Convolutional Neural Networks for Medical Imaging and Clinical Informatics. ACVPR, pp. 69–91. Springer, Cham (2019).  https://doi.org/10.1007/978-3-030-13969-8_4CrossRefGoogle Scholar
  42. 42.
    Liao, F., Liang, M., Dong, Y., Pang, T., Hu, X., Zhu, J.: Defense against adversarial attacks using high-level representation guided denoiser. In: CVPR (2018)Google Scholar
  43. 43.
    Lin, T.-Y., et al.: Microsoft COCO: common objects in context. In: Fleet, D., Pajdla, T., Schiele, B., Tuytelaars, T. (eds.) ECCV 2014. LNCS, vol. 8693, pp. 740–755. Springer, Cham (2014).  https://doi.org/10.1007/978-3-319-10602-1_48CrossRefGoogle Scholar
  44. 44.
    Liu, L., et al.: Deep neural network ensembles against deception: ensemble diversity, accuracy and robustness. In: 2019 IEEE 16th International Conference on Mobile Ad Hoc and Sensor Systems (MASS), pp. 274–282. IEEE (2019)Google Scholar
  45. 45.
    Liu, X., Cheng, M., Zhang, H., Hsieh, C.-J.: Towards robust neural networks via random self-ensemble. In: Ferrari, V., Hebert, M., Sminchisescu, C., Weiss, Y. (eds.) ECCV 2018. LNCS, vol. 11211, pp. 381–397. Springer, Cham (2018).  https://doi.org/10.1007/978-3-030-01234-2_23CrossRefGoogle Scholar
  46. 46.
    Liu, Y., Chen, X., Liu, C., Song, D.: Delving into transferable adversarial examples and black-box attacks. In: ICLR (2017)Google Scholar
  47. 47.
    Ma, X., et al.: Characterizing adversarial subspaces using local intrinsic dimensionality. In: ICLR (2018)Google Scholar
  48. 48.
    Madry, A., Makelov, A., Schmidt, L., Tsipras, D., Vladu, A.: Towards deep learning models resistant to adversarial attacks. In: ICLR (2018)Google Scholar
  49. 49.
    Mao, X., Chen, Y., Li, Y., He, Y., Xue, H.: GAP++: learning to generate target-conditioned adversarial examples. arXiv preprint arXiv:2006.05097 (2020)
  50. 50.
    Moosavi-Dezfooli, S.M., Fawzi, A., Fawzi, O., Frossard, P.: Universal adversarial perturbations. In: CVPR (2017)Google Scholar
  51. 51.
    Mopuri, K.R., Garg, U., Babu, R.V.: Fast feature fool: a data independent approach to universal adversarial perturbations. In: BMVC (2017)Google Scholar
  52. 52.
    Naseer, M.M., Khan, S.H., Khan, M.H., Khan, F.S., Porikli, F.: Cross-domain transferability of adversarial perturbations. In: Advances in Neural Information Processing Systems, pp. 12905–12915 (2019)Google Scholar
  53. 53.
    Poursaeed, O., Jiang, T., Yang, H., Belongie, S., Lim, S.N.: Fine-grained synthesis of unrestricted adversarial examples. arXiv preprint arXiv:1911.09058 (2019)
  54. 54.
    Poursaeed, O., Katsman, I., Gao, B., Belongie, S.: Generative adversarial perturbations. In: CVPR (2017)Google Scholar
  55. 55.
    Qiu, H., Xiao, C., Yang, L., Yan, X., Lee, H., Li, B.: SemanticAdv: generating adversarial examples via attribute-conditional image editing. arXiv preprint arXiv:1906.07927 (2019)
  56. 56.
    Ren, S., He, K., Girshick, R., Sun, J.: Faster R-CNN: towards real-time object detection with region proposal networks. In: NeurIPS (2015)Google Scholar
  57. 57.
    Roth, H.R., et al.: DeepOrgan: multi-level deep convolutional networks for automated pancreas segmentation. In: Navab, N., Hornegger, J., Wells, W.M., Frangi, A.F. (eds.) MICCAI 2015. LNCS, vol. 9349, pp. 556–564. Springer, Cham (2015).  https://doi.org/10.1007/978-3-319-24553-9_68CrossRefGoogle Scholar
  58. 58.
    Ruder, S.: An overview of multi-task learning in deep neural networks. arXiv preprint arXiv:1706.05098 (2017)
  59. 59.
    Rudin, L.I., Osher, S., Fatemi, E.: Nonlinear total variation based noise removal algorithms. Physica D Nonlinear Phenomena 60(1–4), 259–268 (1992)MathSciNetCrossRefGoogle Scholar
  60. 60.
    Shafahi, A., Najibi, M., Xu, Z., Dickerson, J., Davis, L.S., Goldstein, T.: Universal adversarial training. In: AAAI (2020)Google Scholar
  61. 61.
    Simonyan, K., Zisserman, A.: Very deep convolutional networks for large-scale image recognition. In: ICLR (2015)Google Scholar
  62. 62.
    Sun, M., et al.: Data poisoning attack against unsupervised node embedding methods. arXiv preprint arXiv:1810.12881 (2018)
  63. 63.
    Sun, Y., Wang, S., Tang, X., Hsieh, T.Y., Honavar, V.: Adversarial attacks on graph neural networks via node injections: a hierarchical reinforcement learning approach. In: Proceedings of the Web Conference 2020, pp. 673–683 (2020)Google Scholar
  64. 64.
    Szegedy, C., Ioffe, S., Vanhoucke, V., Alemi, A.A.: Inception-v4, Inception-ResNet and the impact of residual connections on learning. In: AAAI (2017)Google Scholar
  65. 65.
    Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J., Wojna, Z.: Rethinking the inception architecture for computer vision. In: CVPR (2016)Google Scholar
  66. 66.
    Szegedy, C., et al.: Intriguing properties of neural networks. In: ICLR (2014)Google Scholar
  67. 67.
    Tang, X., Li, Y., Sun, Y., Yao, H., Mitra, P., Wang, S.: Transferring robustness for graph neural network against poisoning attacks. In: Proceedings of the 13th International Conference on Web Search and Data Mining, pp. 600–608 (2020)Google Scholar
  68. 68.
    Tramèr, F., Kurakin, A., Papernot, N., Boneh, D., McDaniel, P.: Ensemble adversarial training: attacks and defenses. In: ICLR (2018)Google Scholar
  69. 69.
    Ulyanov, D., Vedaldi, A., Lempitsky, V.: Instance normalization: the missing ingredient for fast stylization. arXiv preprint arXiv:1607.08022 (2016)
  70. 70.
    Wu, Y., He, K.: Group normalization. In: Ferrari, V., Hebert, M., Sminchisescu, C., Weiss, Y. (eds.) ECCV 2018. LNCS, vol. 11217, pp. 3–19. Springer, Cham (2018).  https://doi.org/10.1007/978-3-030-01261-8_1CrossRefGoogle Scholar
  71. 71.
    Xiao, C., Zhong, P., Zheng, C.: Enhancing adversarial defense by k-winners-take-all. In: ICLR (2020)Google Scholar
  72. 72.
    Xiao, C., Li, B., Zhu, J.Y., He, W., Liu, M., Song, D.: Generating adversarial examples with adversarial networks. In: IJCAI (2018)Google Scholar
  73. 73.
    Xie, C., Tan, M., Gong, B., Yuille, A., Le, Q.V.: Smooth adversarial training. arXiv preprint arXiv:2006.14536 (2020)
  74. 74.
    Xie, C., Wang, J., Zhang, Z., Ren, Z., Yuille, A.: Mitigating adversarial effects through randomization. In: ICLR (2018)Google Scholar
  75. 75.
    Xie, C., Wu, Y., Maaten, L.v.d., Yuille, A.L., He, K.: Feature denoising for improving adversarial robustness. In: CVPR (2019)Google Scholar
  76. 76.
    Xie, C., Yuille, A.: Intriguing properties of adversarial training at scale. In: ICLR (2020)Google Scholar
  77. 77.
    Xie, C., Zhang, Z., Zhou, Y., Bai, S., Wang, J., Ren, Z., Yuille, A.L.: Improving transferability of adversarial examples with input diversity. In: CVPR (2019)Google Scholar
  78. 78.
    Yang, C., Kortylewski, A., Xie, C., Cao, Y., Yuille, A.: PatchAttack: a black-box texture-based attack with reinforcement learning. arXiv preprint arXiv:2004.05682 (2020)
  79. 79.
    Zhang, Z., Zhu, X., Li, Y., Chen, X., Guo, Y.: Adversarial attacks on monocular depth estimation. arXiv preprint arXiv:2003.10315 (2020)
  80. 80.
    Zhou, W., et al.: Transferable adversarial perturbations. In: Ferrari, V., Hebert, M., Sminchisescu, C., Weiss, Y. (eds.) Computer Vision – ECCV 2018. LNCS, vol. 11218, pp. 471–486. Springer, Cham (2018).  https://doi.org/10.1007/978-3-030-01264-9_28CrossRefGoogle Scholar

Copyright information

© Springer Nature Switzerland AG 2020

Authors and Affiliations

  1. 1.Johns Hopkins UniversityBaltimoreUSA
  2. 2.University of OxfordOxfordUK
  3. 3.Kuaishou TechnologyPalo AltoUSA
  4. 4.ByteDance ResearchMountain ViewUSA

Personalised recommendations