Document Open Access Logo

Robustness Should Not Be at Odds with Accuracy

Authors Sadia Chowdhury, Ruth Urner

PDF
Thumbnail PDF

File

LIPIcs.FORC.2022.5.pdf
  • Filesize: 5.71 MB
  • 20 pages

Document Identifiers

Author Details

Sadia Chowdhury
  • EECS Department, York University, Toronto, Canada
Ruth Urner
  • EECS Department, York University, Toronto, Canada

Funding

This research was supported by an NSERC discovery grant.

Acknowledgements

Ruth Urner is also faculty affiliate member of Toronto’s Vector Institute.

Cite As Get BibTex

Sadia Chowdhury and Ruth Urner. Robustness Should Not Be at Odds with Accuracy. In 3rd Symposium on Foundations of Responsible Computing (FORC 2022). Leibniz International Proceedings in Informatics (LIPIcs), Volume 218, pp. 5:1-5:20, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2022) https://doi.org/10.4230/LIPIcs.FORC.2022.5

Abstract

The phenomenon of adversarial examples in deep learning models has caused substantial concern over their reliability and trustworthiness: in many instances an imperceptible perturbation can falsely flip a neural network’s prediction. Applied research in this area has mostly focused on developing novel adversarial attack strategies or building better defenses against such. It has repeatedly been pointed out that adversarial robustness may be in conflict with requirements for high accuracy. In this work, we take a more principled look at modeling the phenomenon of adversarial examples. We argue that deciding whether a model’s label change under a small perturbation is justified, should be done in compliance with the underlying data-generating process. Through a series of formal constructions, systematically analyzing the relation between standard Bayes classifiers and robust-Bayes classifiers, we make the case for adversarial robustness as a locally adaptive measure. We propose a novel way defining such a locally adaptive robust loss, show that it has a natural empirical counterpart, and develop resulting algorithmic guidance in form of data-informed adaptive robustness radius. We prove that our adaptive robust data-augmentation maintains consistency of 1-nearest neighbor classification under deterministic labels and thereby argue that robustness should not be at odds with accuracy.

Subject Classification

ACM Subject Classification
  • Theory of computation → Machine learning theory
Keywords
  • Statistical Learning Theory
  • Bayes optimal classifier
  • adversarial perturbations
  • adaptive robust loss

Metrics

  • Access Statistics
  • Total Accesses (updated on a weekly basis)
    0
    PDF Downloads
    0
    Metadata Views

Related Versions

References

  1. Naveed Akhtar and Ajmal Mian. Threat of adversarial attacks on deep learning in computer vision: A survey. IEEE Access, 6:14410-14430, 2018. Google Scholar
  2. Hassan Ashtiani, Vinayak Pathak, and Ruth Urner. Black-box certification and learning under adversarial perturbations. In Proceedings of the 37th International Conference on Machine Learning, ICML 2020, 2020. Google Scholar
  3. Idan Attias, Aryeh Kontorovich, and Yishay Mansour. Improved generalization bounds for robust learning. In Algorithmic Learning Theory, ALT, pages 162-183, 2019. Google Scholar
  4. Pranjal Awasthi, Abhratanu Dutta, and Aravindan Vijayaraghavan. On robustness to adversarial examples and polynomial optimization. In Advances in Neural Information Processing Systems, NeurIPS, pages 13760-13770, 2019. Google Scholar
  5. Yogesh Balaji, Tom Goldstein, and Judy Hoffman. Instance adaptive adversarial training: Improved accuracy tradeoffs in neural nets. CoRR, abs/1910.08051, 2019. URL: http://arxiv.org/abs/1910.08051.
  6. Robi Bhattacharjee and Kamalika Chaudhuri. When are non-parametric methods robust? In Proceedings of the 37th International Conference on Machine Learning, ICML 2020, 2020. Google Scholar
  7. Robi Bhattacharjee and Kamalika Chaudhuri. Consistent non-parametric methods for adaptive robustness. CoRR, abs/2102.09086, 2021. URL: http://arxiv.org/abs/2102.09086.
  8. Sébastien Bubeck, Yin Tat Lee, Eric Price, and Ilya P. Razenshteyn. Adversarial examples from computational constraints. In Proceedings of the 36th International Conference on Machine Learning, ICML, pages 831-840, 2019. Google Scholar
  9. Nicholas Carlini, Anish Athalye, Nicolas Papernot, Wieland Brendel, Jonas Rauber, Dimitris Tsipras, Ian J. Goodfellow, Aleksander Madry, and Alexey Kurakin. On evaluating adversarial robustness. CoRR, abs/1902.06705, 2019. URL: http://arxiv.org/abs/1902.06705.
  10. Anirban Chakraborty, Manaar Alam, Vishal Dey, Anupam Chattopadhyay, and Debdeep Mukhopadhyay. Adversarial attacks and defences: A survey. CoRR, abs/1810.00069, 2018. URL: http://arxiv.org/abs/1810.00069.
  11. Kamalika Chaudhuri and Sanjoy Dasgupta. Rates of convergence for nearest neighbor classification. In Advances in Neural Information Processing Systems, NIPS, pages 3437-3445, 2014. Google Scholar
  12. Jeremy M. Cohen, Elan Rosenfeld, and J. Zico Kolter. Certified adversarial robustness via randomized smoothing. In Proceedings of the 36th International Conference on Machine Learning, ICML, pages 1310-1320, 2019. Google Scholar
  13. Daniel Cullina, Arjun Nitin Bhagoji, and Prateek Mittal. Pac-learning in the presence of adversaries. In Advances in Neural Information Processing Systems, NeurIPS, pages 230-241, 2018. Google Scholar
  14. Gavin Weiguang Ding, Yash Sharma, Kry Yik Chau Lui, and Ruitong Huang. MMA training: Direct input space margin maximization through adversarial training. In 8th International Conference on Learning Representations, ICLR, 2020. Google Scholar
  15. Dimitrios Diochnos, Saeed Mahloujifar, and Mohammad Mahmoody. Adversarial risk and robustness: General definitions and implications for the uniform distribution. In Advances in Neural Information Processing Systems 31, NeurIPS, pages 10359-10368, 2018. Google Scholar
  16. Uriel Feige, Yishay Mansour, and Robert Schapire. Learning and inference in the presence of corrupted inputs. In Conference on Learning Theory, COLT, pages 637-657, 2015. Google Scholar
  17. Yarin Gal and Lewis Smith. Sufficient conditions for idealised models to have no adversarial examples: a theoretical and empirical study with bayesian neural networks, 2018. URL: http://arxiv.org/abs/1806.00667.
  18. Ian J. Goodfellow, Patrick D. McDaniel, and Nicolas Papernot. Making machine learning robust against adversarial inputs. Commun. ACM, 61(7):56-66, 2018. Google Scholar
  19. Pascale Gourdeau, Varun Kanade, Marta Kwiatkowska, and James Worrell. On the hardness of robust classification. In Advances in Neural Information Processing Systems 32, NeurIPS, pages 7444-7453, 2019. Google Scholar
  20. Ruitong Huang, Bing Xu, Dale Schuurmans, and Csaba Szepesvári. Learning with a strong adversary. CoRR, abs/1511.03034, 2015. URL: http://arxiv.org/abs/1511.03034.
  21. Andrew Ilyas, Shibani Santurkar, Dimitris Tsipras, Logan Engstrom, Brandon Tran, and Aleksander Madry. Adversarial examples are not bugs, they are features. In Advances in Neural Information Processing Systems 32: NeurIPS, pages 125-136, 2019. Google Scholar
  22. Marc Khoury and Dylan Hadfield-Menell. Adversarial training with voronoi constraints. CoRR, abs/1905.01019, 2019. URL: http://arxiv.org/abs/1905.01019.
  23. Samory Kpotufe. k-nn regression adapts to local intrinsic dimension. In Advances in Neural Information Processing Systems, NIPS, pages 729-737, 2011. Google Scholar
  24. Aleksander Madry, Aleksandar Makelov, Ludwig Schmidt, Dimitris Tsipras, and Adrian Vladu. Towards deep learning models resistant to adversarial attacks. In 6th International Conference on Learning Representations, ICLR, 2018. Google Scholar
  25. Omar Montasser, Surbhi Goel, Ilias Diakonikolas, and Nathan Srebro. Efficiently learning adversarially robust halfspaces with noise. arXiv preprint, 2020. URL: http://arxiv.org/abs/2005.07652.
  26. Omar Montasser, Steve Hanneke, and Nathan Srebro. VC classes are adversarially robustly learnable, but only improperly. In Conference on Learning Theory, COLT, pages 2512-2530, 2019. Google Scholar
  27. Chongli Qin, James Martens, Sven Gowal, Dilip Krishnan, Krishnamurthy Dvijotham, Alhussein Fawzi, Soham De, Robert Stanforth, and Pushmeet Kohli. Adversarial robustness through local linearization. In Advances in Neural Information Processing Systems 32, NeurIPS, pages 13824-13833, 2019. Google Scholar
  28. Hadi Salman, Jerry Li, Ilya P. Razenshteyn, Pengchuan Zhang, Huan Zhang, Sébastien Bubeck, and Greg Yang. Provably robust deep learning via adversarially trained smoothed classifiers. In Advances in Neural Information Processing Systems 32, NeurIPS, pages 11289-11300, 2019. Google Scholar
  29. Ludwig Schmidt, Shibani Santurkar, Dimitris Tsipras, Kunal Talwar, and Aleksander Madry. Adversarially robust generalization requires more data. In Advances in Neural Information Processing Systems, NeurIPS, pages 5014-5026, 2018. Google Scholar
  30. Shai Shalev-Shwartz and Shai Ben-David. Understanding Machine Learning: From Theory to Algorithms. Cambridge University Press, 2014. Google Scholar
  31. Ingo Steinwart and Clint Scovel. Fast rates for support vector machines using gaussian kernels. The Annals of Statistics, 35(2):575-607, 2007. Google Scholar
  32. Christian Szegedy, Wojciech Zaremba, Ilya Sutskever, Joan Bruna, Dumitru Erhan, Ian J. Goodfellow, and Rob Fergus. Intriguing properties of neural networks. In 2nd International Conference on Learning Representations, ICLR, 2014. Google Scholar
  33. Dimitris Tsipras, Shibani Santurkar, Logan Engstrom, Alexander Turner, and Aleksander Madry. Robustness may be at odds with accuracy. In 7th International Conference on Learning Representations, ICLR, 2019. Google Scholar
  34. Ruth Urner, Sharon Wulff, and Shai Ben-David. PLAL: cluster-based active learning. In COLT 2013 - The 26th Annual Conference on Learning Theory, pages 376-397, 2013. Google Scholar
  35. Yizhen Wang, Somesh Jha, and Kamalika Chaudhuri. Analyzing the robustness of nearest neighbors to adversarial examples. In Proceedings of the 35th International Conference on Machine Learning, ICML, pages 5120-5129, 2018. Google Scholar
  36. Huanrui Yang, Jingchi Zhang, Hsin-Pai Cheng, Wenhan Wang, Yiran Chen, and Hai Li. Bamboo: Ball-shape data augmentation against adversarial attacks from all directions. In Workshop on Artificial Intelligence Safety 2019 co-located with the Thirty-Third AAAI Conference on Artificial Intelligence, 2019. Google Scholar
  37. Yao-Yuan Yang, Cyrus Rashtchian, Yizhen Wang, and Kamalika Chaudhuri. Adversarial examples for non-parametric methods: Attacks, defenses and large sample limits. CoRR, abs/1906.03310, 2019. URL: http://arxiv.org/abs/1906.03310.
  38. Yao-Yuan Yang, Cyrus Rashtchian, Yizhen Wang, and Kamalika Chaudhuri. Robustness for non-parametric classification: A generic attack and defense. In The 23rd International Conference on Artificial Intelligence and Statistics, AISTATS, pages 941-951, 2020. Google Scholar
  39. Yao-Yuan Yang, Cyrus Rashtchian, Hongyang Zhang, Russ R. Salakhutdinov, and Kamalika Chaudhuri. A closer look at accuracy vs. robustness. In Advances in Neural Information Processing Systems 33 NeurIPS, 2020. Google Scholar
  40. Dong Yin, Kannan Ramchandran, and Peter L. Bartlett. Rademacher complexity for adversarially robust generalization. In Proceedings of the 36th International Conference on Machine Learning,ICML, pages 7085-7094, 2019. Google Scholar
  41. Hang Yu, Aishan Liu, Xianglong Liu, Gengchao Li, Ping Luo, Ran Cheng, Jichen Yang, and Chongzhi Zhang. Pda: Progressive data augmentation for general robustness of deep neural networks, 2020. URL: http://arxiv.org/abs/1909.04839.
  42. Hongyang Zhang, Yaodong Yu, Jiantao Jiao, Eric P. Xing, Laurent El Ghaoui, and Michael I. Jordan. Theoretically principled trade-off between robustness and accuracy. In Proceedings of the 36th International Conference on Machine Learning, ICML, pages 7472-7482, 2019. Google Scholar
Questions / Remarks / Feedback
X

Feedback for Dagstuhl Publishing


Thanks for your feedback!

Feedback submitted

Could not send message

Please try again later or send an E-mail
© 2023-2024 Schloss Dagstuhl – LZI GmbH About DROPS Imprint Privacy Contact