Towards evolving robust neural architectures to defend from adversarial attacks

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Neural networks are known to misclassify a class of subtly modified images known as adversarial samples. Recently, numerous defences have been proposed against these adversarial samples; however, none have improved the robustness of neural networks consistently. Here, we propose to use adversarial samples as a function evaluation to explore for robust neural architectures that can resist such attacks. Experiments on existing neural architecture search algorithms from the literature reveal that although accurate, they are not able to find robust architectures. An essential cause for this lies in their confined search space. We were able to evolve an architecture that is intrinsically accurate on adversarial samples by creating a novel neural architecture search. Thus, the results here demonstrate that more robust architectures exist as well as opens up a new range of possibilities for the development and exploration of neural networks using neural architecture search.

Original languageEnglish
Title of host publicationGECCO 2020 Companion - Proceedings of the 2020 Genetic and Evolutionary Computation Conference Companion
PublisherAssociation for Computing Machinery, Inc
Pages135-136
Number of pages2
ISBN (Electronic)9781450371278
DOIs
Publication statusPublished - Jul 8 2020
Event2020 Genetic and Evolutionary Computation Conference, GECCO 2020 - Cancun, Mexico
Duration: Jul 8 2020Jul 12 2020

Publication series

NameGECCO 2020 Companion - Proceedings of the 2020 Genetic and Evolutionary Computation Conference Companion

Conference

Conference2020 Genetic and Evolutionary Computation Conference, GECCO 2020
CountryMexico
CityCancun
Period7/8/207/12/20

All Science Journal Classification (ASJC) codes

  • Computational Mathematics

Fingerprint Dive into the research topics of 'Towards evolving robust neural architectures to defend from adversarial attacks'. Together they form a unique fingerprint.

  • Cite this

    Kotyan, S., & Vargas, D. V. (2020). Towards evolving robust neural architectures to defend from adversarial attacks. In GECCO 2020 Companion - Proceedings of the 2020 Genetic and Evolutionary Computation Conference Companion (pp. 135-136). (GECCO 2020 Companion - Proceedings of the 2020 Genetic and Evolutionary Computation Conference Companion). Association for Computing Machinery, Inc. https://doi.org/10.1145/3377929.3389962