Bridged adversarial training
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Kim, Hoki | - |
dc.contributor.author | Lee, Woojin | - |
dc.contributor.author | Lee, Sungyoon | - |
dc.contributor.author | Lee, Jaewook | - |
dc.date.accessioned | 2024-06-26T09:30:17Z | - |
dc.date.available | 2024-06-26T09:30:17Z | - |
dc.date.issued | 2023-10 | - |
dc.identifier.issn | 0893-6080 | - |
dc.identifier.issn | 1879-2782 | - |
dc.identifier.uri | https://scholarworks.bwise.kr/hanyang/handle/2021.sw.hanyang/194772 | - |
dc.description.abstract | Adversarial robustness is considered a required property of deep neural networks. In this study, we discover that adversarially trained models might have significantly different characteristics in terms of margin and smoothness, even though they show similar robustness. Inspired by the observation, we investigate the effect of different regularizers and discover the negative effect of the smoothness regularizer on maximizing the margin. Based on the analyses, we propose a new method called bridged adversarial training that mitigates the negative effect by bridging the gap between clean and adversarial examples. We provide theoretical and empirical evidence that the proposed method provides stable and better robustness, especially for large perturbations. | - |
dc.format.extent | 17 | - |
dc.language | 영어 | - |
dc.language.iso | ENG | - |
dc.publisher | Elsevier Ltd | - |
dc.title | Bridged adversarial training | - |
dc.type | Article | - |
dc.publisher.location | 영국 | - |
dc.identifier.doi | 10.1016/j.neunet.2023.08.024 | - |
dc.identifier.scopusid | 2-s2.0-85170099365 | - |
dc.identifier.wosid | 001072083900001 | - |
dc.identifier.bibliographicCitation | Neural Networks, v.167, pp 266 - 282 | - |
dc.citation.title | Neural Networks | - |
dc.citation.volume | 167 | - |
dc.citation.startPage | 266 | - |
dc.citation.endPage | 282 | - |
dc.type.docType | Article | - |
dc.description.isOpenAccess | N | - |
dc.description.journalRegisteredClass | scie | - |
dc.description.journalRegisteredClass | scopus | - |
dc.relation.journalResearchArea | Computer Science | - |
dc.relation.journalResearchArea | Neurosciences & Neurology | - |
dc.relation.journalWebOfScienceCategory | Computer Science, Artificial Intelligence | - |
dc.relation.journalWebOfScienceCategory | Neurosciences | - |
dc.subject.keywordPlus | CLASSIFICATION | - |
dc.subject.keywordAuthor | Adversarial defense | - |
dc.subject.keywordAuthor | Adversarial robustness | - |
dc.subject.keywordAuthor | Adversarial training | - |
dc.subject.keywordAuthor | Neural networks | - |
dc.identifier.url | https://www.sciencedirect.com/science/article/pii/S0893608023004434?via%3Dihub | - |
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.
222, Wangsimni-ro, Seongdong-gu, Seoul, 04763, Korea+82-2-2220-1365
COPYRIGHT © 2021 HANYANG UNIVERSITY.
Certain data included herein are derived from the © Web of Science of Clarivate Analytics. All rights reserved.
You may not copy or re-distribute this material in whole or in part without the prior written consent of Clarivate Analytics.