Randomizing SVM against adversarial attacks under uncertainty

Yan Chen; Wei Wang; Xiangliang Zhang

doi:10.1007/978-3-319-93040-4_44

Randomizing SVM against adversarial attacks under uncertainty

Yan Chen, Wei Wang, Xiangliang Zhang^*

^*Corresponding author for this work

Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

3 Scopus citations

Abstract

Robust machine learning algorithms have been widely studied in adversarial environments where the adversary maliciously manipulates data samples to evade security systems. In this paper, we propose randomized SVMs against generalized adversarial attacks under uncertainty, through learning a classifier distribution rather than a single classifier in traditional robust SVMs. The randomized SVMs have advantages on better resistance against attacks while preserving high accuracy of classification, especially for non-separable cases. The experimental results demonstrate the effectiveness of our proposed models on defending against various attacks, including aggressive attacks with uncertainty.

Original language	English (US)
Title of host publication	Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Proceedings
Editors	Geoffrey I. Webb, Dinh Phung, Mohadeseh Ganji, Lida Rashidi, Vincent S. Tseng, Bao Ho
Publisher	Springer Verlag
Pages	556-568
Number of pages	13
ISBN (Print)	9783319930398
DOIs	https://doi.org/10.1007/978-3-319-93040-4_44
State	Published - 2018
Event	22nd Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2018 - Melbourne, Australia Duration: Jun 3 2018 → Jun 6 2018

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	10939 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	22nd Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2018
Country/Territory	Australia
City	Melbourne
Period	06/3/18 → 06/6/18

Bibliographical note

Publisher Copyright:
© Springer International Publishing AG, part of Springer Nature 2018.

Keywords

Adversarial learning
Randomization
Robust SVM

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-319-93040-4_44

Cite this

Chen, Y., Wang, W., & Zhang, X. (2018). Randomizing SVM against adversarial attacks under uncertainty. In G. I. Webb, D. Phung, M. Ganji, L. Rashidi, V. S. Tseng, & B. Ho (Eds.), Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Proceedings (pp. 556-568). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10939 LNAI). Springer Verlag. https://doi.org/10.1007/978-3-319-93040-4_44

Chen, Yan ; Wang, Wei ; Zhang, Xiangliang. / Randomizing SVM against adversarial attacks under uncertainty. Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Proceedings. editor / Geoffrey I. Webb ; Dinh Phung ; Mohadeseh Ganji ; Lida Rashidi ; Vincent S. Tseng ; Bao Ho. Springer Verlag, 2018. pp. 556-568 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{a33e7c1054f44dc58e29f1ee9d3c6d31,

title = "Randomizing SVM against adversarial attacks under uncertainty",

abstract = "Robust machine learning algorithms have been widely studied in adversarial environments where the adversary maliciously manipulates data samples to evade security systems. In this paper, we propose randomized SVMs against generalized adversarial attacks under uncertainty, through learning a classifier distribution rather than a single classifier in traditional robust SVMs. The randomized SVMs have advantages on better resistance against attacks while preserving high accuracy of classification, especially for non-separable cases. The experimental results demonstrate the effectiveness of our proposed models on defending against various attacks, including aggressive attacks with uncertainty.",

keywords = "Adversarial learning, Randomization, Robust SVM",

author = "Yan Chen and Wei Wang and Xiangliang Zhang",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing AG, part of Springer Nature 2018.; 22nd Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2018 ; Conference date: 03-06-2018 Through 06-06-2018",

year = "2018",

doi = "10.1007/978-3-319-93040-4_44",

language = "English (US)",

isbn = "9783319930398",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "556--568",

editor = "Webb, {Geoffrey I.} and Dinh Phung and Mohadeseh Ganji and Lida Rashidi and Tseng, {Vincent S.} and Bao Ho",

booktitle = "Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Proceedings",

address = "Germany",

}

Chen, Y, Wang, W & Zhang, X 2018, Randomizing SVM against adversarial attacks under uncertainty. in GI Webb, D Phung, M Ganji, L Rashidi, VS Tseng & B Ho (eds), Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10939 LNAI, Springer Verlag, pp. 556-568, 22nd Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2018, Melbourne, Australia, 06/3/18. https://doi.org/10.1007/978-3-319-93040-4_44

Randomizing SVM against adversarial attacks under uncertainty. / Chen, Yan; Wang, Wei; Zhang, Xiangliang.
Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Proceedings. ed. / Geoffrey I. Webb; Dinh Phung; Mohadeseh Ganji; Lida Rashidi; Vincent S. Tseng; Bao Ho. Springer Verlag, 2018. p. 556-568 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10939 LNAI).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Randomizing SVM against adversarial attacks under uncertainty

AU - Chen, Yan

AU - Wang, Wei

AU - Zhang, Xiangliang

N1 - Publisher Copyright: © Springer International Publishing AG, part of Springer Nature 2018.

PY - 2018

Y1 - 2018

N2 - Robust machine learning algorithms have been widely studied in adversarial environments where the adversary maliciously manipulates data samples to evade security systems. In this paper, we propose randomized SVMs against generalized adversarial attacks under uncertainty, through learning a classifier distribution rather than a single classifier in traditional robust SVMs. The randomized SVMs have advantages on better resistance against attacks while preserving high accuracy of classification, especially for non-separable cases. The experimental results demonstrate the effectiveness of our proposed models on defending against various attacks, including aggressive attacks with uncertainty.

AB - Robust machine learning algorithms have been widely studied in adversarial environments where the adversary maliciously manipulates data samples to evade security systems. In this paper, we propose randomized SVMs against generalized adversarial attacks under uncertainty, through learning a classifier distribution rather than a single classifier in traditional robust SVMs. The randomized SVMs have advantages on better resistance against attacks while preserving high accuracy of classification, especially for non-separable cases. The experimental results demonstrate the effectiveness of our proposed models on defending against various attacks, including aggressive attacks with uncertainty.

KW - Adversarial learning

KW - Randomization

KW - Robust SVM

UR - http://www.scopus.com/inward/record.url?scp=85049373377&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-93040-4_44

DO - 10.1007/978-3-319-93040-4_44

M3 - Conference contribution

AN - SCOPUS:85049373377

SN - 9783319930398

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 556

EP - 568

BT - Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Proceedings

A2 - Webb, Geoffrey I.

A2 - Phung, Dinh

A2 - Ganji, Mohadeseh

A2 - Rashidi, Lida

A2 - Tseng, Vincent S.

A2 - Ho, Bao

PB - Springer Verlag

T2 - 22nd Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2018

Y2 - 3 June 2018 through 6 June 2018

ER -

Chen Y, Wang W, Zhang X. Randomizing SVM against adversarial attacks under uncertainty. In Webb GI, Phung D, Ganji M, Rashidi L, Tseng VS, Ho B, editors, Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Proceedings. Springer Verlag. 2018. p. 556-568. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-93040-4_44

Randomizing SVM against adversarial attacks under uncertainty

Abstract

Publication series

Conference

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this