Evasion attack of multi-class linear classifiers

Han Xiao; Thomas Stibor; Claudia Eckert

doi:10.1007/978-3-642-30217-6_18

Evasion attack of multi-class linear classifiers

Han Xiao, Thomas Stibor, Claudia Eckert

Informatics 20 - Chair of IT Security

Technical University of Munich

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

4 Scopus citations

Abstract

Machine learning has yield significant advances in decision-making for complex systems, but are they robust against adversarial attacks? We generalize the evasion attack problem to the multi-class linear classifiers, and present an efficient algorithm for approximating the optimal disguised instance. Experiments on real-world data demonstrate the effectiveness of our method.

Original language	English
Title of host publication	Advances in Knowledge Discovery and Data Mining - 16th Pacific-Asia Conference, PAKDD 2012, Proceedings
Pages	207-218
Number of pages	12
Edition	PART 1
DOIs	https://doi.org/10.1007/978-3-642-30217-6_18
State	Published - 2012
Event	16th Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2012 - Kuala Lumpur, Malaysia Duration: 29 May 2012 → 1 Jun 2012

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Number	PART 1
Volume	7301 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	16th Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2012
Country/Territory	Malaysia
City	Kuala Lumpur
Period	29/05/12 → 1/06/12

Access to Document

10.1007/978-3-642-30217-6_18

Cite this

Xiao, H., Stibor, T., & Eckert, C. (2012). Evasion attack of multi-class linear classifiers. In Advances in Knowledge Discovery and Data Mining - 16th Pacific-Asia Conference, PAKDD 2012, Proceedings (PART 1 ed., pp. 207-218). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7301 LNAI, No. PART 1). https://doi.org/10.1007/978-3-642-30217-6_18

Xiao, Han ; Stibor, Thomas ; Eckert, Claudia. / Evasion attack of multi-class linear classifiers. Advances in Knowledge Discovery and Data Mining - 16th Pacific-Asia Conference, PAKDD 2012, Proceedings. PART 1. ed. 2012. pp. 207-218 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 1).

@inproceedings{4d414aa6274449e4960b5e5cd5a96460,

title = "Evasion attack of multi-class linear classifiers",

abstract = "Machine learning has yield significant advances in decision-making for complex systems, but are they robust against adversarial attacks? We generalize the evasion attack problem to the multi-class linear classifiers, and present an efficient algorithm for approximating the optimal disguised instance. Experiments on real-world data demonstrate the effectiveness of our method.",

author = "Han Xiao and Thomas Stibor and Claudia Eckert",

year = "2012",

doi = "10.1007/978-3-642-30217-6_18",

language = "English",

isbn = "9783642302169",

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

number = "PART 1",

pages = "207--218",

booktitle = "Advances in Knowledge Discovery and Data Mining - 16th Pacific-Asia Conference, PAKDD 2012, Proceedings",

edition = "PART 1",

note = "16th Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2012 ; Conference date: 29-05-2012 Through 01-06-2012",

}

Xiao, H, Stibor, T & Eckert, C 2012, Evasion attack of multi-class linear classifiers. in Advances in Knowledge Discovery and Data Mining - 16th Pacific-Asia Conference, PAKDD 2012, Proceedings. PART 1 edn, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 1, vol. 7301 LNAI, pp. 207-218, 16th Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2012, Kuala Lumpur, Malaysia, 29/05/12. https://doi.org/10.1007/978-3-642-30217-6_18

Evasion attack of multi-class linear classifiers. / Xiao, Han; Stibor, Thomas; Eckert, Claudia.
Advances in Knowledge Discovery and Data Mining - 16th Pacific-Asia Conference, PAKDD 2012, Proceedings. PART 1. ed. 2012. p. 207-218 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7301 LNAI, No. PART 1).

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

TY - GEN

T1 - Evasion attack of multi-class linear classifiers

AU - Xiao, Han

AU - Stibor, Thomas

AU - Eckert, Claudia

PY - 2012

Y1 - 2012

N2 - Machine learning has yield significant advances in decision-making for complex systems, but are they robust against adversarial attacks? We generalize the evasion attack problem to the multi-class linear classifiers, and present an efficient algorithm for approximating the optimal disguised instance. Experiments on real-world data demonstrate the effectiveness of our method.

AB - Machine learning has yield significant advances in decision-making for complex systems, but are they robust against adversarial attacks? We generalize the evasion attack problem to the multi-class linear classifiers, and present an efficient algorithm for approximating the optimal disguised instance. Experiments on real-world data demonstrate the effectiveness of our method.

UR - http://www.scopus.com/inward/record.url?scp=84861429312&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-30217-6_18

DO - 10.1007/978-3-642-30217-6_18

M3 - Conference contribution

AN - SCOPUS:84861429312

SN - 9783642302169

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

SP - 207

EP - 218

BT - Advances in Knowledge Discovery and Data Mining - 16th Pacific-Asia Conference, PAKDD 2012, Proceedings

T2 - 16th Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2012

Y2 - 29 May 2012 through 1 June 2012

ER -

Xiao H, Stibor T, Eckert C. Evasion attack of multi-class linear classifiers. In Advances in Knowledge Discovery and Data Mining - 16th Pacific-Asia Conference, PAKDD 2012, Proceedings. PART 1 ed. 2012. p. 207-218. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 1). doi: 10.1007/978-3-642-30217-6_18

Evasion attack of multi-class linear classifiers

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Cite this