Evasion Attack of Multi-Class Linear Classifiers
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.
Evasion Attack of Multi-Class Linear Classifiers
| Authors: | Han Xiao, Thomas Stibor, and Claudia Eckert |
| Year/month: | 2012/5 |
| Booktitle: | Proceedings of the 16th Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD) |
| Address: | Malaysia |
| Publisher: | Springer |
| Note: | AR: 37%) |
| Fulltext: | HanXiao2.pdf |
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. | |
Bibtex:
@conference { hanxiao2012-evasion,author = { Han Xiao and Thomas Stibor and Claudia Eckert },
title = { Evasion Attack of Multi-Class Linear Classifiers },
year = { 2012 },
month = { May },
booktitle = { Proceedings of the 16th Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD) },
address = { Malaysia },
note = { AR: 37%) },
publisher = { Springer },
url = {https://www.sec.in.tum.de/i20/publications/evasion-attack-of-multi-class-linear-classifiers/@@download/file/HanXiao2.pdf}
}