1. Near-Optimal Evasion of Convex-Inducing Classifiers
- Author
-
Nelson, Blaine, Rubinstein, Benjamin I. P., Huang, Ling, Joseph, Anthony D., Lau, Shing-hon, Lee, Steven J., Rao, Satish, Tran, Anthony, and Tygar, J. D.
- Subjects
Computer Science - Learning ,Computer Science - Cryptography and Security - Abstract
Classifiers are often used to detect miscreant activities. We study how an adversary can efficiently query a classifier to elicit information that allows the adversary to evade detection at near-minimal cost. We generalize results of Lowd and Meek (2005) to convex-inducing classifiers. We present algorithms that construct undetected instances of near-minimal cost using only polynomially many queries in the dimension of the space and without reverse engineering the decision boundary., Comment: 8 pages; to appear at AISTATS'2010
- Published
- 2010