Adversarially Robust Learning with Unknown Perturbation Sets
arXiv:2102.02145
Abstract
We study the problem of learning predictors that are robust to adversarial examples with respect to an unknown perturbation set, relying instead on interaction with an adversarial attacker or access to attack oracles, examining different models for such interactions. We obtain upper bounds on the sample complexity and upper and lower bounds on the number of required interactions, or number of successful attacks, in different interaction models, in terms of the VC and Littlestone dimensions of the hypothesis class of predictors, and without any assumptions on the perturbation set.
References in corpus (6)
- Provable defenses against adversarial examples via the convex outer adversarial polytope
- Certified Adversarial Robustness via Randomized Smoothing
- Exploring the Landscape of Spatial Robustness
- Testing Robustness Against Unforeseen Adversaries
- Black-box Certification and Learning under Adversarial Perturbations
- Online Learning with Simple Predictors and a Combinatorial Characterization of Minimax in 0/1 Games