JMLR
2012
11 years 7 months ago
2012
We present a minimax framework for classification that considers stochastic adversarial perturbations to the training data. We show that for binary classification it is equivale...