Sample Complexity of Robust Linear Classification on Separated Data

                Robi Bhattacharjee 1 Somesh Jha 2 Kamalika Chaudhuri 1

             Abstract                 thus aims to find a classifier that maximizes accuracy on
                                  examples that are distance r or more from the decision
   We consider the sample complexity of learning         boundary, where distances are measured according to the
   with adversarial robustness. Most prior theoretical      m ...