Văn bản này: Robust optimization for adversarial learning with finite sample complexity guarantees