Improving the Tightness of Convex Relaxation Bounds for Training Certifiably Robust Classifiers

Open in new window