Global Optimality Beyond Two Layers: Training Deep ReLU Networks via Convex Programs.

Tolga Ergen, Mert Pilanci

Browse the full ICML paper archive.