Neural Networks are Convex Regularizers: Exact Polynomial-time Convex Optimization Formulations for Two-layer Networks.

Mert Pilanci, Tolga Ergen

Browse the full ICML paper archive.