Width Provably Matters in Optimization for Deep Linear Neural Networks.

Simon S. Du, Wei Hu

Browse the full ICML paper archive.