@inproceedings{polylogarithmicwidthsufficesforgradient,
  title = {Polylogarithmic width suffices for gradient descent to achieve arbitrarily small test error with shallow ReLU networks},
  author = {Ziwei Ji and Matus Telgarsky},
  year = {2019},
  booktitle = {ICLR 2020 1},
  eprint = {1909.12292},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1909.12292v4},
}