NIPS Proceedingsβ

Xiao Wang

2 Papers

  • First-order methods almost always avoid saddle points: The case of vanishing step-sizes (2019)
  • Understanding Weight Normalized Deep Neural Networks with Rectified Linear Units (2018)