Fetching the paper…

Beating the Perils of Non-Convexity: Guaranteed Training of Neural Networks using Tensor Methods · Around