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

Open in new window