On the Promise of the Stochastic Generalized Gauss-Newton Method for Training DNNs

Open in new window