Convergence and Implicit Regularization Properties of Gradient Descent for Deep Residual Networks

Open in new window