On the Proof of Global Convergence of Gradient Descent for Deep ReLU Networks with Linear Widths

Open in new window