Improved Overparametrization Bounds for Global Convergence of Stochastic Gradient Descent for Shallow Neural Networks

Open in new window