Proxy Convexity: AUnified Framework for the Analysis of Neural Networks Trained by Gradient Descent

Open in new window