The Computational Advantage of Depth: Learning High-Dimensional Hierarchical Functions with Gradient Descent

Open in new window