Towards Problem-dependent Optimal Learning Rates

Open in new window