A Catalyst Framework for Minimax Optimization

Neural Information Processing Systems 

We introduce a generic \emph{two-loop} scheme for smooth minimax optimization with strongly-convex-concave objectives.