Training variational quantum algorithms with random gate activation

Abstract

A new training strategy with progressive random gate activation for variational quantum algorithms greatly mitigates barren plateaus and avoids the local minima while being quantum resource efficient and noise resilient.