Dear reviewers

We do this by using a new Lyapunov-style analysis. The setting of our paper is for strongly convex learning problems. We will make sure to clarify this in our final submission. We will include these new experiments in our final submission. We will also include this update on the improved theory in our final submission.