Exponential Convergence of Sinkhorn Under Regularization Scheduling

07/02/2022
by   Jingbang Chen, et al.
0

The optimal transport (OT) problem or earth mover's distance has wide applications to machine learning and statistical problems. In 2013, Cuturi [Cut13] introduced the Sinkhorn algorithm for matrix scaling as a method to compute solutions to regularized optimal transport problems. In this paper, we introduce the exponentially converging Sinkhorn algorithm ExpSinkhorn, by modifying the Sinkhorn algorithm to adaptively double the regularization parameter η periodically. The resulting method has the benefits of automatically finding the regularization parameter η for a desired accuracy ε, and has iteration complexity depending on log(1/ε), as opposed to ε^-O(1) as in previous analyses [Cut13] [ANWR17]. We also show empirically that our algorithm is efficient and robust.

READ FULL TEXT

Please sign up or login with your details

Forgot password? Click here to reset