1: \begin{abstract}

2: We study the sample complexity of entropic optimal

3: transport

4: in high dimensions using computationally

5: efficient plug-in estimators. We significantly advance the state

6: of the art by establishing dimension-free, parametric rates

7: for estimating various quantities of interest, including  the entropic regression function which is a natural analog to the optimal transport map. As an application, we propose a practical model

8: for transfer learning based

9: on entropic optimal transport and establish parametric rates of convergence for  nonparametric  regression and classification.

10: \end{abstract}

11: