1: \begin{abstract}

2: We develop a general framework for data-driven approximation of

3: input-output maps between infinite-dimensional spaces. The proposed

4: approach is motivated by the recent successes of neural networks and deep learning, in combination

5: with ideas from model reduction. This combination results in a neural

6: network approximation which, in principle, is defined on infinite-dimensional

7: spaces and, in practice, is robust to the dimension of finite-dimensional

8: approximations of these spaces required for computation. For a class of 

9: input-output maps, and suitably  chosen probability measures on the inputs, 

10: we prove convergence of the proposed approximation methodology. 

11: We also include numerical experiments which 

12: demonstrate the effectiveness of the method, showing convergence and

13: robustness of the approximation scheme with respect to

14: the size of the discretization, and compare it

15: with existing algorithms from the literature; our examples

16: include the mapping from coefficient to solution in a divergence form 

17: elliptic partial differential equation (PDE) problem, 

18: and the solution operator for viscous Burgers' equation.

19: \end{abstract}

20: