Practical Session on Approximations with (Deep) Neural Networks

Practical Session


We start with a basic introduction to the deep learning framework Tensorflow. We then review some recent results by D. Yarotsky [1] regarding the approximation of functions in Sobolev spaces by ReLU neural networks. Finally, we will evaluate these theoretical results numerically in a hands-on session, building upon the learned Tensorflow techniques.

Joint session with Rafael Raisenhofer.

[1] Yarotsky, D. (2017). Error bounds for approximations with deep ReLU networks. Neural Networks, 94, 103-113.

Oct 16, 2018
Mathematisches Forschungsinstitut Oberwolfach
Jan Macdonald
Jan Macdonald

My research is at the interface of applied and computational mathematics and scientific machine learning. I am interested in inverse problems, signal- and image recovery, and robust and interpretable deep learning.