A talk that explains the ideas behind our paper on unrolling for learning problems (esp. that it’s worth integrating a non-differentiable solver as well) is online now. If you get to try it, please let us know how it works!

Published by GE_at_TUM