Neural Networks with Cheap Differential Operators

               Ricky T. Q. Chen, David Duvenaud
               University of Toronto, Vector Institute
              {rtqichen,duvenaud}@cs.toronto.edu

                       Abstract
     Gradients of neural networks can be computed efficiently for any architecture,
     but some applications require computing differential operators with higher time
     complexity. We describe a family of neural network architectures that allow easy
     acce ...