residual neural networks (ResNets) are a promising class of deep neural
networks that have shown excellent performance for a number of learning tasks,
e.g., image classification and recognition. Mathematically, ResNet
architectures can be interpreted as forward Euler discretizations of