Let's merge domain-specific models with generic learning. With Julia it is easy to combine these different worlds, differential equations on the one hand and (deep) neural networks on the other. The ...