I like "differentiable programming" instead of "neural networks". The crux is that you are programming in such a manner that you can differentiate your functio

I like "differentiable programming" instead of "neural networks". The crux is that you are programming in such a manner that you can differentiate your function w.r.t. parameters which are then tuned.