There's a long backlog of major achievements in physics that haven't gotten a Nobel Prize in Physics . Giving the prize to something that has essentially nothing to do with physics is just a slap in the face to the physics community.
The latter makes a bit more sense, computer science wasn't really a thing when Alfred Nobel was around, but mathematics certainly was! It seems like it would be perfectly reasonable to add a category for math, and I think Neural Nets would fit in there considerably better.