Could rosetta@home use a neural network to improve it's ability to predict a protein structure and use less CPU time? I think it be cool to implement some AI algorithm to Rosetta, to make it more efficient. I don't know how to do this or if it could work, I was just thinking about it.