You've described the architecture of a ~15-year old engine called NeuroGo: http://webdocs.cs.ualberta.ca/~emarkus/neurogo/ http://webdocs.cs.ualberta.ca/~emarku

You've described the architecture of a ~15-year old engine called NeuroGo: http://webdocs.cs.ualberta.ca/~emarkus/neurogo/ http://webdocs.cs.ualberta.ca/~emarkus/neurogo/ All of the different solvers ran on the same machine, as far as I know. It turns out that the Monte Carlo engines are doing far, far better than the grand hierarchical neural net designs of the past, though.