Yeah, I would expect that encoding stones as "signed liberty count" would train faster/better/stronger. You could imagine a follow-up paper where Go features are supplied.
But, if I think about this on a large scale, wouldn't it be huge to put together a general game-playing program just based on visual-NN and simple MCTS? In the extended section is a comparison of checkers/backgammon/poker/scrabble(thanks!)/... that have been addressed by rollouts, and how those research efforts differed from the tabula rasa / ab initio formulation. The list of games that are played above human caliber now includes deterministic and non-deterministic, and perfect and imperfect information, so there are examples of every type out there. I'm hoping that DeepMind will pursue general game-playing. It would be awesome to see strong AIs emerging from first principles. I would be happy to supply Maven as a sparring partner for Scrabble. I also have a world-class Pineapple Poker player that is based on rollouts and hand-coded features. -----Original Message----- From: Computer-go [mailto:computer-go-boun...@computer-go.org] On Behalf Of Gian-Carlo Pascutto Sent: Wednesday, October 18, 2017 4:38 PM To: computer-go@computer-go.org Subject: Re: [Computer-go] AlphaGo Zero On 18/10/2017 22:00, Brian Sheppard via Computer-go wrote: > A stunning result. The NN uses a standard vision architecture (no Go > adaptation beyond what is necessary to represent the game state). The paper says that Master (4858 rating) uses Go specific features, initialized by SL, and the same technique. Without go features, and without initialization, it's Zero (5185 rating). The obvious question is, what would be the result of using go features and not initializing? I would expect that providing liberties is a useful shortcut (see my remark about game history!). But I'm willing to be surprised :-) -- GCP _______________________________________________ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go _______________________________________________ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go
