|
July 01, 2018, 12:40:17 AM |
|
Sorry if posted in wrong place. Just a vague idea I've had last few days, wanted some feedback on practicalities etc.. Recently there has been a big rise in machine learning chess engines, from alphazero, and the open source leelazero (chess). It got me thinking, from the way the machine learns, it plays through many many games, and creates a tree of the most optimal moves from various positions. What if the algorithm that you would solve to check validity was actually a chess game, and this was congregated into some kind of multi-node chess engine, that would get more powerful overtime. Something like solidity I believe could make this somewhat possible, I just dunno how practical it is. Also don't mind if someone wants to take this idea, as I have no where near the technical skills as of yet to undertake such a project. Would like to hear thoughts, thanks.
|