hassabis 的原话


所有跟贴·加跟贴·新语丝读书论坛

送交者: pgss 于 2016-03-13, 23:55:19:

https://www.youtube.com/watch?v=yCALyQRN3hw

6:09 分左右开始

i just want to say SPECIFICALLY, that alphago was not trained specifically for, say Lee Sedol,
we trained it in a general way, and in fact the game that we used, the human games that we used, to start the training, were STRONG AMATEUR GAMES, from the internet go servers, and there were no Lee Sedol games in those database. and then u know already, that the way that alphago grows stronger, is to play (against) itself. So in fact though, i think actually it is quite equal in terms of information (symmetry), that we didn't train on Lee Sedol's game.

I just add one mroe thing, that even if we wanted to do that, alphago would have need to acquire millions of games, to train itself.

so...a few hundred games, or a thousand game, is not going to be enough, to, to change how alphago really plays. because it needs millions and millions, tens of millions of games.

(my transcipt is quite rough, may have some minor mistakes)




所有跟贴:


加跟贴

笔名: 密码: 注册笔名请按这里

标题:

内容: (BBCode使用说明