The list of AlphaGo's imperfections I found


所有跟贴·加跟贴·新语丝读书论坛

送交者: conner 于 2016-03-14, 14:54:31:

1. The deep mind team leader said AlphaGo did not learn from the matches with Lee. That is not what we understand about intelligence and learning. If that is true, then I predict AlphaGo will continue to make the same error with a big chance.

2. AlphaGo does not have a user model. It treats every opponent the same, even if it is a random generator. In this way, AlphaGo is stable, but it wastes resource and is not the human way to handle problems.




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