@ CCTV CCTV5 reported that Google’s Deepmind company recently announced in Munich, Germany DLD Innovation Conference, will launch the real 2 edition AlphaGo.
version of "a dog" will officially abandon human chess, only by learning the computer depth way to grow — Exploration of "extreme deep learning" approach, to reach "go God" realm.
2.0 version of AlphaGo, will learn through the mathematical model on self and depth, under real artificial intelligence (AI) go.
opened in 2017, a dog incarnation of the Master 60 only The wind puffs the clouds away. wins, a draw or because the players dropped.
was "grandmaster Nie Weiping said:" go away as we imagined so simple, there is a huge space waiting for us to dig. A dog is good, Master worth mentioning, is "go God" sent to the human guide."
and Ke Jie said he did not sleep a night, the human Millennium combat exercise evolution, the computer tells us that human beings are all wrong……"
since the human wisdom now go into the wind, so artificial intelligence will bring about the change of go?
DeepMind actually had the idea of abandoning human chess routines, AI go your own way.
the company founder Hassabis earlier in an interview revealed that they are trying to train a team has not studied the human artificial intelligence chess.
and Master kill the Quartet, is also speculation that Master could not study human chess.
but whether it is a human routine or a faction, a French dog is more frightening or its ability to learn.
In October 2015
, a dog in the game with the European champions fan Hui in the fair in the case of 5 to 0 victory over. Fan Hui in a media interview that a dog is terrible "never make mistakes".
in the fan Hui and a dog five set battle chess are put on the network, there are many amateur go master comment on the network, many people think that this fan Hui chess is not good.
but at that time, "nature" magazine wrote, "people have a disadvantage in playing chess, after a long time, they will make mistakes, but the machine will not. And humans may be able to play 1000 games a year, but the machine can play the game in the 1 million game."
"AI go learning ability is very strong, like the original" deep blue "first lost to Kasparov, but less than second times a year the game won the game." ;