Alpha dog seized 60 wins machine can really beat the human

Source：Updated：2017-01-09 08:16:01

Who also didn't expect, in 2017 the first refresh human through, unexpectedly is artificial intelligence. 
 
Go to fame last march AlphaGo (alpha), at the beginning of the end to the "Master", to 60 games Internet chess wheel war, beating the almost all of China, South Korea and Japan to excel mogul legend and talent. 
 
Then on the morning local time on January 4, Google DeepMind founder and CEO of demi, sabe si (DemisHassabis) release tweets, said the Master is AlphaGo, officially upgraded AlphaGo come out. 
 
Much like the terminator movies, the plot Arnold Schwarzenegger ruthlessly to jilt to the sentence "I will be back!" 
 
"Challenged" AlphaGo over to master. Head of South Korea's house to say: "after the team behind the Master DeepMind has said that the last decade, want to excel Master various countries and, once the pride of the moment. It should be in this situation, 'bear comparison with chess saint wei-ping nie nine and nine Chang Hao etc and the Master." 
 
 
 
And DeepMind also hope that South Korea cao nine segment and lee nine section of the two "go mogul" battle. However, cao nine section has become a new national congress of the communist party of China and South Korea's ruling party due to the chaos in the party, not to accept the invitation. 
 
And South Korea to excel another leading lee nine more "obvious" reasons for refusing. He said he didn't like the Internet game, rejected the request. 
 
"If only be judged, artificial intelligence, after all, no fatigue, no mistakes, this is different from human's absolute advantage, and can be conducted in accordance with the rules of a particular training conditions, has created good conditions for it, the artificial intelligence to win, not an accident." South Korean chess player Yin Yingmin female (transliteration) four said. 
 
A shocking development speed 
 
In march last year, AlphaGo professional go player ever in South Korea defeated the south Korean lee se-dol, shocked the world. When Yin Yingmin was convinced lee se-dol will to win, even said "if lee se-dol really lose, so we go the mentor, may be will face unemployment". 
 
Yin Yingmin in Seoul, South Korea on go running a training institution. After nearly a year, the reporter visited Yin Yingmin again, she was directed to parents with children go. 
 
"In fact, the level of man-machine war over the past year, not only make I lost my job, but also the focus of many people on the go, more and more parents begin to pay close attention to go to promote children's intelligence development; it may also be benefits of man-machine war to excel." Yin Yingmin said. 
 
And South Korea go institute director also confirmed that "after the man-machine war last year, the advisory watching chess game, and apply for to go Dan also had significantly increased the number of". 
 
For Master's string of good news, Yin Yingmin seemed to cheer up a lot. "To see a lot of China, Japan and South Korea's top players have lost to AlphaGO, watching these their idols in failure, there are some surprise and lost in my heart." She said. 
 
To Master QiFeng and level, Yin Yingmin admitted "is much higher than the previous man-machine AlphaGO of wartime QiFeng healthy and very few mistakes", and said "AlphaGO artificial intelligence development speed is surprising". 
 
Ha sabe, said his team has been dedicated to the advancement AlphaGo performance, they use the new test several times over the past few days the unofficial the game, thanks to the net to Magister and Master account, respectively in Tygem city go net (implement) and FoxGo (fox) running on the server; Also want to thank all the people who watch the game. They are very satisfied with the result of the match. "More importantly, the new AlphaGo to excel in the process of game provided valuable experience." 
 
Hasa in also specially mentioned the game failed in the peel, "p. Posting blog, according to human and artificial intelligence will soon find the mystery of the deep learning together." 
 
"We are looking forward to the official test, and will be held again this year for the chess game, the game appears in the course of mining AlphaGo creativity." Ha sabe si said. 
 
Can so play go? 
 
On December 29, 2016, the Master city first appeared in implement, and in the evening soon made 10 in a row of domestic high level professional guy; 30, at noon, and after many times in a row, with the first person of "Korean" park jeong-hwan levels and nine suspected KeJie "China first". 
 
When KeJie after watching ever wonder: "have never seen such a trick, can so play go?" "Human thousands of years of practical experience evolution, the computer tells us that mankind is all wrong. I think, even with no one to go edge of truth." 
 
Late January 4, in a row against ShenZhen Xu nine, peel, Zhou Rui sheep nine paragraphs, after nine AlphaGo expanded winning streak to 60 games, according to plan in advance and go programs of informal test temporarily. 
 
Network blitz is currently one of the main methods of professional practice, because every step think time is very short, compared with the official match with bonuses, with none in the two ends are more prone to errors, so the upgrade AlphaGo actual go level how still need further test of formal rules of the game. 
 
Master shows human chess players reach the blitz level, but if you give enough human thinking time, whether the results will be different? 
 
Impact has begun to 
 
It is worth mentioning that in December 13, a number of senior Google had made a surprise visit to China, they go institute in China with wei-ping nie, KeJie and others communication, and reached a cooperation agreement. Then in Japan's go institute JournalistClub ceremony, ha sabe si once said: "for AlphaGo and excel in 2017 will be a year full of excitement." People don't realize that in the New Year has not yet arrived, artificial intelligence impact to the ancient art of circulated in one thousand has already begun. 
 
Sesame, founder and CEO of science and technology Zhu Zhi told reporters: "this game compared to the last three games and lee se-dol reflects AlphaGo better stability and confidence. This area is in go, artificial intelligence proves that it is better, but this does not mean that the machine has the wisdom. In fact, machines are better at dealing with these complex rational, so in many specific areas, brought by the machine learning intelligence will gradually change the traditional pattern." 
 
The Master version than AlphaGO whether there are upgrade and improve, and what is optimized to make it become so powerful that specific improvement scheme there is still no open. Artificial intelligence company founder and CEO of the fourth paradigm Dai Wenyuan speculation, "after more than half a year's training, the data quantity should be promoted, trained 8 million games last time, estimated that several hundred million this time. In addition to the last time (to lee se-dol) in the fourth game of the weaknesses of the estimation on the training methods have improved." 
 
Zhihu also triggered discussion, Master whether to upgrade a remark for "the founder of the" future knowledge map users said that the stronger the reason is that the Master is on the basis of AlphaGo adjusted algorithm and parameters, rather than how many games under it, after all, the machine won't really 'learning', it is just in the repeat user set algorithm, provide data for optimizing parameters." 
 
But Dai Wenyuan thinks, upgraded AlphaGo "there may be some random in the game or the random disturbance, the choice is not optimal position move later, try a new possible, will not be stupid to themselves and their". 
 
Association of electrical and electronics engineers (IEEE), a senior member, British el curran, a computer science professor at the university of rochester (KevinCurran) told reporters: "the success of AlphaGo even its creators is also a big surprise. We all know AlphaGo already very advanced, but I didn't think it was so good. So, strengthen the learning plus' tree search algorithm, and effectively increases the odds, indeed this is the most powerful AlphaGo superior to where humans go players. At the same time we should not forget, when Google DeepMind team research the purpose of the artificial intelligence technology, not only to develop a set of chess game tools, but in order to better in the future the application of artificial intelligence technology to the medical treatment, robot, driverless cars, agriculture, finance and other industry for the service of mankind. It is now the development of science and technology of the best places are amazing." 
 
Poet t.s. Eliot once said, "last words belong to last year's language, if waiting for next year another voice." It used in artificial intelligence instant no change.