China's man-machine battle to determine AlphaGo will play against Ke Jie

[Global Technology Reporter Li Wenyao] On April 7, the General Administration of Sport issued an announcement that the Go artificial intelligence program “AlphaGo” (AlphaGo) will be held in Wuzhen, Zhejiang Province and Ke Jie in late May this year. China's top chess players showdown. This news was further confirmed today, Google China said: May 23-27 will be held in Wuzhen, "China Wuzhen. Weiqi Summit", when AlphaGo will play against Ke Jie in three games.

It is understood that the Wuzhen Weiqi Summit lasted for five days, but unlike the previous five-game three-game system with Li Shishi, AlphaGo and Ke Jie played only three games. According to the introduction of Google and Chinese chess schools, in addition to the three games that AlphaGo and Ke Jie will perform, the Challenge will also arrange a match match and a team match. In the match match, the two players will team up with AlphaGo to challenge the player to understand and cooperate with AlphaGo's unique style. In the team competition, there will be five Chinese top professional players who will form a group to fight against artificial intelligence with collective intelligence.

Last year's game between AlphaGo and Li Shishi

In the match between Korea's Jiushi player Li Shishi and "Alpha Go" held last year, Li Shishi lost 1:4 to the Alpha dog developed by the DeepMind team. This is also the first time that the artificial intelligence program has let human chess players bow in the field of Go. Li Shishi said that this is not a failure of mankind, but his personal failure. However, Li Shishi also frankly stated that if the two sides fight again, he still has no confidence.

It is understood that in the man-machine war last year, Google used a separate computer instead of using a distributed system, and its performance was exactly the same as when it was played in October 2015 with European Champion and professional Go II fan.

"AlphaGO conducts in-depth learning based on countless score data, continues to improve, and enhances its strength through self-simulation, and has conducted many targeted tests before the game. AlphaGO's Go level has been significantly improved over the past few months. "Diamond Hasabis, CEO of Google DeepMind and the father of AlphaGo, once said," and AlphaGo will not be afraid and timid and will not be tired. We used different versions of the program and different chess It was tested by hand and we hope to use this game to discover the disadvantages of AlphaGo."

At the beginning of this year, a mystery player named “Master” appeared on multiple chess platforms and scored 60 uninterrupted successes against the top players of Japan, South Korea and South Korea. Later, Google confessed that the "master" is an upgraded version of the "Alpha Dog" and is more powerful.

The DeepMind team also announced the new "AlphaGo" in late January of this year. Compared with the previous Alpha Dog, the 2.0 version of AlphaGo has continued to deepen its “deep learning” and has abandoned human chess. In addition to continuing to use deep learning methods, it also incorporates reinforcement learning methods and focuses more on the latter. This method allows the artificial intelligence program to start directly from 0. Without artificially summarizing any patterns and training, its success will be completely independent of the human experience and wisdom of Go.

After Li Shishi lost his first match in the 2016 man-machine war, Ke Jie said on Weibo: "Although the Alpha Dog won Li Shishi, it won't win me."

According to the Beijing Evening News, according to the Beijing Evening News, Ke Jie said: “It is not possible to follow the Alpha dog's tactics to disrupt the other side and win the game. It is also necessary to rely on real skills to win.”

Drone Defense

multi band noise source anti drone system,Drone Defense,Confronting fpv drone

Jiangsu Yunbo Intelligent Technology Co., Ltd , https://www.fmodel-ai.com