Photo caption: On the 23rd, Ke Jie was in the man-machine go tournament. [This newspaper went to Wuzhen special correspondent Li Qiaoyi special correspondent Wei Yunfeng reporter Zhaojue Zhang Wang] only a poor 1/4 goal! In the human-machine go tournament held on the 23rd, Ke Jie, the number one Chinese chess player, lost his first game to Go AlphaGo. Despite defeating only the smallest gap in Go rules, Ke Jie himself admitted after the game that Alpha Dogs are more and more like God. Various players and artificial intelligence experts generally believe that it is inevitable that artificial intelligence programs such as alpha-fascias have defeated human players, and that it is more necessary to focus on how to apply this capability of artificial intelligence to more societies. field. There has been no good chance of winning chess Ke Jie took the lead in the game, but soon got into a bitter fight. A member of the chess field watching the game told the Global Times reporter that Ke Jie played well as a whole, but he could see that this game was very difficult. Ke Jie was quite nervous. There were many small gestures like scratching his head and shaking his head. His expression was also very rich. From time to time, he smiled and smiled. The person said that most players watching the game did not think this game is a suspense, Ke Jie really difficult to win the upgraded Alpha dog. Half an hour before the end of the game, most of the players in the chess community have basically determined that Ke Jie does not have a chance to come back. Compared to Ke Jie's game with the Alpha Dog, I am more concerned about the situation of the match between Alpha Dogs and humans. Artificial intelligence and people collaborate together to make progress on Go. Prior to this, national player Gu Lili had predicted that Ke Jie could only win 10% of the game. Nie Weiping, a chess player, also stated that Alpha Dog's layout concept is far better than human beings, and Ke Jie’s move was unknowingly made. The Alpha Dog had a few moves that were completely unexpected to the professional chess player. At that time, he looked strange, but after a closer look, he did a good job. What is the progress of the Alpha 2.0 dog? Ke Jie also acknowledged at the press conference after the game that the Alpha Dog really did a wonderful job and changed our views on the game of chess. There is no game that can't be followed. Feeling Alpha Dog and two judges last year, at the time felt that its chess was very close to people, and now it feels more and more like Go. Zhang Yunquan, a researcher at the Institute of Computing Technology of the Chinese Academy of Sciences and director of the National Supercomputer Jinan Center, told the Global Times reporter that an upgrade of the alpha dog was mainly due to the use of a new TPU specifically designed for machine learning. . At the press conference after the game, Alpha Dog founder Hasabis said that Alpha Dog used 10 TPUs to run on Google Cloud. Compared with the version of Li Shishi, the new version of Alpha Dog has reduced the computational complexity to 1/10 and has a stronger self-playing ability. It is easier to operate and consumes less power. Zhang Yunquan said that Alpha Dog 2.0 does not require human chess as before, but learns through self-opposition. Zhang Yunquan believes that this can be understood as a kind of troubleshooting method. Through self-diagnosis, the wrong choices in different situations are eliminated, leaving others unable to find their own weaknesses. However, Alpha Dogs are still in the supervised learning stage and cannot reach the level of unsupervised learning. It still needs humans to tell it how to learn. Zhang Yunquan said that the idea of ​​artificial intelligence playing chess is not the same as that of a chess player. In the chess world there is a saying called gold corner silver grass belly. In general, the human chess player will choose to walk or go in order to reduce the amount of calculation, and artificial intelligence is not afraid of complex calculations, may choose to fall in the grass belly, pull the war situation To the center of the board to play the advantage of power. This is exactly what makes Ke Jie and other players lament that there is no reason why the following games are impossible. During the game, Global Times reporter also noticed a detail. In the final stage of the game, Alpha Dog's speed has gradually slowed down from the first ten seconds. Zhang Yunquan told the “Global Times†reporter that the more and more pieces of Go, the more complex the computational complexity of Alpha Dogs, which is the distinguishing feature of artificial intelligence from human players. Alpha Dog's Future Path According to the game setting, there will be several match-ups in the next few days. Even if Ke Jie wins one of the games, Ke Jie's room for improvement is limited, while the Alpha Dog has huge room for improving computing power. Xiangyang Electronic Information Industry Research Center industry research expert Xiangyang was interviewed by a Global Times reporter and said that the general trend is that the computational power of artificial intelligence has exceeded humanity, especially in the field of Go. If you completely defeat humans in the field of chess, what will the Alpha Dog do next? Earlier reports said that it may continue to modify and will challenge the more difficult computer game StarCraft II. Zhang Yunquan said that Alpha Dogs rely on technologies that are universal technologies. In addition to such challenges, it can also be used in other areas where the algorithm is applicable. In particular, deep learning algorithms do a good job of comprehending images, and they need a great improvement in their ability to understand. Later, in areas where the demand for such technologies is relatively high, such as transportation planning and decision-making, the predictions used for the situation will be of great use. ▲ Offshore Wind Turbine,Custom Wind Turbine,Best Offshore Wind Turbine,Intelligence Wind Turbine CRRC SHANDONG CO., LTD. , https://www.crrcsd.com