DeepMind
DeepMind, now part of the Alphabet holding company, has been working for several years on the AlphaGo software platform. In the past year, she won a go in one of the world champions Lee Sedol. Thus, the developers of this neural network were able to prove that the computer is quite able to compete with a person in areas that were previously considered the prerogative of people.
The same game was included in the rating of the most complex games for “understanding” computer games. Experts believed that in the near future AI (its weak form) will not be able to defeat not that of world champions, but also ordinary players of high level. However, what happened happened, Alpha Go beat Lee Sedol in four games out of five. And now it’s time to fight with another champion, Ke Jie, a player in go 9 is given. Jie takes the first line of the rating of China on the points scored.
Another professional player approved the contribution of artificial intelligence to go, saying that Alpha Go gives a lot to the development of strategy and tactics. “The Alpha Go game allowed us all to feel free, to understand that any move is possible. Now everyone is trying to play in a new way, in a style that has not been tested before, “said Ruiyan, player 9 is given in go
The creators of the software platform have long declared their desire to reduce AlphGo and Ke Jie, now These plans are destined to come true. There will be two champions, a car and a man on the holiday of go who will be held for five days from 23 to 27 May inclusive in the homeland of this game, in China, in the city of Vuzhen, Zhejiang province.
27 May. The event will be held jointly by China Go Association and the Government of China. The championship, which will be held during the festival, will be visited by famous players in go, as well as experts in artificial intelligence.
As for AI, AlphaGo will not only play against one person. The organizers proposed several stages of the competition, including the following:
- “Two Go” – a game where one of the champions from China will play against the other … Everything is traditional, except that both will have their own partner AlphaGo . It is expected that in this format, a computer and a person will be able to learn from each other;
- “Go Team” is a competition in which five pros will play against AlphaGo. Professionals will try to win from the system, having studied the features of its game;
- “Ke Jie vs. AlphaGo” is perhaps the central element of the whole holiday. There will be two champions – a computer and a man.
By the way, Ke Jie is known for his balanced game. He is not too keen on attacking the enemy, remembering the “rear” and developing a reasonable system of attack and defense. When Lee Sedol played with AlphaGo, Jie explained the advantages of certain shortcomings to the head of Google Sundaru Pichai.
According to TheVerge, Jie lost several unofficial games to Alpha Go. Perhaps in the official game, he will be the champion from China and will show the best he is capable of. In any case, playing with AlphaGo is not a way to show the superiority of AI over a person, but the ability to prove that the machine is no worse than a person can cope with a number of tasks. “Instead of playing” to destroy, “as many fear, AI helps its opponents become stronger by showing original ways of thinking,” says Demis Hassabis, one of the founders of DeepMind.
AlphaGo consists of a combination of the Monte Carlo method and Neural networks policies (policy networks) and values (value networks). To the current level of his professionalism, the system came out a long time after his “birth.” She had to play hundreds of thousands of times, analyzing the games from the server KGS, where masters of different levels play – from the sixth to the ninth dan. Gradually the system self-taught, and the game was conducted not only with people, but with them. Gradually AlphaGo began to beat professionals, and after that she won the European champion in go, and then – at Lee Sedola.
The game itself is considered more difficult for computers because the possible positions of stones on a standard board in googol (10 100 ) are larger than in the same chess. This is more than the number of atoms in the universe. At the very beginning of the game the opponents have stones of different colors on the board and begin to attack each other in the hope of occupying an area larger than the rival with stones of their own color.
By the way, DeepMind does not only work on training its AI for playing board games, but on training a software platform in a virtual environment, in particular, when playing StarCraft. This game DeepMind chose because for 20 years it is the benchmark among cyber sportsmen.
