Go – a board game originated in ancient China, about 3 thousand. years ago. Today, it is popular not only in East Asia but also in Western countries. The goal – to capture as a greater area of the playing field (which is lined with vertical and horizontal lines of the board) with special chips.
Despite the fact that the rules of fairly simple, the number of possible positions of the parties and the game of chess is superior and, therefore, until recently, remained first real challenge for artificial intelligence.
In 2014, Google acquired the company Deep Mind, which is engaged in the development of artificial intelligence. The acquisition will enable the corporation to move forward in the creation of a virtual mind and has borne fruit. Computer software Alpha Go, improved Google, failed to win a professional go-player, three-time European champion in a match of the five parties Fan Hui. The article, which the authors talk about the program, published in the magazine Nature. The program is able to estimate the location of the stones on the board and select the most appropriate in this situation moves.
Neural networks Alpha Go excel program play of, for example, Pachi and Fuego, due to the fact that she is learning to play, based on the information parties, played out by people, and analyze information about their own game.
Alpha Go outright won 99.8% of the tournament against programs-analogues, which are based on the Monte Carlo method and three-time European champion Fan Hui defeated in a match of the five parties with a score of 5: 0.
Chess supercomputer Deep Blue , created by IBM and in 1996 beat world chess champion Garry Kasparov, gives Alpha Go, because it is based on an algorithm. One of the researchers of the computer, John Diamond, said:
«Until now, many programs can not beat amateur players. I assumed that in the next five – ten years, the computer will not be able to win the professional players, but now the victory of artificial intelligence seems inevitable ».
Though the car and was able to beat a man to go, it does not mean that will be the end of the game between people. Winning computer chess tournament in 1996 has served as an incentive for players to learn from computer programs. Perhaps these prospects and expect the first computer. Nevertheless, this breakthrough, along with Monte Carlo and tree search, making a significant contribution in the field of non strategic games.
In the future, pass a test program Alpha Go Lee Sedol offered – it is considered the world’s best player of the last decade.
The tournament will take place in March. The player acknowledges that the Alpha Go makes a revolution in the history of the game, but the Secretary of Korea Baduk Association (the local name for the game of) suggests: “The computer can win one or two games, but not the entire tournament».
Artificial Intelligence able to compete with a man not only in the game – the computer learned to recognize printed and handwritten text regardless of the language, as well as to distinguish human emotions. Scientists have achieved some success in machine learning – a computer model that simulates the brain,
was able to learn English on their own and talk to him.
The artificial neural network called ANNABEL helped researchers to understand the mechanism of language acquisition without any algorithms. Another sensational program that simulates a person could go through Turing Test and convince the jury that they talking 13-year-old Jack from Odessa.
No comments:
Post a Comment