[ad_1]
Developers have stated that games such as StarCraft II serve only as a training ground for artificial intelligence.
AlphaStar gave orders to three men, located in different areas simultaneously / deepmind.com
The artificial intelligence AlphaStar, created by DeepMind (owned by Alphabet Inc.), defeated two professional players The 19659004] Naked Science reports on this site.
Matches of a two-player professional algorithm – Dario "TLO" Wünsch and Grzegorz "MaNa" Komincz) – were held in December, but this week, January 24, the last game was held, before which properties of the program have changed. As a result, the victory first arrived at the man.
Previously, the DeepMind algorithms exceeded the most powerful player of the hole and allowed to better predict the shape of the protein compared to professional biologists. However, in each of these cases, all available information has been discovered for artificial intelligence. With a real-time strategy, such as StarCraft II, it's all different: it's a game with private information. The machine does not know what is happening behind the fog of war and can not predict the actions of the player. It must therefore adapt instantly to its maneuvers.
Read also Google engineers learned artificial intelligence to play Quake III (video)
The developers of the program were studying the actual games that Blizzard provided them . Mastering behavioral patterns of players, she learned to gain the built-in artificial intelligence of the greatest complexity in 95% of cases. Then his agents started to compete in the tournament for the round of 16. As it was in accelerated mode, each of them spent about 200 years.
After choosing the most effective agent, the authors opposed him to TLO and MaNa. First, he won five out of five matches at the TLO, and another version of the neural network managed to defeat MaNa at the same time. In this case, the algorithm was doing fewer actions per minute than its opponents. This is probably due to the small advantage that he still had. Unlike a person, he did not see a part of the known map (the same screen), but the whole of it, so that he did not have to switch between different areas. In addition, viewers noticed that he was giving orders to three troops located in different areas simultaneously, which were absolutely not capable of a person.
Video: DeepMind
In the last game, the developers made concessions. and disabled this ability of artificial intelligence – this time won MaNa. However, the authors noticed that the version of the camera displacement algorithm, studied in the tournament, lasted only seven days, half of the previous training period. MaNa herself hailed the actions of the program: "I play with a real person, he could not manage his stylists also filigree (a unit of races – ed.)."
Developers have said that games such as StarCraft II, serve only as a training ground for artificial intelligence. In the future, he will attack for other purposes. For example, earlier in DeepMind, they said their algorithm would diagnose eye diseases.
Help. StarCraft II is a computer game of the kind strategy in real time. The scenario of the game takes place on several planets, where takes place the conflict between the three races – people, zerges and protoss. Gambling is a discipline in many cyber sports leagues around the world. The game has entered the IGN version of the Top 100 Games of All Time (19659003).
If you find an error, point your mouse over it and press Ctrl + Enter.
[ad_2]
Source link
