Google alpha go really going to human Star 2 was singled out-winavi

Google Alpha Go to early with humans singled out the "Star 2", Google Alfa dog (AlphaGo) shocked the world, the discussion about artificial intelligence also came to the peak, "beat the best human players, this is full of legendary events in fact, so many people overestimate the artificial intelligence power. After that, there are a lot of news, Alfa dog next to challenge the "StarCraft 2", artificial intelligence over the occupation player point the day and await for it. In March this year, a lot of Chinese media is reported in the WCS China Station finals, Blizzard Entertainment production director Tim Morten personally confirmed that AlphaGo will challenge the StarCraft 2. This news has two points: first, almost no foreign media to "AlphaGo challenge" StarCraft 2 "for the title were reported; second, the name of the AlphaGo also shows, this is only DeepMind for the chess project (English: Go) artificial intelligence project specialized training, if the challenge of StarCraft, should not be the name. More precisely, the "wealth" report: DeepMind’s CEO, Demis Hassabis, said that "StarCraft" is an interesting challenge to consider. Now, DeepMind and Starcraft’s "scandal" can finally be implemented. DeepMind Engineer Oriol Vinyals announced on the company blog, Blizzard will cooperate with DeepMind, so that the "StarCraft 2" has become a research environment of artificial intelligence and machine learning, and in the first quarter of next year will be open to the environment are all artificial intelligence research. In other words, even DeepMind’s artificial intelligence, has just begun to learn "StarCraft 2". Oriol Vinyals was a top junior StarCraft player in Spain, and he said, "we have a long way to go before we can defeat a professional player." Unlike the turn of the game, StarCraft is an instant strategy game in which players need to choose one of the three races, each with its own advantages and disadvantages. According to the rapid response of the different situations, and long-term planning of the war, this is a very difficult challenge for artificial intelligence program. And in Li Shishi when playing the game, AlphaGo from the Monte Carlo tree search algorithm, the optimal evaluation may be the next solution (can eliminate some of their less favorable, may) to reduce the search depth. Even so, has a powerful computing capability of AlphaGo up before long, still need time to think. In addition, unlike the game, in the "StarCraft", artificial intelligence can get the opponent’s information is limited (such as the game is not explored in the region of the fog of war, where the opponent can not see). You may think of the "StarCraft" comes with the AI, that is, computer players. Its working mechanism is quite different from the current artificial intelligence, because the AI in the game is to get information directly from the game code,.相关的主题文章:

Comments are closed.