The toolset, labelled SC2LE, includes a machine-learning API from Blizzard; an open-source iteration of DeepMind's PySC2 toolset; a dataset of 65,000 anonymised game replays, which will be expanded to more than 500,000 in the next few weeks and will aid imitation learning for sequence prediction and long-term memory; a set of mini games to test AI performance on specific StarCraft II tasks, such as collecting minerals, compiling gas, and selecting units; and a joint paper outlining the environment and initial baseline results on AI performance. But so far, space is proving a hard frontier for the company's algorithms.
Blizzard Entertainment's classic strategy game StarCraft 2 is now being used as a basis for improved AI learning. In addition to a machine-learning API to allow both researchers and developers access to the game, Blizzard is also providing a dataset of anonymised game replays - now at 65,000, this will increase to "more than half a million" in the next few weeks.
Whilst it seems odd to spend so much money, time and effort to teach a computer to play a ideal game of StarCraft II, it's obvious that, on a larger scale, the idea of computer A.I having a self-taught understanding of predictability, adaptation, counter-intelligence and decision-making is absolutely fascinating.More news: Gene editing spurs hope for transplanting pig organs into humans
StarCraft and StarCraft II are among the biggest and most successful games of all time, with players competing in tournaments for more than 20 years. The hope is that compartmentalizing these areas of play will allow testing and comparison of techniques from different researchers on each, along with refinement, before their eventual combination in complex agents that attempt to master the whole game.
Because of these factors, StarCraft II comes much closer to approximating many real-world situations than games such as chess, Go or even Poker.
DeepMind said that SC2LE's mini-games, which are created to test specific skills, have shown promising results in training and evaluating AI, but the company said that its strongest AI agents could not win a full Starcraft II match against even the easiest opponents.More news: Climate report: Earth hotter, seas higher
Video games have become a popular testing ground for AI because they allow developers to control a number of variables while still giving the AI the freedom to interact with a semi-realistic environment.
The company's decision to make its StarCraft II toolsets available to researchers for free differs from the more proprietary approach the company took when it was first working on algorithms that could master Atari games and Go. While AI agents usually perform tasks well in isolated mini-games, they struggle to perform tasks when exposed to the full game.More news: Dallas Stars oppose 'bathroom bill'
- Canada province to challenge Kinder Morgan pipeline
- CNN Bounces Polarizing Contributor For 'Sieg Heil' Tweet
- NY high school football player dies in log drill mishap
- North Korea missiles can reach Guam in 14 minutes, local official says
- Tranquil weather on tap is the calm before some weekend storms
- Isaac Makwala runs by himself, qualifies for 200m semis
- Chris Cornell's Family Commissions Memorial Statue For Seattle
- Jaganmohan Reddy should introspect before criticising CM: Nara Lokesh
- Deathmatch in Two Flavors Coming Soon in Arcade Mode
- Jeffress: God sanctions 'fire and fury'