DeepMind, the Alphabet Inc-owned artificial intelligence (AI) company best known for creating software capable of beating the world’s best go players, has targeted StarCraft II as its next big research milestone, but so far, space is proving a difficult frontier for the company’s algorithms.
DeepMind’s algorithms, including those that performed with superhuman skill across a host of classic Atari titles, “cannot win a single game against the easiest built-in AI” in StarCraft II, let alone challenge skilled humans, the company said in a blog post on Wednesday.
The built-in agents, which are created by StarCraft publisher Activision Blizzard Inc, use hard-coded rules to determine their gameplay rather than the kinds of advanced machine learning techniques in which DeepMind specializes.
Photo: EPA
The company said new breakthroughs in machine learning would be required for its software agents to master the game.
The post did not reveal how close DeepMind might be to such breakthroughs.
The algorithm that mastered the Atari games was unveiled in June last year. Since then, DeepMind has published a number of research papers that hint it might be closing in on creating software capable of many of the tasks — such as prioritizing goals, long-term planning and memory — that any system would need to play StarCraft II successfully.
The company said that its algorithms performed well at learning some basic steps, such as moving around the game environment and selecting units, that would be critical to mastering the game.
StarCraft is considered an important target for machine-learning researchers because, unlike go, in which both players can see the entire board and take turns moving pieces, players in StarCraft cannot see what is happening in the entire game environment at one time and both players move their units simultaneously.
The game also requires players to carry out sub-tasks, such as building structures and mining resources, while also conducting reconnaissance, mounting attacks and defending territory.
To succeed, a player needs to have a good memory, prioritize among tasks and plan under conditions of uncertainty.
Because of these factors, StarCraft II comes much closer to approximating many real-world situations than games such as chess, go or even poker.
StarCraft II is also used in e-sports competitions, so there are highly skilled human opponents with which an AI can match wits.
To help computer scientists use StarCraft II as a testbed for AI, DeepMind has partnered with Blizzard to create an interface that allows outside software to access and play the game.
The two companies this week unveiled the interface, along with a set of tools to help other computer scientists train AI agents to play the game, at a machine learning conference in Australia.
Among the tools Blizzard is making public are a dataset of anonymized game replays — recordings of humans playing the game — that computer scientist would be able to use to help train their systems.
“One technique that we know allows our agents to learn stronger policies is imitation learning,” DeepMind said in its blog post. “This kind of training will soon be far easier thanks to Blizzard, which has committed to ongoing releases of hundreds of thousands of anonymized replays.”
DeepMind also said it was releasing a series of “mini-game” environments that would help researchers train their AI agents on basic components of the game.
It follows efforts by other companies, such as Facebook Inc and Microsoft Corp, to open game environments, including the original StarCraft and Minecraft, to the entire AI research community.
BUSINESS UPDATE: The iPhone assembler said operations outlook is expected to show quarter-on-quarter and year-on-year growth for the second quarter Hon Hai Precision Industry Co (鴻海精密) yesterday reported strong growth in sales last month, potentially raising expectations for iPhone sales while artificial intelligence (AI)-related business booms. The company, which assembles the majority of Apple Inc’s smartphones, reported a 19.03 percent rise in monthly sales to NT$510.9 billion (US$15.78 billion), from NT$429.22 billion in the same period last year. On a monthly basis, sales rose 14.16 percent, it said. The company in a statement said that last month’s revenue was a record-breaking April performance. Hon Hai, known also as Foxconn Technology Group (富士康科技集團), assembles most iPhones, but the company is diversifying its business to
Apple Inc has been developing a homegrown chip to run artificial intelligence (AI) tools in data centers, although it is unclear if the semiconductor would ever be deployed, the Wall Street Journal reported on Monday. The effort would build on Apple’s previous efforts to make in-house chips, which run in its iPhones, Macs and other devices, according to the Journal, which cited unidentified people familiar with the matter. The server project is code-named ACDC (Apple Chips in Data Center) within the company, aiming to utilize Apple’s expertise in chip design for the company’s server infrastructure, the newspaper said. While this initiative has been
GlobalWafers Co (環球晶圓), the world’s No. 3 silicon wafer supplier, yesterday said that revenue would rise moderately in the second half of this year, driven primarily by robust demand for advanced wafers used in high-bandwidth memory (HBM) chips, a key component of artificial intelligence (AI) technology. “The first quarter is the lowest point of this cycle. The second half will be better than the first for the whole semiconductor industry and for GlobalWafers,” chairwoman Doris Hsu (徐秀蘭) said during an online investors’ conference. “HBM would definitely be the key growth driver in the second half,” Hsu said. “That is our big hope
The consumer price index (CPI) last month eased to 1.95 percent, below the central bank’s 2 percent target, as food and entertainment cost increases decelerated, helped by stable egg prices, the Directorate-General of Budget, Accounting and Statistics (DGBAS) said yesterday. The slowdown bucked predictions by policymakers and academics that inflationary pressures would build up following double-digit electricity rate hikes on April 1. “The latest CPI data came after the cost of eating out and rent grew moderately amid mixed international raw material prices,” DGBAS official Tsao Chih-hung (曹志弘) told a news conference in Taipei. The central bank in March raised interest rates by