Google is paying attention, so is Yahoo.
But the University of Alberta-based Alberta Ingenuity Centre for Machine Learning (AICML) hasn't fully played its hand yet.
Established as a centre for pure and applied machine- learning research with financial support in 2006 of $2.3 million from the province's Alberta Ingenuity Fund, AICML's machine-learning operation brings together experts with backgrounds in artificial intelligence, computer science, statistics and mathematics.
Combining this knowledge, it finds useful data patterns in databases - but at speeds with which humans would be hard-pressed to compete.
It has enjoyed early success with work in the gaming sector - its Poker Academy software, described as a full-feature poker simulator, has generated revenue for Edmonton spinoff firm BioTools - and software that helps with the identification of brain tumours.
And while it seems at first glance that gaming and cancer research are completely unrelated, they're not - it's the machine-learning component that is the key.
"We've come up with a learning framework that we can plug into existing legacy databases and that system will help the client make informed decisions from their data," says AICML centre manager Brant Coghlan.
AICML scientific director Russ Greiner, a U of A professor of computing science, says a machine can look at these gigabytes of data and find patterns.
"Data mining is a technique applied to databases. Machine learning is a step beyond data mining," adds Greiner.
Web search engines such as Google and Yahoo wouldn't exist without machine learning, he notes. "Google and Yahoo have hired our highly qualified personnel recently. We're on their radar screen when they're looking for ideas. They know this is a good hunting ground."
But machine learning also addresses the challenge of turning data into useful information. "Our centre has become one of the world centres tackling the big problem of machine learning," says Greiner.
"We're also trying to help diversify the Alberta economy, to leverage our strengths in oil and gas, and to build capability in ICT," adds Coghlan, who notes the centre's mandate is to operate provincewide with an outreach component.
"We haven't yet found a business partner, but we're looking at (the issue of) compressor failure in the oilpatch. We believe there's an opportunity to identify and predict when these compressors will need maintenance, or fail, and want to do this before it (becomes) critical," says Coghlan.
As part of its outreach work, the centre is also extending an offer to the province's business sector.
"We'd like them to talk to us about their data challenges - ideally, companies that have large amounts of data and challenging problems," adds Coghlan.
"We can help in two ways: One way is by providing some advice - we're a source of knowledge. The other is through programs we're familiar with, we can assist companies in subsidizing the hiring of our recent graduates as well as interns."
One company already working with AICML is Edmonton-based Xymbiant Systems Inc., a startup firm that is building intelligent conversational systems based on natural language - in this case, everyday English - for e-mail, instant messaging and short-message service for cellphones.
"The problem with language is that it involves a lot of data and you have to get that data from someplace," says company CEO Bruce Matichuk. "It requires machine learning. If you get people to do all the work - actual programmers to program every last detail in - it's going to take forever and it's not going to work that well."
Matichuk says the company is probably about a year away from a large-scale commercial product.
"What we want to do is take the toolkit to do these projects and make it available as a general product for the marketplace. You download and build a bot (a program to do automated tasks) yourself without us having to be involved," says Matichuk.
AICML also has ties with major gaming companies such as Redwood City, Calif.-based Electronic Arts and Edmonton's BioWare.
Electronic entertainment is a multibillion-dollar industry, notes Coghlan. "There's lots of money to be had by doing it right, whether we're building better fantasy games or poker.
"We did some work with (Electronic Arts') FIFA soccer. We've certainly been supplying people to BioWare and BioTools. All three of those game companies have given us access to source code for some of their computer games - it's very rare for a game company to release their source code to anyone and it shows the value that machine learning has," says Coghlan.
"We're trying to make computer games more fun to play, game characters more intelligent. We're also helping the designers identify weaknesses in their games."
Working with the Edmonton-based Cross Cancer Institute, machine learning is also being used in the medical field by identifying and predicting how a brain tumour will grow.
"We have patent protection and we're pursuing licensing of the technology and it's being tested by two companies currently. We hope it will be licensed next year," says Greiner.
(Laura Severs can be reached at laura@businessedge.ca)
