The Canadian research institute uses neural networks to develop business solutions and push the boundaries of artificial intelligence.

In their lab at the MaRS Discovery District, Borealis AI is using the way the human brain works to teach computers how to learn like one. "All of the tools we use are inspired by the human brain and how humans interact with the world," explains co-founder and Borealis AI head Dr. Foteini Agrafioti.

Borealis AI is a research institute spearheaded by RBC to pursue the latest developments in machine learning and artificial intelligence. Agrafioti's team works on both fundamental and applied research, producing published academic papers, presenting at top-tier machine learning conferences, and finding applications that can help solve business problems. "AI is a strategic opportunity for Canada," Agrafioti says, "As a bank we do well when the country does well, so we support more research and more entrepreneurship but also leverage the power of this technology for our business needs. That's how Borealis AI came to be."

Life at MaRS

At MaRS, they have found a perfect home for their Toronto mission. "MaRS has made a really big push into bringing AI tenants. Everywhere you look there are like-minded people having an exchange of ideas," says Agrafioti. "The Vector Institute is located here too. Actually, last night I had a group of PhDs from Vector come over for dinner to talk about our research."

Influenced by Geoffrey Hinton

Toronto's Discovery District is dense with AI talent. "Next door there is Geoff Hinton, the Godfather of Deep Learning," Agrafioti explains. Hinton, a University of Toronto professor, pioneered the process of teaching computers to learn from data via neural networks, much the same way the human brain does.

"We deliberately chose to be close to the university. It has to do with the concentration of talent," says Agrafioti, "U of T has done a very good job of training people. You start to see entire industries built around these academic concentrations."

Agrafioti studied at the University of Toronto herself. While she was completing her PhD she won the University of Toronto's Inventor of the Year award, which recognizes commercialization of technology that stands to effect real social change. Her graduate project, Nymi, uses electrocardiogram (ECG) to authenticate user identity with the accuracy of a fingerprint.

Teaching machines to think

Today Agrafioti and the Borealis AI team work to invent new applications for machine learning and push state-of-the-art in machine learning theory. "Machine learning powers most of the pattern recognition technologies out there, whether you're browsing Netflix or shopping on Amazon, it's there," she explains. While accurate movie recommendations are appreciated, these algorithms are different than the work at Borealis. "Our business relies on data to make decisions every day. What happens around the world impacts how we respond, so we're a very news-driven business too."

One of the latest projects to emerge from the lab involves analyzing the world's news to help inform financial analysts of incoming trends. "Heavy rainfall in China can affect the North American electronics market, the world is very connected. The problem is that there is so much information, where do you start?" she asks. "Humans can make these connections very easily but are limited by capacity. We're automating the process to make these connections and understand what's going on around the world, to show you what's relevant even though it might appear unconnected."

Her team has also developed AI applications in cybersecurity, from analyzing traffic from malware to detecting credit card fraud. "It has learned very well, reaching a human rate of accuracy, but much faster than a human."

Despite the rise of AI, Agrafioti does not think we are about to be replaced by machines. "I've never seen a case where we've built an AI that can completely replace a human. I've seen many cases where we have built AIs that augment and allow humans to focus on what they do well."

Northern lights

Borealis AI is expanding rapidly across Canada. In addition to their labs in Toronto, Edmonton and Montreal they are opening locations in Vancouver and Waterloo later this year. Agrafioti thinks Canada is a fertile environment for AI start-ups. "Canada has done well in funding for research; it was the government stepping in that helped build the Vector Institute." She also says flexible immigration policy is critical. "Talent in this field is scarce and spread out across the globe, so being able to bring them quickly to Canada is important." She herself came to Canada from Greece to pursue her graduate studies.

Of course, the research must be brought to market for it to have an economic impact, and institutes like Borealis AI are stepping in to help commercialize the cutting-edge ideas being explored in Canada's universities. "After research, entrepreneurship needs support to be able to scale with the help of VCs or accelerators. With MaRS and our institute we are doing that to build continuing innovation. I think it's a self-fulfilling ecosystem."

Sector/subsector: 

November 14, 2018

Share on

Feedback: Was this page useful?

Talk to a business consultant


Request a conversation

Subscribe to our newsletter


Sign up now

Newsletter Sign Up

Form is for business purposes.

Back to top