How Waterloo’s artificial brain is powering the next tech revolution
Chris Eliasmith
Professor, Faculty of Arts and Faculty of Engineering
> Co-Founder, Applied Brain Research inc.
> Director, Centre for Theoretical Neuroscience
The notion that robots could possess something akin to a human brain seems like science fiction.
But not for Chris Eliasmith who has dreamed of building an artificial brain for many years. In fact, he wrote a book in 2013 with the ambitious title How to Build a Brain. Now, at the Centre for Theoretical Neuroscience, he is taking the first steps toward doing that.
Eliasmith is a professor in philosophy, systems design engineering and computer science at the University of Waterloo. He leads a team that built Spaun, the world’s biggest functional brain model. It is a kind of a “brain in a box,” with the equations in powerful algorithms mapping out what neurons actually do in our own brains as we perform mental and physical tasks.
The brain model can simulate approximately 6.6 million neurons, which is relatively small compared to the 86 billion neurons in the average human brain. Nevertheless, Spaun can do such things as recognize lists of numbers, solve some reasoning problems, do simple arithmetic and even write out the answers using an arm.
“It was one of the first models to be able to switch between a number of different tasks, which is something the human brain does quite well,” Eliasmith says.
Peter Suma
Alumnus, Faculty of Engineering
> Co-Founder, Applied Brain Research inc.
Peter Sumas, a venture capitalist, was on the lookout for new technology investments and the next big thing in machine intelligence. He discovered that Eliasmith was building something like a brain — or at least a simulation of part of a brain.
Suma was so impressed that he went back to school to study under Eliasmith.
Although Spaun is a lab research tool that is used to better understand the human brain and explore possibilities in artificial intelligence, Suma and Eliasmith realized that some of the technologies used to build it could be commercialized.
The startup, Applied Brain Research (ABR) Inc., was launched in 2014 by Eliasmith, Suma, and co-founders Terry Stewart, Daniel Rasmussen, Travis DeWolf, Trevor Bekolay and Xuan Choo, who were researchers in the Computational Neuroscience Research Group at Waterloo.
The company has entered into a partnership with Intel to put its software on the new Intel neuromorphic processor called Loihi. Several artificial intelligence applications, including a keyword speech recognition app and a robotic controller, were demonstrated at the Ontario Centres of Excellence Discovery conference last year.
“We hope that every chip that goes out there with our partner Intel will have a little bit of ABR on it,” Suma says.
The startup has been turning heads, winning several awards in international technology competitions and rankings. Last year, ABR was the only technology company in Canada to be named a 2018 Technology Pioneer by the World Economic Forum in Davos, Switzerland.
Corporations like Intel are betting on neuromorphic chips because they are designed more like the human brain, where neurons will only “spike” and send signals to other neurons when the activity reaches a certain threshold. That makes the chips efficient and able to use less power.
Most of today’s artificial intelligence applications, such as the Siri voice interactions that people with Apple smartphones are familiar with, are happening over the internet. But neuromorphic chips make it possible to run artificial intelligence software right on the phones themselves, reducing costs.
“One of the most expensive things to do is move data around,” Eliasmith says.
Self-driving cars, robots and medical devices are just a few examples of areas where the neuromorphic chips and software could be deployed. Transparency Market Research estimates that neuromorphic processors will be a huge market, with $1.8 billion in global annual sales by 2023.
But first ABR needs to prove and demonstrate the technology. It has contracts with several companies, including car and other manufacturers, to explore the applications.
“We are putting algorithms on a chip to demonstrate that the neuromorphic chips can perform the application faster and more efficiently,” Eliasmith says.
Whether researchers can ever build anything that approaches the power, speed and ability of the human brain is an unknown fraught with ethical considerations. But Eliasmith says there is no reason we can’t build “cognitive tools” that work alongside humans.
“We had the industrial revolution where we built physical tools that significantly outperformed what people could physically do and that helped us make things more efficiently,” he says. “I think the same thing could happen on the cognitive side with neural computation using neuromorphic chips.”
Suma says he is more convinced than ever that this is the next technological revolution.
“It seems to me that the next step is moving toward intelligence, where it is not just data and information being processed dumbly, but tools that actually leverage the whole cognitive realm. It’s clear the next opportunity is in that direction,” he says.
