After several years in the creative wilderness, the 35-year-old whiz kid returns to Quebec, where his AI career began.
Don’t be fooled by Hugo Larochelle’s youthful looks. He’s one of the world’s brightest stars in artificial-intelligence research. Since 2012, he has been cited 7,686 times in the Google Scholar index. “He was involved in the very first article on deep learning that we wrote in 2006, which sparked interest in this growing field,” recalled professor Yoshua Bengio, a leader in the field and Larochelle’s thesis advisor at Université de Montréal’s Department of Computer Science and Operations Research.
What exactly is deep learning? Simply put, it’s how computers simulate the type of calculations that human brains perform. The information we process travels through various areas of our brain, each with its own function and specialization. These areas are can be thought of as levels (hence the “deep” part of learning) that need to be tapped to analyze information. Deep learning is so innovative it's often been described as a technological revolution.
Larochelle, 35, recently bought a house in Mont-Saint-Hilaire, an hour's drive east of Montreal, finally settling down after many years wandering the world of technology. “I knew I wanted to raise my family and develop research projects here in Quebec, so I’m thrilled with Google’s proposal,” he said, referring to his hiring as the California tech giant's man in Montreal for its innovative Google Brain project. "The Google opportunity also appealed to his paternal instinct. “My wife and I had been living all over the place for quite a few years. We dreamed of providing our four daughters a stable home environment where they could hear French being spoken.”
Twitter bought his start-up
A foodie and video game fan, Larochelle was already a star in UdeM's computer science department even before he started doing research. He aced all his undergraduate courses (with a 4.3 GPA) and showed exceptional maturity. “He was the best, but he never let it go to his head," Bengio recalled. "He became a hotshot in deep-learning research, which is the spearhead of today’s tangible progress in artificial intelligence.” He’s also a team player who loves to share knowledge, Bengio added.
Larochelle earned three degrees at UdeM (a Bachelor of Mathematics, a Master’s and a Doctorate in Computer Science) before moving on to do postdoc work at the University of Toronto. In 2011, upon completion of his postdoctoral studies, Larochelle accepted an assistant professor position at Université de Sherbrooke.
In addition to his academic pursuits, at the start of his career Larochelle joined forces with a colleague to create Whetlab, a start-up that developed software to accelerate data-processing operations. Simply put, they came up with a tool that provided access to advanced methods of artificial intelligence. Whetlab quickly became a serious player in emerging technology, attracting the attention of industry giants like Twitter, which bought the company in 2015. After selling his share, Hugo Larochelle joined the Twitter team and worked at their headquarters for two years. Then Google asked to pick his brain.
“Google already had a development group in Montreal for the Chrome browser,” Bengio said, “but Hugo’s mission is to add a world-class deep-learning research team, and I have no doubt that he will be successful."
Thanks to his many publications, Larochelle had no difficulty landing a promotion to associate professor at Sherbrooke in 2016. He continues to work for the university and still supervises some graduate students, but now devotes part of his energy to his new challenge, heading up the Google Brain group in Montreal.
Google, which began in 1988 as a concept hatched in a California garage and is a top global brand, is much more than a search engine. The multinational has launched a number of projects to diversify its operations, moving into artificial neural networks in 2012. The Google Brain division, whose goal is to advance understanding of this cutting-edge sector and promote technology transfer, needed a leader in Montreal, a city known as the 'Silicon Valley of the North' in deep learning.
And Larochelle – a mathematics whiz kid from Asbestos, Quebec, who had managed to become part of the global elite in his field in less than a decade – was the ideal candidate. “Google wants to partner with state-of-the-art research centres," he said. "I see myself as a facilitator in meeting that objective."
Larochelle had just returned to Montreal when Google announced a $4.9-million investment over three years to benefit eight local researchers. Having been the recipient of many scholarships and research grants, he now finds himself on the side of the investors. He said he was “surprised by his own life” and did not expect to be so successful when he started his Ph.D in Artificial Neural Networks in the fall of 2004. “It’s a bit surreal,” he admitted with a grin.
So, what does the future hold for him now? Something positive, he replied: the freedom to concentrate on the fertile ground of fundamental research and commercial applications. “Google represents a new industrial culture. The artificial intelligence-research division is a good example of that.”