Google has entrusted Hugo La Rochelle, a professor at the University of Sherbrooke, with the task of building a high-quality center for deep learning research in Montreal.
artificial intelligence file Several big players in the Artificial Intelligence (AI) sector have set foot in Montreal over the past few months. Who are they and above all what do they do? To see more clearly, here is an overview of the work of Google, Facebook, Microsoft and IBM in Montreal.
Google: talent first, wherever it is
When, less than a year ago, the American multinational Google announced the opening of a research center in downtown Montreal, All-Montreal Scientific had reason to rejoice.
First, Google could have chosen to establish itself elsewhere. It then reached out to pay $4.5 million to fund the work of eight local AI researchers. Above all, by coming to Google, Google has brought back from the United States one of the brightest minds that Quebec has produced in this field: Hugo Larochelle, a professor at the University of Sherbrooke, whom Twitter managed to lure to its offices in Cambridge, not far from Boston, a few years ago. Years. It is for the former student of Professor Yoshua Bengio, leader of the Montreal Institute for Learning Algorithms (MILA), that Google has entrusted the task of building a high-level research center on deep learning.
The research team at the Artificial Intelligence Laboratory in Montreal is still modest. On the second floor of a building overlooking McGill Avenue College, the new team had barely six people, the equivalent of the Toronto team. This number should grow to about ten people.
Hugo La Rochelle explains that every researcher in Montreal works on different aspects of AI. For example, the latter is interested in visual recognition and computer vision. One of his goals is to create algorithms that allow the computer to learn to recognize objects by drastically reducing the number of images required.
He explains that others are working to improve systems in machine learning or reinforcement learning. Essentially, Google hires talent wherever they are and gives those it recruited the freedom to pursue work in their chosen field, says La Rochelle.
He continues, “We work primarily in basic research. This does not prevent our work from leading to concrete applications, but there is no pressure in this direction. What we are asked above all is to try to influence by giving conferences on our work, through Get involved in our community or by publishing articles in journals that make an impact.
Facebook: machine learning
On September 15, the American giant Facebook announced the establishment of a new research laboratory specializing in artificial intelligence in Montreal. This lab, the first in the Facebook family to see the light of day in Canada, is run by Quebecer Joëlle Pineau, professor for more than a decade in the Department of Computer Science at McGill University and co-director of the Thinking and Learning Lab, the Radiation Research and Learning Laboratory.
With this, Montreal became the fourth city on the planet to join the Facebook Artificial Intelligence Research (FAIR) team, the lead person responsible for artificial intelligence research within the Facebook family. The other three research centers are located in Menlo Park, California, New York, and Paris.
This international network of laboratories brings together just over a hundred researchers, including a dozen in Montreal. According to Ms. Pino, the Montreal team should grow rapidly, reaching 20 to 30 researchers within 18 months. They will initially specialize in machine learning, but it is expected that their field of research will rapidly expand to also include reinforcement learning and dialogue systems.
Facebook researchers will enjoy great freedom, FAIR focuses on basic research, explains the new director of the Facebook laboratory in Montreal. Applied research is outsourced to other Facebook entities, outside of Montreal.
Ms Benio explains that she was particularly tempted by Facebook when she learned that FAIR practices so-called open research, that is, it encourages the sharing of its research findings outside of Facebook’s walls, throughout the scientific community.
Facebook has announced $7 million grants to some of Montreal’s leading artificial intelligence organizations, including the Canadian Institute for Advanced Research, the Montreal Institute for Learning Algorithms (MILA), McGill University, and the University of Montreal.
Microsoft: Artificial Intelligence Corridor
By acquiring Ontario Maluuba in January, US Microsoft had plans to make Montreal its main laboratory for artificial intelligence research in Canada. Nine months later, the forty or so Maluuba employees who worked at Waterloo had moved to Montreal. Many of them have already started learning French, confirms Jennifer Chase, director general of Microsoft Research in Cambridge, who also became responsible for Microsoft’s activities in Montreal in the field of artificial intelligence.
“I am completely overwhelmed by the extraordinary potential of the Maluuba-Microsoft team in Montreal and by the implementation of government measures to protect this gem that you have built here,” Ms Chase told Les Affaires between two scientific conferences on the topic in Montreal. University of Montreal.
The gem that the Microsoft representative is talking about is the ecosystem that the Montreal science community has put in place to make the city essential in the highly competitive world of artificial intelligence.
Jennifer Chase has known something about this since she co-founded Microsoft Research New England in Massachusetts in 2008, and co-founded the Microsoft Research Center in New York in 2012. “Without anything, with our lab in Montreal, Microsoft now has one of the most Enviable knowledge in the field of artificial intelligence.” Today, with an estimated forty researchers, the team at the Microsoft Research Center in Montreal should double in size to around 80 people by the end of 2018. In addition to collaborating with teams in New York and Cambridge, the Montreal Center will also be invited to work with those in Vancouver and Seattle.
The Montreal Lab intends to specialize in deep learning, reinforcement learning, and dialogue systems. Founded in 2011, Maluuba has distinguished itself in its short existence with the development of technology that allows computers to communicate with each other.
Jennifer Chayes sums up: “Not only can we now make machines answer questions, but we can also make these computers, by interacting with each other, managing problems or asking questions we could not have otherwise imagined.”
Finally, Microsoft announced that it wants to pay $7 million over five years for AI research at the University of Montreal ($6 million) and McGill University ($1 million).
IBM: Elementary, my dear Watson
Last April, the American company IBM announced the opening of a laboratory in Montreal to strengthen its relationship with MILA, led by Professor Joshua Bengio.
Recruitment has been launched. The team will conduct basic scientific research on artificial intelligence, machine learning, and deep learning, as well as explore areas of research such as unsupervised learning.
Professor Bengio’s team has already been working with IBM for more than a year on deep learning algorithms to help computers improve their interpretation of language, speech and vision. This work is believed to have the potential to contribute rapidly to the improvement or birth of new scientific or commercial applications.
The best example of an application is Watson, a cognitive computing system developed by IBM a few years ago. By analyzing millions of diverse data, Watson was able to guide professionals in complex decisions. Moreover, it has already been widely talked about due to its potential contribution to the field of medicine. By traversing millions of data that a clinician or even the entire team couldn’t interpret, Watson was able to deliver life-changing treatments for cancer patients and Parkinson’s patients, says Natalie Le Prohon, vice president of health industry at IBM Canada proudly.
And that’s only the beginning because the areas of Watson’s application seem limitless. At the latest eCOM MTL conference, IBM partner Imad Osmani gave an example of his potential contribution to the creative industry. For example, once trained, Watson is able to create a movie trailer in one day, while it usually takes 20.
IBM says it conducts intelligence research in dozens of countries, including Canada (mainly Toronto and Montreal). Arguing for reasons of competitiveness, IBM on the other hand refuses to share the address of its lab, as well as the number of AI researchers it currently gathers in Montreal.
Click here for a dose AI consultation: How Quebec is taking part in the revolution