With the opening of its newest AI research lab in Montreal, Facebook is tapping into Montreal’s thriving AI ecosystem. The fourth Facebook AI Research (FAIR) lab will be led by Dr. Joelle Pineau, an expert in the field of reinforcement learning and co-director of the Reasoning and Learning Lab at McGill University’s School of Computer Science.
Q. Tell us about your area of AI research?
A. I am working on the question of how to enable machines to make good decisions (or a sequence of decisions) in complex real-world situations, even when there is incomplete or incorrect information. Reinforcement learning enables machines to learn new tasks through experimentation, feedback and rewards.
For example, if you’re playing chess and thinking about the next move, it’s very predictable. It’s one decision and we have a clear model of the effects of that decision. In the real world, there is a lot of uncertainty about the effects of your actions. If you make several decisions, what is relationship between all of them? How do you plan out a course of action? How do you enable a machine to learn good strategies to optimize its behavior?
Reinforcement learning can be applied to tackle this problem. There are many use cases including robotics, transportation, healthcare, and dialogue systems.
Q. What areas of AI research will the Montreal lab be working on?
A. The Montreal lab will be working on all aspects of AI research. One main focus will be on reinforcement learning and the framework for sequential decision-making. Conversational agents will be a key part of our research. Some of team will also work on deep learning, optimization, computer vision and video understanding.
Q. What are some ways people might experience your lab’s research in Facebook products?
A. I’m still exploring the Facebook organization. There are a lot of projects focused on the interaction between humans and bots where AI-driven bots can carry out transactions. They can do short interactions and you either get the answer or not. But if you want to have a longer conversation, to say something and refer to something else, you have to consider the persistence of information in multiple forms. Reinforcement learning and dialogue systems will play an important role in accelerating the development of more sophisticated conversation agents.
Q. Why did you decide to join Facebook?
A. There are many factors that contributed to my decision. I’ve been reading about the work of the people in FAIR for years. It’s exciting to have the opportunity to work more closely with these fantastic AI researchers in the New York, Paris and Menlo Park labs.
The culture of FAIR is unique. I like the philosophy of open science, to be able to publish our papers and code, and talk about our research openly. Some companies share a little, some don’t at all. FAIR is an extreme model and I appreciate that open approach as it makes it feasible for me to still spend half of my time in academia.
Also, I knew I didn’t want to leave Montreal. When Facebook was interested in opening a lab in Montreal, the discussions became much more real.
Q. What do you think about the transition from academia to Facebook?
A. I’m lucky that I get to continue to participate in academia. There are opportunities on both sides. In industry, there are more resources and better access to computational resources. On the academic side we have the students—and a constant flow of new ideas and talents pouring into this research. It you get together with very senior researchers, you have lots of good ideas, but not a lot gets done. With a room full of students, things get done. They are so driven and imaginative. I have the best of both worlds.
Q. Why is Montreal such a hotbed of AI research?
A. It’s a fantastic city and many people who live here don’t want to leave.
Montreal has had expertise in AI for many years. Our public universities and government have invested in it. We collaborate together very well. There is a shared institute, an incubator, and a whole ecosystem that has accelerated our research.
The focus of the machine learning specialization at McGill University is reinforcement learning. At University of Montreal, the focus is deep learning. The decision for FAIR to come to Montreal is to be close to this talent as those students graduate.
Q. How will AI change how we live over the next 3-5 years?
A. The future is hard to predict. There will be a lot of changes in our urban environments with autonomous driving. We’re seeing changes in how we access information. AI already mediates a lot of decisions in our digital lives. Right now we’re interacting through websites, chat bots will mediate those interactions more in the future. I’m excited about all of it. I’m a curiosity-driven researcher. I can’t resist a new challenge.