Computer Vision News - April 2022

23 Claire Vernade work. We try, as much as possible, to have a lot of visibility of other teams. When they come, they ask us questions. They have challenges in their work, and we can help. On our side, we feel no pressure to have strict deadlines to deliver products. This is working super well here.  You live in London working for DeepMind. Can you tell us where you come from? I am French. I grew up in France, I went through the classic French path at the Grandes Écoles and I studied engineering at Telecom ParisTech. I did everything in France, from the beginning until the end of my PhD, except for one short internship at Adobe in California with Branislav Kveton, which was a really great time. I learned a lot about the interactions between industry and research. I had a roughly one-year, part- time Postdoc at Amazon in Berlin. After this math problem that I can solve? ” Ha! I’d say yes, it happens to me, especially during Covid. You had all these vaccine slot booking services where you had to go on the platform and book your slot. Then it will tell you where to go, but you realize that this is not the place you wanted to go. Then you go back and the appointment that you wanted is no longer available.  It’s because you are competingwith other internet users who are trying to book the same thing at the same time. If you hesitate, the slot is no longer available!  It’s a typical computer science problem to have this queuing system and memory allocation trying to optimize peoples’ path for a several-stage system. That’s typical computer science. There are lots of results in scheduling tasks. It sounds like common sense, but it's just that formulating and teaching this theory to people allow us to have more efficient systems since the first run. The problem with Covid was that people didn’t have time! I guess, in the long run, we would have the knowledge to improve the systems, but the challenge is to have the system work as quickly as possible. These are super interesting problems. There are so many! How does your work answer your drive to solve problems? DeepMind has a very long-term strategy. My team looks at very basic theoretical problems. I mean basic in a good way; fundamental. Our team is called Foundations, so it’s really foundational, fundamental problems. They really give us the time we need, and they value our I focus on the situation where the environment is not always giving us the same reward