Computer Vision News - July 2021

Women in Computer Vision 42 Best of CVPR 2021 young. I was also interested in biology, so I split the difference by studying biomedical engineering in undergrad, learning a bit about biological systems but also learning math and learning how to code. That balance was important to me. My undergrad institution had a couple people who were doing modeling of the nervous system. That was just fascinating to me. It wasn’t something that I knew was really an option until my senior year of undergrad, so I applied to a bunch of neuroscience PhD programs, not knowing what neuroscience research was at the time. I was lucky enough that I was accepted into one. Whatwas your biggest “wowmoment”? This specifically, what I’m doing now, it’s hard to say. When I was a kid I was always super into robotics; realizing how hard it was to make robots do things was eye-opening to me. These things that seemed trivial to me, like recognizing an object or walking without falling over, that you don’t even think about, turned out to be really hard to do when you sit down and try to reproduce those behaviors in a program or a machine. That was always fascinating to me, this distinction between what biology can do and what humans can build machines to do. Then a more concrete example is when I was in undergrad. I had a class where we did a problem set on training a perceptron to discriminate pictures What gives them particular instincts? One last question about your work: of all the differences about humans and animals, which one is the most striking? I like to think there are some differences between humans and mice! We have a much more developed cortex. We have more executive control than these animals do. Mice are prey animals, and they are very skittish. It takes a long time to teach a mouse to do something, for certain kinds of tasks. That’s one of the frustrating things about mouse research. Say you want to study how an animal makes a decision, how it integrates sensory evidence and then make some choice about that evidence. It can take days to get a mouse to understand the kind of task you’re trying to get it to do, and you have to just hope that it’s doing the task the way you think it is. I could talk to you for hours about your work, but I also want to ask a bit about you. How did a young woman like yourself end up here? My parents were both computer programmers, so I learned how to code from my mom when I was pretty “What makes animals act in a certain way? What gives them particular instincts?"