Computer Vision News - July 2021

Women in Computer Vision 40 Best of CVPR 2021 It doesn’t seem like themost attractive part of the job anyway. Do we really need to work with animals in order to understand ourselves better? That’s definitely a thought that has arisen in theoretical neuroscience since the emergence of deep learning - there are people who are studying artificial neural networks as opposed to working with animal data. I think it really depends on the question that you’re trying to ask. If you care about ways that any neural system could learn, you could do that in an artificial neural network. If you’re looking at how things are specifically happening in the human brain and animal brains, and how we treat diseases and disorders, then we still need some access to the biology. We need to see the real side of things. It really just depends on your priorities. I guess the priority is studying mammals. Is that right? There’s a vibrant invertebrate neuroscience community as well. I collaborate with a postdoc who studies the nervous system of jellyfish. My postdoctoral lab has a cohort of people who work on the fly nervous system. I think there’s a lot you can learn from that too just because insect nervous systems are so different from mammalian nervous systems. In mammals, a given task will be associated with a pool of thousands or tens of thousands of neurons your brain or a new neural network to recognize a fire extinguisher with a very few number of examples. Do you ever feel uncomfortable working with mice? I don’t personally work with mice - mostly I collaborate with experimental labs that collect this data. Although I have a small component of my lab that is not up and running yet but will be collecting behavioral datasets in mice. So somebody else is doing the dirty work? [ laughs ] I guess so, yeah. It’s a part of the job!