Computer Vision News - March 2022

11 Objects, Structure, and Causality (OSC) and causal researchers, the workshop will play host to speakers from a range of fields and perspectives, including core causality people like Bernhard Schölkopf , Director of the Max Planck Institute for Intelligent Systems and professor at ETH Zürich, and big object people like Nikolaus Kriegeskorte , a professor at Columbia University. There are also speakers fromthe worlds of machine learning, neuroscience, and psychology, including Alison Gopnik , a professor of psychology and affiliate professor of philosophy at the University of California at Berkeley. “ We’re very excited by all the different perspectives, ” Wilka adds. “ Integrating different expertise on the same problem is such an important part of making progress. With our collection of speakers and participants, I do not doubt that we are poised to do that! ” “ I am so fascinated by the human ability to discover the pieces of the world and to leverage that to adapt our behavior, ” he tells us. “ For example, I’m in Boston now to give a talk at Harvard, and even though the kitchen here in this apartment is different, I have no problem generalizing my behavior to this setting. I know the pieces, and I know roughly how they relate, so I can recognize them in new settings. Last night, I needed to find a fork, so I looked around this completely new kitchen I’d never been in before and was able to find one. AI is not quite there yet. ” Reinforcement learning defines a framework for learning behavior that maximizes reward. Using the fork example, Wilka’s reward for finding the fork was that he got to eat. That reward guided his behavior. The next step is training agents to learn behavior guided by reward . As well as bringing together object-centric