Thursday, September 19th
ABSTRACT: I will present the essence of "machine learning" using as little math and computer science as possible, but enough to convey that modern machine learning algorithms depend on impenetrably complicated mathematical computations. (Very, very little math, I promise.) Along the way, I will explain what constitutes "learning" to a machine, and how machines exercise what they've learned. Finally, I will discuss the fact that self-driving cars are essentially answering trivial real-world trolley problems constantly, and absolutely nobody knows what they will do when faced with an unforeseen non-trivial problem. The goal of this nontechnical talk is to demystify machine learning for an audience of philosophers, and, I hope, promote a discussion about how to think about what applications based on machine learning are actually doing and not doing.
The Freedom Center Fall 2019 colloquium series presents Todd Proebsting, Professor and Department Head of Computer Science (University of Arizona). His research interests are centered on reproducibility in Computer Science and the impact of programming tools on programmer productivity. Before coming to the U of A in 2012, he worked at Microsoft for 15 years, where he founded Microsoft's efforts in using prediction markets to help forecast future events.
We welcome faculty, students, and staff of the Philosophy and Moral Science Departments as well as members of the wider University community. RSVP to Lucy Schwarz at firstname.lastname@example.org.