Machine Learning, Parsing, and Zippers: Derivative Works in Computer Science

I'm excited to announce that Brian Howard (Associate Professor at Depauw University) will be giving us a really interesting talk on the user of derivatives in programming. If you are interested in friendly introduction to some deeper concepts in the programming world, you should definitely attend. As always, we will have free Bazbeaux's Pizza so please RSVP. You can see the full talk description below.

---------------

While high school calculus might not seem all that relevant to most programming tasks, this talk will explore some unexpected connections between derivatives and (functional) programming. Starting with a brief review of numeric and symbolic differentiation, we will see how operations on "dual" numbers provide an explanation of back-propagation in machine learning. Next we'll look at derivatives of sets of strings as a way to construct efficient parsers from regular expressions and context-free grammars. Finally, by taking the derivative of a parameterized type, we will see how to work with "zippers" -- a technique for navigating around tree-like data structures in a purely functional way.