Abstract: In this talk, I will talk about the emerging area of neurosymbolic programming, a new approach to learning that sits at the intersection of program synthesis and machine learning. Neurosymbolic programming combines neural and symbolic reasoning to learn models that have program-like structure. The goal is to develop models that are more interpretable, that better capture domain knowledge or that generalize better from small amounts of data. The talk will describe some of the challenges in neurosymbolic programming as well as some of the efforts from my group in this direction.
Bio: Armando Solar-Lezama is a professor at MIT where he leads the Computer Aided Programming group and is also Associate Director and COO of the Computer Science and Artificial Intelligence Lab. He is also the founder of playskript.com, an online platform for creating interactive presentations and visualizations.