Sebastijan Dumančić - Delft University of Technology
Nov. 26, 2021, 10 a.m. - Nov. 26, 2021, 11 a.m.
Virtual (see link below)
Hosted by: Breandan Considine
Abstract: Learning and reasoning paradigms are often seen as being on the opposing side in AI. One advertising for pure blank state and the other one for innate knowledge. The research directions trying to unite both paradigms have only recently risen to prominence. This recent surge of interest has made the field somewhat prone to baby issues — what exactly is meant by reasoning and learning is left ambiguous, and many different flavours are often considered interchangeable. This often makes the contribution of each of the components ‘murky’. In this talk, I will show you how explicitly acknowledging the role of a reasoning system, i.e., logic, can help us make better and more reliable program synthesis/induction and neuroscience-symbolic AI approaches.
Bio: Sebastijan is an Assistant Professor at the Delft University of Technology, the Netherlands. He finished his PhD at KU Leuven. Belgium, in 2018 and has been as PostDoc at the Research Foundation Flanders & KU Leuven. His interests include program synthesis, probabilistic programming, and neuro-symbolic AI. He has received an honorary mention for his PhD thesis at the EurAI Dissertation Award.
Zoom link: https://mcgill.zoom.us/j/84280715850
Virtual reception after the talk in Gather: https://gather.town/app/tYHHMh7tPcPw9037/reception