Midwest Computability Seminar

XXVIII
Part iii


The Midwest Computability Seminar will meet remotely in the winter and spring of 2022, jointly with the Computability Theory and Applications Online Seminar. The recurring Zoom link is:

https://notredame.zoom.us/j/99754332165?pwd=RytjK1RFZU5KWnZxZ3VFK0g4YTMyQT09

Meeting ID: 997 5433 2165

Passcode: midwest


slides    Panopto video    YouTube video


DATE: Monday, February 21st, 2022

TIME: 3:30 - 4:30 PM Central Time


SPEAKER: Ekaterina Fokina - Vienna University of Technology

TITLE: Algorithmic learning of structures

ABSTRACT: Assume we have a class of structures closed under isomorphism. Assume further, we receive information about one of these structures step by step: finitely much information at each step. Our goal is to determine, after finitely many steps, which structure from the class we are observing. If we can reach the goal, we call the class learnable. In the talk we formalise various aspects of this problem using ideas from computable structure theory and computational learning theory. We give syntactic characterisations for several notions of learnability and apply these results to get examples of learnable and non-learnable classes of structures.



Past and Future Sessions

Previous Seminars:

If you haven't been receiving the announcements and would like to be included in the list, send an email to drh@math.uchicago.edu.