The seminar is based on Yoshua Benjio's presentation "From System 1 Deep Learning to System 2 Deep Learning" (NeurIPS-2019), which examines the transformation of AI research directions.
The first part of the workshop is given an introduction to the problem area namely what are the considered system-1 and system-2, as well as an answer to the question about the current state of affairs.
In the second part of the seminar, new and key "players" of transformation is discussed are out-of-distribution data, the concepts of generalization and agency, etc., and their interrelationships are highlighted.
The third and final part of the seminar is about the target object of modern research which is consciousness, the possibilities of its mathematical representation, as well as the stumbling blocks on the way of "conquering" this object.
Speaker: Natalya Shevskaya, Assistant of the Department of Software Engineering and Computer Applications, ETU "LETI"