La División Académica de Ingeniería los invita al Seminario Híbrido que se llevará a cabo el próximo 31 de marzo a las 12:00 en el salón SA3. Generative models for graphs-challenges and responses, que impartirá la Dra. Magda Gregorová, Head of Center for Artificial Intelligence and Robotics, Technical University of Applied Science Wúrzburg Schweinfurt, Germany.

Abstract: Deep generative models aim at learning the probability distributions of high-dimensional data. Having learned the distributions, the models can sample from these, producing new realistic data examples. Great progress have been achieved on domains such as images, videos or language. The task of graph generation poses several important challenges due to the discreteness of the data structures and complex internal relationships.

In this talk, after a review of the basic classes of generative models, we will speak about the challenges of graph generative modeling and discuss some of the existing responses and our own approach to the problem.