Place: Online Seminar: Please sign up for our mailing list at www.physicsmeetsml.org for zoom link
Speaker: Siamak Ravanbaksh, McGill University
Abstract: A principled approach to modeling structured data is to consider all transformations that maintain structural relations. Using this perspective in deep learning leads to the design of models that are invariant or equivariant to the symmetry transformations of the data. Symmetry groups can be composed in different ways, and in this talk, I explore the utility of the compositionality of symmetry groups in deep learning. In particular, I review different notions of group composition and their application in deep model design for new and compositional data structures.