.
Speaker: Markus Reineke
Title: Quiver representations and neural networks
Abstract: Quiver representations formalize classification problems of linear algebra. We will review some basic concepts and aims of the representation theory of quivers, in particular the algebraic-geometric approach via quiver moduli spaces. We will model (certain aspects of) neural networks using the quiver language, and discuss what quiver representation theory can say, both qualitatively and quantitatively, about the space of network functions.
We will meet in the following coordinates: