Quiver representations and neural networks

Title

Quiver representations and neural networks

Speaker

Claudio Bartocci - Università di Genova

Abstract

The theory of quiver representations has recently been used by some authors to provide an algebro- geometric formulation for neural networks in machine learning. Two different approaches to this issue will be examined and compared. In particular, G. Jeffrey and S. Lau’s attempt to construct activation functions as toric moment maps will be discussed.

Bio

Claudio Bartocci is a mathematician working at the University of Genova (Italy). His recent research focuses on moduli spaces of quiver representations.

When

March 218th 2022, 15:00

Where

Room 508, UniGe DIMA, Via Dodecaneso 35, Genova, Italy.

Streaming will be available at the link below.