Geometry of stable cliques in neural networks

Meeting Time: Nov. 3, 2009, 2:00-2:50pm

Abstract: We investigate the relationship between the structure (connectivity) and function (stimulus space representation) for a simple class of neural network models. The problem of relating network connectivity to stimulus space properties leads to a number of geometric questions about the set of stable cliques associated to a particular network. I will discuss some recent results relating to these questions. This is joint work with Anda Degeratu and Vladimir Itskov.