Generalized Dot Product Models for Weighted Networks

Daryl Deford

Mathematics, Dartmouth College.


The dot product model was defined in 1998 as a combinatorial method for efficiently representing graphs. More recently, randomized versions of this model have been shown to generate networks with small world properties and a vector embedding based on this model provides an effective framework for statistical inference for stochastic block models. In this talk I will describe a generalized version of the dot product model for networks with weighted edges focusing on the relationship between community structure and the vector embedding.

Back to ACMS schedule