Timezone: »

 
Poster
A Convex Formulation for Learning Scale-Free Networks via Submodular Relaxation
Aaron Defazio · Tiberio Caetano

Wed Dec 05 07:00 PM -- 12:00 AM (PST) @ Harrah’s Special Events Center 2nd Floor #None

A key problem in statistics and machine learning is the determination of network structure from data. We consider the case where the structure of the graph to be reconstructed is known to be scale-free. We show that in such cases it is natural to formulate structured sparsity inducing priors using submodular functions, and we use their Lovasz extension to obtain a convex relaxation. For tractable classes such as Gaussian graphical models, this leads to a convex optimization problem that can be efficiently solved. We show that our method results in an improvement in the accuracy of reconstructed networks for synthetic data. We also show how our prior encourages scale-free reconstructions on a bioinfomatics dataset.

Author Information

Aaron Defazio (Ambiata)
Tiberio Caetano (NICTA Canberra)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors