Ryan Thompson

Structure Learning on Clustered Data

A new approach is introduced that estimates a global structure while accounting for local cluster-level effects, and presents a differentiable graph coupling mechanism that guarantees the union of the fixed- and random-effects graphs remains acyclic.