machine learning @ toronto
[ home ] [ people ] [ projects ] [ courses ] [ meetings ]


Semi-supervised and Hierarchical AP


Monday, March 31th 2008 — Inmar Givoni


Abstract:
 

In this talk I will introduce an alternative graphical model and derivation of the affinity propagation algorithm. This formulation is particularly well suited for simplifying the max-product message derivation for the original problem, as well as for extensions of AP. I will describe two extensions of affinity propagation — semi-supervised AP and hierarchical AP, and discuss some applications, including interactive image segmentation.