Part of Advances in Neural Information Processing Systems 10 (NIPS 1997)
Richard Wilson, Edwin Hancock
Our aim in this paper is to develop a Bayesian framework for match(cid:173) ing hierarchical relational models. The goal is to make discrete la(cid:173) bel assignments so as to optimise a global cost function that draws information concerning the consistency of match from different lev(cid:173) els of the hierarchy. Our Bayesian development naturally distin(cid:173) guishes between intra-level and inter-level constraints. This allows the impact of reassigning a match to be assessed not only at its own (or peer) level ofrepresentation, but also upon its parents and children in the hierarchy.