In this paper we consider planning problems in relationalMarkov processes where objects may “appear” or “disap-pear”, perhaps depending on previous actions or propertiesof other objects. For instance, problems which require to ex-plicitly generate or discover objects fall into this category. Inour formulation this requires to explicitly represent the un-certainty over the number of objects (dimensions or factors)in a dynamic Bayesian networks (DBN). Many formalisms(also existing ones) are conceivable to formulate such prob-lems. We aim at a formulation that facilitates inference andplanning. Based on a specific formulation we investigate twoinference methods—rejection sampling and reversible-jumpMCMC—to compute a posterior over the process conditionedon the first and last time slice (start and goal state). We willdiscuss properties, efficiency, and appropriateness of eachone.
|Publication status||Published - 2013|
|Event||2013 AAAI Fall Symposium: How Should Intelligence Be Abstracted in AI Research - |
Duration: 13 Nov 2013 → 15 Nov 2013
|Conference||2013 AAAI Fall Symposium: How Should Intelligence Be Abstracted in AI Research|
|Period||13/11/2013 → 15/11/2013|