The Frame Problem and Bayesian Network Action Representations* [chapter]

Craig Boutilier, Moisés Goldszmidt
1999 Logical Foundations for Cognitive Agents  
We examine a number of techniques for representing actions with stochastic e ects using Bayesian networks and in uence diagrams. We compare these techniques according to ease of speci cation and size of the representation required for the complete speci cation of the dynamics of a particular system, paying particular attention the role of persistence relationships. We precisely characterize two components of the frame problem for Bayes nets and stochastic actions, propose several ways to deal
more » ... eral ways to deal with these problems, and compare our solutions with Reiter's solution to the frame problem for the situation calculus. The result is a set of techniques that permit both ease of speci cation and compact representation of probabilistic system dynamics that is of comparable size (and timbre) to Reiter's representation (i.e., with no explicit frame axioms). 3 IDs are representational tools used for optimal decision making in decision analysis. Actions are usually referred to as decisions, but for our purposes the two can be considered equivalent. 4 There are a number of other representational methods that deserve analysis (e.g., the event calculus 17], the A language of 12] and its variants, probabilistic STRIPS rules 18, 2], probabilistic Horn rules 24]) which unfortunately we cannot provide here; but see the full paper 6].
doi:10.1007/978-3-642-60211-5_5 fatcat:6qewp2zirzejznhvk72vovupbi