Diagnostic Hypothesis Enumeration vs. Probabilistic Inference for Hierarchical Automata Models
Paul Maier, Dominik Jain, Martin Sachenbacher · mediaTUM – the media and publications repository of the Technical University Munich (Technical University Munich) · 2011
AI problems in engineering domains often lie at the intersection of model-based diagnosis and probabilistic reasoning.A typical example is the plan assessment problem studied in this paper, which comprises the identification of possible faults and the computation of remaining success probabilities based on a model.In addition, solutions to such problems need to be tailored towards the needs of engineers, and thus use high-level, expressive modeling formalisms such as probabilistic hierarchical constraint automata (PHCA).This work introduces a translation from PHCA models to lower-level Bayesian nets, which enables to compare model-based diagnosis approaches with a wide array of probabilistic reasoning methods.Using a state-of-the-art probabilistic solver, we compare this approach to an alternative model-based diagnosis approach that translates the PHCA models to lower-level logic models and computes solutions by enumerating most likely hypotheses.Experimental results on realistic problem instances demonstrate that the probabilistic reasoning approach is a promising alternative to the model-based diagnosis approach.