A graphical framework for stochastic model-based diagnosis

Diagnosing systems with uncertainty has significant practical importance. Many different methods for performing diagnostics inference on stochastic systems have been developed in fields such as FDI and AI. We provide a factor graph framework that integrates several of these approaches for diagnosing stochastic systems. This integration provides several advantages, e.g., showing inter-relationships among the inference algorithms, a computational toolbox for solving diagnostics problems, and an a priori means for predicting inference complexity based solely on the graph structure.