Probabilistic and Relational Inferences in Dynamic Environments (PRIDE)

Reasoning under uncertainty is a critical capability for systems that support intelligent decision making in defense applications. Probability theory is a sound, widely accepted basis for reasoning under uncertainty. The environments in which decisions are made are rich and complex. Logical languages are powerful and expressive, providing the capability for making general statements and describing the world concisely and naturally. In order to support reasoning under uncertainty in rich, complex environments, we need a language that combines support for probabilistic reasoning with the ability to talk about the world in terms of objects and relationships and make general statements. We will build a system for Probabilistic and Relational Inferences in Dynamic Environments (PRIDE) that supports such a language, based on Probabilistic Relational Models (PRM). PRMs allow a model to be described in terms of classes of objects in the world, instances of those classes, and relationships between them. They make it possible to describe uncertainty about general logical statements involving entities in the world. We will extend PRMs to handle both discrete and continuous variables in full generality. We will also develop dynamic PRMs, an extension of PRMs to handle scenarios taking place over time.