Charles River Analytics, a developer of intelligent systems solutions, sponsored the 2012 Conference on Uncertainty in Artificial Intelligence (UAI), held in Catalina Island, CA, on August 14-18, 2012. UAI is the leading international conference on uncertainty management and reasoning. Charles River’s sponsorship of UAI helps enable researchers and students from around the world to come together to discuss innovative research in this key area.
As part of the conference’s workshop on Statistical Relational Artificial Intelligence (StarAI), Dr. Avi Pfeffer presented a poster on Charles River’s Figaro probabilistic programming language, which enables creation of rich, powerful probabilistic models with built-in reasoning capabilities. In addition, Dr. Pfeffer and his colleagues Dr. Siddharth Srivastava and Professor Stuart Russell of the University of California, Berkeley, presented a joint paper on sensing and planning in “open universe” situations where the number and identity of objects is unknown.
Charles River’s sponsorship of UAI reflects its ongoing commitment to the development of novel probabilistic modeling technologies, which Charles River supports through multiple projects. Under a DARPA-funded project called PRIDE, Charles River is developing a framework for probabilistic modeling in a wide range of complex situations. Probabilistic modeling determines a situation’s unknown facts from known facts—a key reasoning task that is difficult because of the uncertainty and unpredictability of many factors, like sensors giving incomplete or noisy readings. Probabilistic modeling creates a model of a situation that captures the known facts about a situation and then uses probability theory to compute beliefs about the unknown facts.
Another project, ACACIA, funded by the Air Force, applies this kind of probabilistic framework to help Department of Defense intelligence analysts understand what an adversary is capable of doing and is likely to do. ACACIA is powered by a probabilistic “engine” that uses Figaro to perform complex assessments of possible adversary intentions and capabilities.