Statistical Relational Artificial Intelligence
Papers from the 2013 AAAI Workshop
Vibhav Gogate, Kristian Kersting, Sriraam Natarajan, David Poole, Workshop Cochairs
The main purpose of the Statistical Relational AI workshop is to bring together researchers and practitioners from two subfields of AI: logical (or relational) AI and probabilistic (or statistical) AI. (The first and second workshops on this topic were held in conjunction with AAAI-2010 and UAI-2012 respectively, and were among the most popular workshops at the respective conferences.) Despite the fact that the two fields share many key features and often solve similar problems and tasks, research in them has progressed independently with little or no interaction. Moreover, the two fields often use different notation and terminology making sharing results rather difficult and cumbersome. Our long term goal is to change this by achieving a synergy between logical and statistical AI and this workshop will serve as a stepping stone towards realizing this big picture view on AI.
Since its inception in the late 1990s, statistical relational AI has enjoyed great success. Perhaps, its main success stories has been lifted probabilistic inference — probabilistic inference algorithms that exploit symmetry. Symmetries of models have been explored in many AI tasks such as (mixed) integer programming, SAT, CSP, and MDPs. Surprisingly, however, until recently, symmetries have not been the subject of interest within probabilistic inference. Powerful representation and reasoning statistical relational AI tools have enabled several new applications in diverse domains such as social networks, natural language processing, bioinformatics, the web, robotics and computer vision.