Beyond Truth & Falsehood: Logic as a Calculus of Events

Logic in AI has traditionally been cast as a tool for representing and reasoning with categorical beliefs (i.e., a calculus of truth and falsehood). However, over the last two decades, the role of logic in AI has been largely as a calculus of events, where it was effectively and profoundly employed in tasks such as probabilistic reasoning, machine learning and explainable AI. In this lecture/tutorial, which celebrates World Logic Day, Adnan Darwiche of UCLA discusses some of the fundamental insights, techniques and tools that have emerged from this broader role of logic in AI. He further provides concrete and intuitive examples of how using logic as a calculus of events has enabled the development of state-of-the-art AI systems that transcend the traditional role of logic that is typically emphasized in AI textbooks. World Logic Day is an international day proclaimed by UNESCO in association with the International Council for Philosophy and Human Sciences (CIPSH) in November 2019 to be celebrated on 14 January every year. This is the date of death of Kurt Gödel and the date of birth of Alfred Tarski, two of the most prominent logicians of the twentieth century (). 02:07 events are everywhere 05:00 basics of events 07:55 probabilistic reasoning 27:25 machine learning 34:48 explainable AI 54:57 the calculus of events 55:46 pointers 56:33 final messages Related references/tutorials by Adnan Darwiche: --Three Modern Roles for Logic in AI: --Logic for Explainable AI: --Tractable Boolean and Arithmetic Circuits: