报告题目：Abstraction in Situation Calculus Action Theories
主 讲：Yves Lesperance, Associate Professor, EECS, York University
时 间：上午9:30 – 10:30
主 持：刘咏梅 教授
In this talk, I will present a general framework for agent abstraction based on the situation calculus and the ConGolog agent programming language. We assume that we have a high-level specification and a low-level specification of the agent, both represented as basic action theories. A refinement mapping specifies how each high-level action is implemented by a low-level ConGolog program and how each high-level fluent can be translated into a low-level formula. We define a notion of sound abstraction between such action theories in terms of the existence of a suitable bisimulation between their respective models. Sound abstractions have many useful properties that ensure that we can reason about the agent's actions (e.g., executability, projection, and planning) at the abstract level, and refine and concretely execute them at the low level. We also characterize the notion of complete abstraction where all actions (including exogenous ones) that the high level thinks can happen can in fact occur at the low level. I will also outline how the framework can be extended to online executions where the agent can acquire new knowledge through sensing as it executes. Finally, I will briefly discuss how the framework can be used to more efficiently perform agent supervision, i.e., minimally restrict the agent's behaviour to ensure it satisfies some specification.
This is joint work with my PhD student Bita Banihashemi as well as with Giuseppe De Giacomo of Sapienza University of Rome; the core aspects of the framework have appeared in the Proc. of AAAI 2017.
Yves Lesperance is Associate Professor in the Department of Electrical Engineering and Computer Science, York University, Toronto, Canada. He received a Ph.D. in Computer Science from the University of Toronto in 1991. His research is in the areas of knowledge representation and reasoning and intelligent agents and multiagent systems. His work on agent programming languages (Golog and ConGolog) and on reasoning about action and mental states in the situation calculus has had a major impact. He is the author of over 120 scholarly publications. He is Associate Editor of both Artificial Intelligence Journal and the Journal of Artificial Intelligence Research (JAIR), the two top journals in the field. He was Local Organization Chair of the International Joint Conference on Autonomous Agents and Multiagent Systems (AAMAS) in 2010. He was also a member of the Board of the International Foundation for Autonomous Agents and Multiagent Systems (IFAAMAS) from 2007 to 2010.