College of Information and Computer Sciences (CICS) alum Daniel Bernstein, professor Shlomo Zilberstein and professor Neil Immerman were selected to receive the 2019 Influential Paper Award in Autonomous Agents and Multiagent Systems by the International Foundation for Autonomous Agents and Multiagent Systems (IFAAMAS) for their two papers, both titled, “The Complexity of Decentralized Control of Markov Decision Processes.” The first version of the paper, based on Bernstein’s doctoral synthesis project co-advised by Zilberstein and Immerman, was presented at the Conference on Uncertainty in Artificial Intelligence in 2000, and the extended version, co-authored by Robert Givan of Perdue University, was published in Mathematics of Operations Research in 2002.
As the IFAAMAS award committee explains, “These papers formally introduced the decentralized partially observable Markov decision process (Dec-POMDP), launching a subfield on principled models and solution methods for multiple cooperative agents with uncertainty and limited communication. Since then, the influence of the paper has spread widely and was followed by numerous other publications that include many theses, journal articles, conference papers and a recent book. Dec-POMDP methods have become well known in the AI community (e.g., becoming a popular model for deep multi-agent reinforcement learning) and have begun to be used in fields such as robotics and networking.”
The award was presented at the AAMAS 2019 (International Conference on Autonomous Agents and Multiagent Systems) held in Montréal, Canada from May 13-17. AAMAS is the flagship conference of IFAAMAS, a non-profit organization chartered to promote science and technology in the areas of artificial intelligence, autonomous agents and multiagent systems.
“I am thrilled to see the work recognized by the AAMAS community,” said Zilberstein. He noted that Dan Bernstein’s doctoral dissertation created the foundation for studying Dec-POMDPs, and that other CICS alumni have made significant contributions to the development of the model and its applications.