DAI-List Digest Friday, 2 March 1990 Issue Number 4 Topics: DAI Tutorial at AAAI-90 DAI Tutorial at IEA/AIE-90 Job Opportunity in DAI Please send submissions to DAI-List@mcc.com. Send other requests, such as changes in your e-mail address, to DAI-List-Request@mcc.com. ---------------------------------------------------------------------- Date: Fri, 2 Mar 90 10:50:17 CST From: gasser@pollux.usc.edu Subject: DAI Tutorial at AAAI-90 Distributed Artificial Intelligence by Les Gasser and Jeffrey Rosenschein Preface Organized systems of coordinated problem solvers are now a research reality, and are rapidly becoming practical partners in critical human problem-solving environments. At a panel on high-impact directions for AI research at IJCAI-89, several panelists cited Distributed AI as the primary high-impact area. The growth and importance of coordinated problem-solving systems is indisputable; reasons for studying them include: individual intelligent processes have limited capacity to solve large problems; the distributed computing technology to support coordinated problem solving is now available; we would like to build coordinated problem solvers systems for research purposes, as testbeds for distributed reasoning, and for performance studies; and distributed intelligence systems exist, and we would like explanatory and prescriptive theories that account for them. Content We will begin with the motivation for distributed AI (DAI) and a brief summary of the history of research in the area. In discussing motivation, we will highlight the insight provided by group interaction and social organization as metaphors for computation. We will then survey some domains in which DAI appears to be particularly well suited, including: organizational information systems; manufacturing and robotics; design; monitoring, control, and diagnosis; and distributed sensing and interpretation. We will then discuss approaches to each of six basic DAI problems. The first is problem decomposition; we will discuss the use of multiple interacting knowledge bases, distributed planning, and alternative ways of decomposing a problem. The second is communication and interaction among the distributed agents; here we will discuss communication and interaction protocols and the use of Speech Act theory as a basis for communication. The third problem is the maintenance of coherence during problem solving; we will discuss issues of distributed control and organization. The fourth problem is modeling other agents; we will talk about modeling problem-solving capabilities, beliefs, plans, goals, and the collaborative process itself. The fifth problem is recognizing and resolving disparities; here we will discuss the FA/C Model, negotiation, belief nets, ATMSs, defaults, open systems, microtheories, and Due Process. The final problem is implementation; we will talk about object-based concurrent programming languages such as Actors, reflective languages, frameworks such as ABE, MACE, and AGORA, and blackboard systems. Intended Audience This tutorial is targeted for people who are interested in building DAI systems, for AI researchers interested in learning about DAI approaches, and possibly for technology planners and managers who need to know about leading-edge AI technologies. The tutorial presumes knowledge of AI at the level of an introductory course, and familiarity with such general concepts as first-order predicate calculus, object-oriented systems, Lisp, hierarchical and nonlinear planning, heuristic search, knowledge-based systems, reasoning under uncertainty, and so on. Speakers Les Gasser received his B.A. in English from the University of Massachusetts, and his M.S. and Ph.D. in Computer Science from U.C. Irvine. Dr. Gasser is on the faculty of Computer Science at the University of Southern California, where he leads the Distributed Artificial Intelligence Research Group. He developed and teaches the advanced graduate course on Distributed AI at USC, and has published two books on Distributed AI: Readings in Distributed Artificial Intelligence (With A. Bond, Morgan Kaufmann, 1988) and Distributed Artificial Intelligence, Volume II (With Michael N. Huhns, Pitman Publishers/Morgan Kaufman, 1989). He chaired the 1988 Workshop on Distributed Artificial Intelligence, and has published widely in the field. Jeffrey S. Rosenschein received his A.B. in Applied Mathematics from Harvard University, and his M.S. and Ph.D. in Computer Science from Stanford University. He is currently a Lecturer in the Computer Science Department at Hebrew University, Jerusalem, Israel. His dissertation on "Rational Interaction: Cooperation Among Intelligent Agents" broke new ground in the Artificial Intelligence community with its study of principles of multiple agent interactions. His research since that time has continued to focus on issues of cooperation and competition among high-level problem solving entities. Dr. Rosenschein has taught courses at Hebrew University and in industry on introductory and advanced topics in Artificial Intelligence, and has authored various articles on multiagent problem solving. ---------------------------------------------------------------------- Date: Fri, 2 Mar 90 10:50:17 CST From: huhns@mcc.com Subject: DAI Tutorial at IEA/AIE-90 Distributed Artificial Intelligence by Michael N. Huhns and Larry M. Stephens to be presented at the Third International Conference on Industrial and Engineering Applications of Artificial Intelligence and Expert Systems, Charleston, SC, July 15-18, 1990. Overview and Motivation Distributed artificial intelligence (DAI) is concerned with the cooperative solution of problems by a decentralized group of agents. It is the appropriate technology for applications where 1) expertise is distributed, as in design; 2) information is distributed, as in office automation; 3) data are distributed, as in distributed sensing; 4) decisions are distributed, as in manufacturing control; and 5) knowledge bases are developed independently but must be interconnected or reused, as in next-generation knowledge engineering. Interconnecting computational agents and expert systems enables them to cooperate in solving problems, to share expertise, to work in parallel on common problems, to be developed and implemented modularly, to be fault tolerant through redundancy, to represent multiple viewpoints and the knowledge of multiple human experts, and to be reusable. Content This tutorial will describe the current state of research in distributed artificial intelligence. It will present architectures, languages, and techniques for achieving cooperative problem solving in a distributed environment, and describe several successful applications of DAI in manufacturing, information retrieval, and distributed sensing. The specific topics to be covered are 1) A brief history of DAI 2) Communication among agents, expert systems, and users: syntax 3) Communication: semantics and speech-act theory 4) Decomposing and distributing problems among agents 5) Distributing control among agents 6) Modeling global goals, the global problem-solving state, and other agents 7) Maintaining consistent beliefs among agents 8) Applications and implementations, including blackboards Intended Audience This tutorial is intended for computer scientists and engineers interested in constructing DAI systems, for computer researchers interested in learning current DAI theory, and for managers interested in staying abreast of next-generation AI technology. The audience is presumed to have a working knowledge of artificial intelligence terminology and techniques at the level of an introductory course. Speakers Dr. Michael N. Huhns is a Senior Member of the Artificial Intelligence Laboratory at MCC, where he has been directing the Argo Project in machine learning and the Antares Project in DAI. Prior to joining MCC in 1985, he was an Associate Professor of Electrical and Computer Engineering at the University of South Carolina, where he also directed the Center for Machine Intelligence. He received the B.S.E.E. degree from the University of Michigan in 1969, and the M.S. and Ph.D. degrees in electrical engineering from the University of Southern California in 1971 and 1975, respectively. Dr. Huhns is a member of IEEE, Sigma Xi, Tau Beta Pi, Eta Kappa Nu, ACM, and AAAI. He is the author of 83 technical papers in machine intelligence and an editor of the books {\it Distributed Artificial Intelligence, Volumes I and II}. His research interests are in the areas of machine learning, distributed artificial intelligence, and computer vision. Larry M. Stephens is an Associate Professor in the Department of Electrical and Computer Engineering, University of South Carolina, which he joined in 1977. He is also a consultant to MCC in the area of distributed artificial intelligence. His previous research includes intelligent retrieval of information in distributed environments, design and microprogramming of a multicomputer system, and computer graphics. His current research interests are distributed problem solving, expert system technology, and knowledge representation. Dr. Stephens was granted a B.S. degree in electrical engineering from the University of South Carolina in 1968 and received the M.S. and Ph.D. degrees, also in electrical engineering, from the Johns Hopkins University in 1974 and 1977, respectively. While at Johns Hopkins University, he was an NSF graduate fellow. Prior to earning his advanced degrees, he served as an officer in the U.S. Navy in the Naval Reactors Program. Dr. Stephens is a member of Phi Beta Kappa, Tau Beta Pi, Eta Kappa Nu, Sigma Xi, IEEE, AAAI, and ACM. ---------------------------------------------------------------------- Date: Tue, 23 Jan 90 16:48:55 MET From: Donald Steiner Subject: Job Opportunity in DAI Dear Colleague, Enclosed is an announcement for research positions that may be of interest to you, your colleagues, or students. I would appreciate it very much if you would pass it on to those who might be interested. The German Research Center for Artificial Intelligence (DFKI) is collaborating with Siemens AG on the KIK project aiming to integrate artificial intelligence and communication technologies. A subproject of KIK (TEAMWARE) is developing foundations for work in a distributed man-machine scenario. This will include methods for support of communication and cooperation in a spatially and temporally distributed team of humans and diverse (partly)autonomous machine agents. The DFKI is looking for researchers to participate in the TEAMWARE project in Kaiserslautern. Along with research into the foundational theoretical aspects of DAI, the candidates should be willing to contribute to the development of a functional application. The candidates should have experience in at least two of the following areas: - Distributed Artificial Intelligence Distributed Architectures Cooperation Formalisms Interaction Languages - Man-Machine Interaction and Cooperation - AI Languages (Lisp, Prolog) - Distributed AI Languages (MACE, ACTOR related languages etc.) - Networks and Protocols - Development of large complex software systems. For more information about this position, contact Donald Steiner at steiner@uklirb.informatik.uni-kl.de (CSNET). If you wish to apply, please send your resume to: Prof. Dr. G. Barth Geschaeftsleitung DFKI Postfach 20 80 D-6750 Kaiserslautern Federal Republic of Germany The DFKI is a young and dynamic research consortium, funded by major German computer companies and the German Ministry for Research and Technology and is located in Kaiserslautern and Saarbruecken. SIEMENS is one of the largest computer and telecommunication companies in Europe.