Distributed Artificial Intelligence Meets Machine Learning Learning in Multi-Agent Environments

The complexity of systems studied in distributed artificial intelligence (DAI), such as multi-agent systems, often makes it extremely difficult or even impossible to correctly and completely specify their behavioral repertoires and dynamics. There is broad agreement that such systems should be equipped with the ability to learn in order to improve their future performance autonomously. The interdisciplinary cooperation of researchers from DAI and machine learning (ML) has established a new and very active area of research and development enjoying steadily increasing attention from both communities. This state-of-the-art report documents current and ongoing developments in the area of learning in DAI systems. It is indispensable reading for anybody active in the area and will serve as a valuable source of information.

Inhalt
Reader's guide.- Challenges for machine learning in cooperative information systems.- A modular approach to multi-agent reinforcement learning.- Learning real team solutions.- Learning by linear anticipation in multi-agent systems.- Learning coordinated behavior in a continuous environment.- Multi-agent learning with the success-story algorithm.- On the collaborative object search team: a formulation.- Evolution of coordination as a metaphor for learning in multi-agent systems.- Correlating internal parameters and external performance: Learning Soccer Agents.- Learning agents' reliability through Bayesian Conditioning: A simulation experiment.- A study of organizational learning in multiagents systems.- Cooperative Case-based Reasoning.- Contract-net-based learning in a user-adaptive interface agency.- The communication of inductive inferences.- Addressee Learning and Message Interception for communication load reduction in multiple robot environments.- Learning and communication in Multi-Agent Systems.- Investigating the effects of explicit epistemology on a Distributed learning system.