Learning of defaults by agents in a distributed multi-agent system environment

Abstract

The paper introduces a novel approach to machine learning in a multi-agents system. A distributed version of Inductive Logic Programming is used, which allows agents to construct new rules based on knowledge and examples, which are available to different memebrs of the system. The learning process is performed in two phases - first locally by each agent and then on the global level while reasoning. © Springer-Verlag Berlin Heidelberg 2013.