LEAF: a toolkit for developing coordinated learning based MAS

This paper describes LEAF, the "Learning Agent based FIPA-Compliant Community Toolkit", a toolkit for developing multiagent systems coordinated using utility function assignment, based on collective intelligence by Wolpert et al. (1999). LEAF agents use machine learning techniques such as reinforcement learning to maximise local utility functions, where local utility functions are assigned to agents such that the maximisation of local utility by agents within a community maximises a global utility. LEAF provides support via a Java API for developing FIPA-compliant agent systems conforming to this framework, utilising the FIPA-OS agent toolkit, a Java based FIPA compliant agent construction toolkit.