Learning Classifier Systems meet Multiagent Environments
Keiki Takadama, Takao Terano, Katsunori Shimohara, W. Stolzmann, S.W. Wilson (Eds.) · 2000
this paper starts by employing our Organizationallearning oriented Classier System (OCS) [Takadama 99], which is an extension of LCS to a multiagent architecture via the introduction of the concept of organizational learning (OL) [Argyris 78] in management and organization science, and explores elements and approaches for coping with multiagent environments by investigating the characteristics of OCS through a comparison of its performance with that of conventional LCSs which include the Michigan and Pittsburgh approaches. Since OCS has several dierent aspects from those of conventional LCSs, such comparisons among LCSs oer the potential of nding elements needed in conventional LCSs to cope with multiagent environments and also the potential of nding approaches which able to contribute to coping with multiagent environments.