Towards a Multiagent Design Principle: Analyzing an Organizational-Learning Oriented Classifier System
Keiki Takadama, Takao Terano, Katsunori Shimohara, Koichi Hori, Shinichi Nakasuka · Studies in fuzziness and soft computing · 2001
This paper addresses a big issue of a multiagent design principle by exploring our model in terms of its generality , scalability , and performance To investigate these aspects in our model, we apply it into another domain, analyze its characteristics in large-scale problems, and compare its performance with that one of conventional models. Intensive simulations on a complex domain problem reveal the following implications: (1) our model shows its effectiveness in another domain, maintains its effectiveness in large-scale problems, and achieve a better performance than conventional models; (2) three key elements derived from our model have the potential to be important and essential factors towards multiagent design principles; and (3) the interpretation of general concepts from a computational viewpoint is one of the useful ways of addressing multiagent design principles.