Bang 3: a computational multi-agent system
Roman Neruda, Pavel Krušina, P. Kudova, Pavel Rydvan, Gerd Beuster · 2004
A multi-agent system targeted toward the area of computational intelligence modeling is presented. The purpose of the system is to allow both experiments and high-performance distributed computations employing hybrid computational models. The focus of the system is the interchangeability of computational components, their autonomous behavior, and emergence of new models. 1. Computational Intelligence Bang 3 is a platform for the development of Multi-Agent Systems (MAS) [3]. Its main areas of application are computational intelligence methods (genetic algorithms, neural networks, fuzzy controllers) on single machines and clusters of workstations. Hybrid models, including combinations of artificial intelligence methods such as neural networks, genetic algorithms and fuzzy logic controllers, seem to be a promising and extensively studied research area [1]. We have designed a distributed multi-agent system [4] called Bang 3 that provides a support for an easy creation of hybrid AI models by means of autonomous software agents [2]. Besides serving as an experimental tool and a distributed computational environment, this system should also allow to create new agent classes consisting of several cooperating agents. The MAS scheme is a concept for describing the relations within such a set of agents. The basic motivation for schemes is to describe various computational methods. It should be easy to ‘connect ’ a particular computational method (implemented as an agent) into hybrid methods, using schemes description. The scheme description should be strong enough to describe all the necessary relations within a set of agents that need to communicate one with another in a general manner. Here we show, how two computational intelligence methods — artificial neural network of the RBF type, and a