Organic Computing (OC) is a field of biologically-inspired computer science which emerged recently as a challenging vision for future information processing systems. OC Systems are large dynamic collections of autonomous systems, which exhibit so called self-x properties. Ideally, such systems are self-organizing, self-healing, self-adaptive, self-configurating, self-protecting etc. In this talk we present our work on ``spezialization'' in OC systems and cover two topics. In the first part we deal with so called Organic Support Systems. These systems consist of autonomous components which use reconfigurable hardware in order to specialize to different support tasks. Social insect inspired methods for reconfiguration and task allocation strategies are introduced. In the second part of the talk a system is presented which uses interacting Pittsburgh-style Learning Classifier Systems to evolve rule sets for solving classification problems on distributed, autonomous, and memory constrained components. This system can overcome the restricted memory size of its components by evolving specialized components and cooperation between the components. The structure and properties of the evolved rule sets, the influence of the communication topologies, and the communication costs of the evolved patterns of cooperation are discussed.
specialization, task allocation, cooperation, organic computing