Modeling Dynamics, Adaptivity and Control by Self-modeling Networks

Networks provide an intuitive, declarative way of modeling which has turned out to be suitable for many types of applications that involve complex dynamics. In many cases also adaptivity plays a role. By using algorithmic or procedural descriptions for the adaptation processes as is often the approach followed, easily leads to less declarative and transparent forms of modeling. This chapter exploits the notion of self-modeling network that has been developed recently to avoid this. According to this approach, adaptivity is obtained by adding a self-model to a given base network, with states that represent part of the network’s structure. This results in a two-level network. The self-modeling construction can easily be iterated so that multiple orders of adaptation can be covered as well. In particular, a three-level self-modeling network can be used to integrate dynamics, adaptivity and control in one network. In this chapter, it is shown how this can provide useful building blocks to design network models for learning within an organisation.