Artificial immune systems for optimisation

Artificial immune systems (AIS) are a class of biologically inspired algorithms which are build after different theories from immunology. While the field of AIS is a relatively new area of research, it has achieved numerous promising results in different areas of application, e.g., learning, classification, anomaly detection, and optimization. In this tutorial we focus in particular on AIS build for the purpose of optimization. From an algorithmic point of view AIS show on a high level similarities to other biologically inspired algorithms, e.g. evolutionary algorithms. Due to their different origin concrete AIS for optimization are quite different from evolutionary algorithms. They constitute an interesting alternative approach to current methods. The tutorial gives an overview over different methods in the field of AIS. It addresses everyone who wants to broaden his or her area of research within this emerging field, both practitioners and theoreticians. It enables attendees without prior knowledge of AIS to learn about a novel kind of optimization method that can be used as an alternative to other biologically inspired algorithms. Moreover, it gives researchers with prior knowledge of AIS the opportunity to deepen their understanding of the considered algorithms.

We start with an overview over the different areas of AIS, including different general approaches and some immunological background. Afterwards, we discuss several examples of AIS for optimization. We introduce concrete algorithms and their implementations and point out similarities and differences to other biologically inspired algorithms. In the last part of the tutorial, we present an overview over recent theoretical results for this kind of algorithms.