Selected Applications of Natural Computing David Corne, Heriot-Watt University, UK Kalyanmoy Deb, IIT Kanpur, India Joshua Knowles, University of Manchester, UK Xin Yao, University of Birmingham, UK 1. Introduction The study of Natural Computation has borne several fruits for science, industry and commerce. By providing exemplary strategies for designing complex biologi- cal organisms, nature has suggested ways in which we can explore design spaces and develop innovative new products. By exhibiting examples of effective co- operation among organisms, nature has hinted at new ideas for search and control engineering. By showing us how highly interconnected networks of simple bio- logical processing units can learn and adapt, nature has paved the way for our de- velopment of computational systems that can discriminate between complex pat- terns, and improve their abilities over time. And the list goes on. It is instructive to note that the methods we use that have been inspired by na- ture are far more than simply ‘alternative approaches’ to the problems and applica- tions that they address. In many domains, nature-inspired methods have broken through barriers in the erstwhile achievements and capabilities of ‘classical’ com- puting. In many cases, the role of natural inspiration in such breakthroughs can be viewed as that of a strategic pointer, or a kind of ‘tie-breaker’. For example, there are many, many ways that one might build complex multi-parameter statistical models for general use in classification or prediction; however, nature has exten- sive experience in a particular area of this design space, namely neural networks – this inspiration has guided much of the machine learning and pattern recognition community towards exploiting a particular style of statistical approach that has proved extremely successful. Similar can be said of the use of immune system metaphors to underpin the design of techniques that detect anomalous patterns in systems, or of evolutionary methods for design. Moreover, it seems clear that natural inspiration has in some cases led to the exploration of algorithms that would not necessarily have been adopted, but have nevertheless proven significantly more successful than alternative techniques. Par- ticle swarm optimisation, for example, has been found enormously successful on a range of optimisation problems, despite its natural inspiration having little to do with solving an optimisation problem. Meanwhile, evolutionary computation, in its earliest days, was subjected to much scepticism and general lack of attention –