Artificial Immune Systems: A New Computational Approach

de Castro, Leandro N. and Timmis, Jon (2002) Artificial Immune Systems: A New Computational Approach. Springer-Verlag, London. UK., 357 pp. ISBN 1852335947. (The full text of this publication is not currently available from this repository. You may be able to access a copy if URLs are provided)

The full text of this publication is not currently available from this repository. You may be able to access a copy if URLs are provided. (Contact us about this Publication)


When one reads through the current literature on artificial immune systems (AIS), a common observation is: ''the field of AIS is too young to be well defined, and its scope and limitations are still unknown''. Indeed, as remarked by I. Stewart while referring to chaos in his best seller Does God Play Dice? ''few were willing to offer a precise definition. This isn't unusual in a 'hot' research area - it's hard to define something when you feel you still don't fully understand it.'' Despite this, we feel that in order to help promote and consolidate the emerging area of AIS, an attempt should be made at drawing together what sometimes seems to be very disparate work. In this book we make an effort to bring many ideas from AIS together into a single text that can provide some basics for AIS. It is our hope that this will make the field more accessible to the wider community and also begin the process of formalizing AIS through the introduction of an engineering framework. The majority of the ideas contained in this volume constitute an outcome of Leandro's Ph.D. thesis undertaken at the State University of Campinas - Unicamp, Brazil. Between the two of us, we reviewed, extended, discussed, and improved these ideas in order to achieve the final result that is now this book. The motivation to write this book came from several parts: from the referees of Leandro's viva, from several conversations we had with one another, and from comments of many researchers and research students in the broad area of computational intelligence. Most importantly, we felt that the field lacked a textbook, but we (of course!) do not claim this text is going to answer all questions. Indeed, we see this very much as a first attempt and hopefully not the last one. We hope it will help to mature the field and inspire researchers in AIS and many other areas to gain a better understanding of such a new, rich, and exciting research area. In order to set the scene for our book, we begin discussing themes such as computing with biological metaphors and computational intelligence. There then follows a discussion on the fundamentals of the biological immune system. It was very difficult to decide how deep we should go into biological terminology, and it is possible that some readers may think that we have gone a bit over the top. However, we feel what we have produced is a compromise of offering a text that makes the biological language simple for computer scientists and engineers, but one that is accurate and provides enough terminology so as to prepare the reader to understand the contents of the book itself and also the related literature on AIS. From biology, there emerges a proposed framework on how to engineer an AIS. We observed from the literature that there were a number of common building blocks, which would make an ideal common framework to design AIS. We then try to exhaustively survey the publications on AIS. Instead of briefly describing every work cited, we identify the major application domains, describe one work of each research school and reference the others. We focus on the immune metaphors employed by the authors and how their approaches suit the framework introduced. We then again turn back to biology. Chapter 5 is strongly biological and one would probably raise the question ''do we actually need all this biology?'' It is worth noting that by presenting biology in a broader context, it allows us to understand the wider picture played by the immune system with other organisms. When viewed in relation to other systems, the evolution of species, and cognition, it is easier to explain some of the immune system's behavior and to compare this behavior with the behavior of other systems. Chapter 5 also serves the purpose of reviewing the biological motivation for the development of several computational intelligence tools, such as neural networks and evolutionary algorithms, to be discussed in Chapter 6. Artificial immune systems are hybrid systems almost by their very nature, and thus, this book could not restrict itself to a discussion of this single theme. It goes far beyond the AIS domain and discusses several other computational intelligence paradigms. Among these, we focus on artificial neural networks and evolutionary algorithms. One of our motivating factors for this is the fact that it is not unusual to hear questions concerning the distinction between an AIS and a genetic algorithm, immune network models and neural networks, and so on. One point to note is that, the flavor of the book might be seen deliberately philosophical in parts. This is an attempt on our end to place emphasis on underlying concepts, knowing that in this rapidly developing area the specifics may change very quickly. That's what our book is about; computational intelligence focused on the emerging field of artificial immune systems. We hope that it helps to shape the field and that it serves as a guide for you to understand and engineer your own AIS. Leandro Nunes de Castro & Jonathan Timmis Canterbury, April 2002

Item Type: Book
Uncontrolled keywords: artificial immune systems, classification, supervised learning, immune networks, case studies, AIS framework, survey
Subjects: Q Science > QA Mathematics (inc Computing science) > QA 76 Software, computer programming,
Divisions: Faculties > Sciences > School of Computing > Applied and Interdisciplinary Informatics Group
Depositing User: Mark Wheadon
Date Deposited: 24 Nov 2008 17:59 UTC
Last Modified: 01 Jul 2014 10:26 UTC
Resource URI: (The current URI for this page, for reference purposes)
  • Depositors only (login required):