Leandro N. de Castro and Jonathan Timmis
|Book Contents | Links to AIS Researchers | Immune Algorithm Implementations | Springer-Verlag|
Artificial Immune Systems: A New Computational Intelligence Approach
Over the past few decades there has been a growing interest in the use of biology as a source of inspiration for solving computational problems. This area of research is often referred to as Biologically Inspired Computing. The motivation of this field is primarily to extract useful metaphors from natural biological systems, in order to create effective computational solutions to complex problems in a wide range of domain areas. The more notable developments have been the neural networks inspired by the working of the brain, and the evolutionary algorithms inspired by neo-Darwinian theory of evolution.
More recently however, there has been a growing interest in the use of the biological immune system as a source of inspiration to the development of these computational systems. The immune system contains many useful information-processing abilities, including pattern recognition, learning, memory and inherent distributed parallel processing. For these and other reasons, the immune system has received a significant amount of interest to use as a metaphor within computing. This emerging field of research is known as Artificial Immune Systems (AIS).
Essentially, AIS are the use of immune system components and processes as inspiration to construct computational systems. AIS is very much an emerging area of biologically inspired computation and has received a significant amount of interest from researchers and industrial sponsors in recent years. Applications of AIS include such areas as machine learning, fault diagnosis, computer security, scheduling, virus detection, and optimisation. The field of AIS is showing great promise of being a powerful computing paradigm and therefore the writing of this book is very timely.
The book will present a general introduction to the field of immunology, stressing the key areas that are currently used within the field of AIS. A framework for engineering AIS is then introduced to the reader, followed by an up to-date review of the state of the art in AIS, in then light of that framework. It is hoped that through these initial chapters the reader will become aware of the powerful metaphor of the immune system and be left with a concrete set of ideas of how to create their own AIS. The book then goes onto describing the natural immune system in context with other biological systems and explores interaction between those systems. This will allow the reader to develop an understanding and appreciation for the richness of biology and its possible inspiration. This is then followed by a discussion of the field of AIS in relation to other computational intelligence paradigms. It is hoped that this chapter will allow the reader to become familiar with other techniques and understand the relative strengths and weaknesses of each and where the use of each (including AIS) would be appropriate.
2. Fundamentals of the Immune System
3. A Framework for Engineering Artificial Immune Systems
4. A Survey of Artificial Immune Systems
5. The Immune System in Context with Other Biological Systems
6. AIS in Context with other Computational Intelligence Techniques
7. Case Studies
8. Conclusions and Future Trends
Appendix I: Glossary
of Biological Terminology
Appendix II: Pseudocodes for Immune Algorithms
Appendix III: Web Resources on AIS
Back to Top
Significant attention has been given to the extraction of metaphors and ideas from the nervous system for the construction of Artificial Neural Networks (ANNs) and of neo-Darwinian evolution for the creation of evolutionary algorithms (EA). More recently however, attention has been drawn to the use of the immune system as another powerful metaphor. The immune system is incredibly robust; it is adaptive, inherently distributed, posses powerful pattern recognition, learning and memory capabilities. It is for these reasons (and more) the immune system is attracting such attention.
The field of research that applies immune metaphors, abstractions, and principles to computational problems is known as Artificial Immune Systems (AIS). The field of AIS is, in computer science and engineering terms, relatively young and is growing at a rapid rate. To date, no single text has ever been written that attempts to consolidate all current research and to focus the many ideas into a general framework. It is primarily for this reason this book is being written. It is introductory and multidisciplinary in nature and will be suitable mainly for final year undergraduates of engineers and computer scientists, research students in various areas from bioinformatics to computer science and for both business people and academics alike, who want to be introduced to this exciting research field.
This first chapter will gently introduce the reader to the field of biologically inspired computation. Reasons why biology is very good at providing inspiration for addressing computational problems will be discussed, along with discussions on the nature of interdisciplinary research.
Full Text (PDF)
Back to Contents
Understanding of the human immune system has significantly advanced in the past few decades. The immune system is known to defend our bodies from attack by foreign invaders (such as viruses and bacteria). It has also been shown to be capable of remembering previous encounters with these invaders and this knowledge has been harnessed to provide vaccines for a whole plethora of viruses and diseases. The immune system also shows highly distributed detection and memory mechanisms, diversity of detection ability across individuals, approximate matching strategies, and sensitivity to most new foreign patterns.
This chapter begins with the basic concepts and a brief history and perspectives on immunology. This is then followed by a general overview of the immune defense mechanisms, including the primary and secondary immune responses. Attention is then given to the main types of cells that make up the immune system, B-cells and T-cells. There then follows an explanation of the theories relating to how the immune system is considered to learn and maintain a memory of past encounters. The fact that the immune system can distinguish self from nonself is then explored, followed by a final section on the immune network theory.
The architecture of the immune system is also discussed along with associated principles and processes. In particular, the clonal selection principle, affinity maturation and immune network theories will be stressed and detailed as important mechanisms to the development of artificial immune systems.
Back to Contents
Recently, there has been a growing interest in the use of immune metaphors and the construction of AIS applied to a wide variety of domains. Although some papers from the literature proposed formal artificial immune systems, no general framework has ever been introduced for the construction of AIS to be applied to several different domain areas. This is in part due to the sheer diversity of cells, molecules and organs that are involved in the immune system, to the young age of the research field, and also to the innumerable possibilities of applications. It is felt that this lack of a formal framework is a potential weakness of the research field and that by creating a general framework, designs and solutions will be more robust and principled, drawing from a more concrete set of ideas.
This chapter begins with an explanation as to why the immune system can be considered to be a good metaphor for solving computational problems. Theoretical immunology plays an important role in the development of AIS, but there is a difference between the two disciplines. Therefore, an important distinction is drawn out between the theoretical immunology (and simulations of the immune system) and the field of AIS. The chapter then goes onto introduce a general framework that can be used for the design of AIS. To achieve this framework, several ideas and proposals from the current research literature have been brought together in a unified framework that allows for the design of AIS. The framework has been defined as a specific set of ideas in such a way that it can be extended in a principled manner as and when required.
Back to Contents
There are currently many different algorithms to simulate and/or model the immune system that are aimed at solving problems in a wide range of domains. This chapter will provide the reader with an overview of the majority of these works. Attention is given to the types of shape-space used (encoding schemes), affinity (match) functions, and common immune metaphors employed. The framework proposed in Chapter 3 will be used throughout the review and through this be shown to be general for the explanation and subsequent development of new AIS.
The focus of this chapter will be on use of the immune metaphor in a variety of applications. However, discussions regarding the results from these applications and their implications will be briefly mentioned.
Back to Contents
Immune system research is, in its very nature, interdisciplinary. There are many researchers studying the relationship between the immune system and other biological systems, such as the nervous and endocrine systems. Some cognitive scientists are interested in studying the immune system recognition, learning and memory capabilities. Researchers on evolutionary biology are interested in verifying the contribution of the immune system to the evolution of the organism and comparing their timescales. Finally, researchers on ecological systems argue that immune systems share a number of similarities with ecological economic systems in terms of function.
In this chapter the nervous and endocrine systems are gently introduced and a discourse is provided about the many similarities, differences and relationships among them. By studying these biological systems and their integration, we expect to provide new insights for the development of novel hybrid systems and also systems that are closer to artificial life and behaviors than the present ones.
The working of the brain has led to the development of one of the most influential computational intelligence paradigms: artificial neural networks (ANN). ANNs have been applied to a vast amount of complex problems such as vision, pattern recognition, classification and approximation, to name a few. By regarding the immune system as cognitive, such as the brain, and making a comparison between them, it is possible to trace new parallels between AIS models and ANN and also to have new insights into how to create novel hybrids between them, as will be fully discussed in Chapter 6.
In this chapter we will also illustrate how an adaptive immune response can be described as an evolutionary process from a Darwinian viewpoint. By taking this approach, one can better understand the nature of immune system population diversity, genetic variation and natural selection. Thus, it will be possible to better understand and explain the behavior of several AIS presented in the literature.
Back to Contents
The area of biologically inspired computing is becoming increasingly large. Moving from the biological viewpoint presented in the previous chapter, it is essential to place the field of AIS in context and relation with other biologically inspired paradigms.
To this end, this chapter stresses the importance and usefulness of biological metaphors for developing new computational intelligence paradigms, such as artificial neural networks and evolutionary algorithms. The central nervous system and Darwinian evolution discussed in the previous chapter were crucial to the development of these approaches. By relating these systems to the development of AIS, a comparative study is presented showing the similarities and differences between the three approaches. The chapter also discusses how each of the separate research fields can enrich one another when a greater understanding of the separate fields is obtained. Additionally, artificial immune systems are also placed in context with other computational intelligence paradigms, such as case-based reasoning, classifier systems, fuzzy systems and DNA computation. This will allow the reader to fully appreciate the context of the work and would hopefully be able to assess where the use of AIS might be more or less appropriate than other techniques.
The chapter is concluded with a tentative list of extensions in the direction of hybridizing one or more of these approaches.
Back to Contents
To fully appreciate the usefulness of the AIS paradigm, a detailed analysis and discussion of some specific AIS will be examined.
Chapter 3 proposed a general framework for the design of AIS and Chapter 4 provided an extensive survey of the field. Chapter 4 focused on the immune metaphors employed and how the framework proposed in Chapter 3 could have been employed in the formalization of these AIS. The present chapter takes a slightly different approach. It will describe in detail, according to the framework and design principles introduced in Chapter 3, some influential works and applications of AIS from the literature. Among these, it will be stressed a computer network security application, an autonomous navigation system for robots, two discrete immune network models applied to data analysis and optimization tasks, and a scheduling problem.
Back to Contents
This book has introduced a large number of novel techniques. In order for the reader to fully benefit from the topics discussed, a comprehensive overview of the concepts introduced in this book will be provided. Attention will be drawn to the power of the new computational paradigm of AIS: in particular the extraction of useful metaphors, principles and abstractions from the immune system and the strength of the AIS framework.
To augment this view, a discussion is provided concerning the future direction of the field of AIS. Issues regarding problems with using the immune metaphor approach will also be discussed, along with arguments on how the authors view the future of AIS in the wider context.
Back to Contents
©2002, L.N De Castro and J. Timmis. (Thanks to Simon Garrett for layout ideas)