Handbook of Research on Artificial Immune Systems and Natural Computing: Applying Complex Adaptive Technologies

Natural computing (NC) or nature inspired computing (NIC) is such a kind of technology. Of course, there are series of theories and methods relating to it. As we know, NC was developed with artificial intelligence, because genetic algorithm, artificial neural network, fuzzy theories, which are classical methods of NC, were developed with AI. Now they are called biologically inspired computing (BIC), which is an important type of NC. The other kinds of NC are social inspired NC, physical inspired computing and chemical inspired computing. Most methods of NC belong to BIC. Besides the classical ones, there are so many kinds of BIC developed today, such as DNA Computing, Membrane Computing, Bacterial Computing, Ant Colony, Swarm Intelligence, and Artificial Immune System (AIS), which is also one of the main topics of this book. So, here we can see that AIS is also a part of NC. That is why we can discuss NC and AIS in one book.

Artificial immune systems are adaptive systems inspired by theoretical immunology and observed immune functions, principles, and models, which are applied to complex problem domains; they are relatively recent examples of using the real world as computational inspiration. Immune system, particularly a human being’s immune system, is a very complex system, so it has many special characteristics and physiology processes which the other biology systems do not own, such as recognizing virus by pattern, keeping memory by immune cells, learning by responding to invaders, clone selection, negative selection, and so on.

In the past ten years, researchers from different fields have investigated immune system from different angles for different aims. In the field of computer science, we mainly focus on the computational ability of it.

In addition to obvious security metaphor, many other application areas of AIS have been developed, including anomaly detection, optimization, fault tolerance, machine learning, data mining, automation control, Internet, and so on. Although AIS do not break all the classic computational paradigms, for example, they do not use concepts from quantum physics; they do challenge some of the major paradigms. That is why they are worthy to be researched.

As a whole, AIS is one of a whole array of novel approaches for computation. In fact, these separate areas are not independent. Rather, their results and insights should provide valuable groundwork for the overarching challenge. Thus they are to produce a fully mature science of all forms of computation unifying the classical and non-classical computing paradigms.

The Overall Objective of the Book
First, it is a good reference book for researchers in the field of NC because it provides latest empirical researching achievements for them. They can learn the main developing tendency, ideas, methods, and technologies of AIS.

Second, it will make the research of the field be more active and absorb much more attention. Because AIS is a relative new research direction in the field of NC, this book will enhance the position of AIS in the field of NC or biologically inspired computing (BIC). It will enhance many more applications of AIS in the near future. It makes researchers learn that in the field of AIS, immune system should not only be used as metaphor for solving computer security or designing optimization algorithms based on clonal selection theory and immune network theory from immunology. It has many potential abilities to be developed, such as its detecting ability and distinct ways of information processing relative to brain. And the applications of AIS could extend to many broad fields besides security, optimization, such as Internet, control, forecast, and so on. Many more theories should be paid attention to for their long-term developing.

Third, as comparison, people can also learn more about the other paradigms of NC in the same book because immune system is only one kind of inspiration source. Researchers can understand the main differences among those different inspiration sources and where they are adaptive to respectively. It is written for professionals who want to improve their understanding of the strategic role of NC in different fields. What is more important, it can be seen that the inspiration resources of NC are so diverse, that is why NC has had such a strong living ability for so many years. Based on these, it can inspire more researchers to develop advanced technologies or theories to solve practice problems. This is the main objective of the handbook.

The Target Audience
The contents of this book will be attractive for graduate students, doctoral candidates, lecturers, academic or professionals, researchers and industry parties in biology, computing, engineering, computer science, artificial intelligence, information technology, especially those who are interested in developing and applying alternative problem solving techniques inspired by immune system, and also the other ways inspired by biology or nature.

Introduction
The book is mainly divided into two parts: The first one is mainly on AIS. Some new ideas, methods are proposed by using metaphor or mechanisms of immune system to solve problems of engineering and science. After reading this part, the readers can know recent applications of AIS in different fields, so they can open their mind much more broadly. There are 15 chapters in this part.

The first 3 chapters focus on the application of optimization of AIS.

Chapter I is a review of multi-objective optimization (MOP) with AIS (MOAIS). This chapter provides a thorough review of the literature on MOAIS algorithms based on the emulation of the immune system for the readers who are interested in this topic. MOP had been given much more attention in recent years. Many different methods had been designed to solve many types of problems of MOP, such as evolutionary algorithm, neural networks and so on. The algorithms of MOAIS were also used for MOP problems recently and showed their strengths or weaknesses in solving these problems. There are roughly 17 algorithms of MOAIS reviewed in this chapter. Most of them are from the papers before 2007.

The multi-objective immune system algorithm (MISA) is considered the real first proposal of MOAIS. After it, many different algorithms of MOAIS were developed, most of which are based on the two main theories of immunology: clonal selection theory and immune network theory.

In this review, we can learn that MOAIS has four features, including elitism, adaptive population, clone size, and memory. Algorithms of MOAIS are designed distinctly based on these common features to adapt to their concrete tasks. They had been applied to power systems, electromagnetism, computer science, biomedicine, control theory, networking, and so on. Although some comparison work proved that neither MOAIS performs better nor is it inferior to the other kinds of MOP, for any class of problems, and such results sounds to be frustration to our researchers, just as the author said, these considerations opens the research trends to study if there is a class of problems which is well suited to be solved by immune inspired algorithms. Chapter 2 tells us a new algorithm of MOAIS and its engineering applications in practice. A population adaptive based immune algorithm (PAIA) and its modified version for MOP are put forward and tested in this chapter. In this chapter, at first, it reviews previous research work about single and multi-objective optimization of the existing immune algorithms. Then, the drawbacks of immune algorithms are analyzed, and also those of the other evolutionary algorithms and the modified PAIA are compared with many important evolutionary algorithms on ZDT and DTLZ test suites. Based on the comparison research work, it presents a general framework from the modified PAIA.

In fact, the main contribution of the new algorithm is that it adapts a multi-stage optimization procedure, which reduces the number of evaluation significantly. The algorithm has its own strengths, including adaptive population, good jumping-off point, good and fast convergence, which make it compete with EA. And what is more important, the proposed algorithm is used to solve a real engineering problem. The results show that it is efficient in real application. So far, since the real applications of MOAIS are few, we are pleased to see in the chapter that it proves that the algorithm of MOAIS can be useful in some real problems.

Chapter 3 tells us a new kind of optimization algorithm called population-based artificial immune dynamical system(PAIS). But it is not for MOP. It is mainly for solving the problems of numerical optimization(NO). Some general descriptions of reaction rules including the set of clonal selection rules and the set of immune memory rules are introduced in PAIS. Thus, a general model of AIS, which models the dynamic process of human immune response and is particularly used for NO, is established. Based on the model, a dynamic algorithm, termed as PAISA, is designed for solving NO problems. The performance of PAISA is tested on nine benchmark functions with 20 to 10000 dimensions and a practical optimization problem, optimal approximation of linear systems, successively. The experiment results indicate that PAISA is good at optimizing some benchmark functions and practical optimization problems, especially large search space problems.

The main contribution of this chapter lies in that it proposes a general model combing the main mechanisms of immune system and then designs dynamic algorithm to solve problem. Based on the model or a platform, many different algorithms for NO can be proposed. And its validation experiments are very sound relative to the other similar researches. This chapter also proves that AIS can solve practice problem of optimization, and embodies the strength of immune-inspired dynamic algorithm.

So till now, we can see that algorithms of AIS can be used to solve optimization problems, though there are some disadvantages waiting for being improved, at least, it is proved by so much hard work that immune optimization algorithms has become an important and competitive branch in the field of NC. But it is clear that the applications of AIS are not limited to optimization since there are so many problems from the other fields waiting for being solved. So, let us see another important application aspect of AIS, that is, Agent, which is a well known field in computer science. The following three chapters are mainly on AIS and Agent.

In chapter 4, the authors tell us an interesting game solving plan by using immune inspired algorithm. Basically, it is interesting because a classical problem of multi-agent encounters-pursuit-evasion problems is researched, and the immune system is considered as a multi-agent system. Then, some parallel relation can be established between the two different systems. There are two kinds of pursuit-evasion problems—two player and multi-player. The immune approach is used for the both. In the problem, the interactions between a pursuer and an evader are modeled by those between an antigen and an antibody. And the interactions between antibodies help choose more and the most suitable pursuer. The well known clonal selection algorithm(CSA) is extended to Multi Agent Immune System Algorithm(MISA) for solving the problem. It is scalability and can be on-line learning, resistance to errors.

The simulation results show that MISA outperforms the other solution both in a number of training epochs and the number of steps, necessary for the goal achievement.

The main contribution of the chapter is that it uses the paradigm of reinforcement learning method in the proposed immune algorithm. And it can solve the complex problem combining with game theory. Although there is no sound theoretic explanation of the approach, just as the authors say: it seems that bridging a gap between traditional, mathematically motivated models and novel, unconventional paradigms can be very fruitful for both parties, producing a kind of stimulating feedback.

Chapter 5 is a detail review about artificial immune systems and agents. In fact, AIS had been used in the field of Agents for many years (since middle 90’s in last century). Most of the work focused on robotic applications. The chapter briefly introduces the background of AIS applied in robotics from the angle of view of single and multi-agent. Then, some single and multi-agent systems based on the principles of immunology and applied in robotics are introduced in detail, including artificial immune network based cooperative control in collective autonomous mobile robots, AIS based multilayered cognitive model for autonomous navigation, AIS based cooperative strategies for robot soccer competition, AIS for a multi-agent robotics system, AIS based intelligent multi-agent model and its applications to a mine detection problem, idiotypic immune network for mobile robot control, a gait acquisition of a 6-legged robot using immune networks, and the application of a dendritic cell algorithm to a robotic classifier. At last, two projects implementing immune inspired multi-agent systems are presented.

We can see that most of the work is usually based on the adaptive immune response characteristics, especially the idiotypic immune networks or immune network theory, the innate immune response has been neglected and there is not much work where innate metaphors are used as inspiration source to develop multi-agent robotic systems.

The contribution of the chapter is that it shows a relative full picture of AIS applied in the field of Agent. And the immune network theory plays an important role in these applications. It is mainly because immune system has basic features of agent. Immune cells are active agents. It is relative easy to make a connection between immune system and agents. But what is disappointed is that there is no general theory about immune agent proposed till now.

Now we turn to another topic to which AIS is applied. As we know, scheduling is a kind of general problem in engineering.Chapter 6 proposes an immune algorithm based robust scheduling method to solve uncertain scheduling problem. The model of workflow simulation scheduling is used to model uncertain scheduling problem. In fact, it is a complex optimization problem to find an optimal robust scheduling scheme. A workflow simulation can be executed by using a workflow model. A variable neighborhood immune algorithm(VNIA) is proposed to find the optimal robust scheduling scheme. The VNIA is based on the immune network theory and uses variable neighborhood strategy to balance the conflict of local and global searches. A two-level immune network is designed to maintain a large diversity of populations during the optimization process.

It is compared with definitive scheduling schemes on 4 problems.. Experimental results show that the method is very effective for generating a robust scheduling scheme and has satisfactory performances under uncertain scheduling environments.

The contribution of the chapter is that it succeeds in solving uncertain scheduling problem by immune inspired method.

In chapter 7, another important scheduling problem--the management of complex small scale cogeneration systems is solved by immune algorithm.

The energy management problem is called the Combined Heat and Power(CHP). Such a problem is to minimize the management costs of power system and fulfill all loads requirements. It is defined as a scheduling period (e.g. one day, one week etc.) with loads, costs, fares changing.

In the chapter, after describing the classical formulation of energy management problem in complex energy nodes, it uses the classical clonal selection algorithm to solve the problem. The problem is coded by binary. And special immune mutation operator is designed to generate feasible solutions. The modified immune algorithm is called artificial immune system-linear programming (AIS-LP). It is compared with mixed integer linear programming (MILP) on a simple but effective energy management problem. The result shows that AIS can efficiently solve such problem.

The contribution of the chapter lies in modifying existing immune algorithm to be adaptive to new problem and get relative good result.

These two chapters let us see that immune inspired algorithms, either new developed ones or existing ones, can be used to solve the complex scheduling problems. Of course, all of them must be specific to problem being solved.

In the following two chapters, we learn something about AIS used for data mining or information retrieval.

Document clustering is a common problem in the field of data mining. In chapter 8, it uses the modified aiNet to solve the problem of document clustering. In the background, it introduces a method named document map and a research project BEATCA, which is a fully-fledged search engine capable of representing on-line replies to queries in graphical form on a document map.

After that, it introduces how to use aiNet to realize document clustering. The initialization of immune network is the first important step, which starts from random document drawn. The other improvements include robust constructing antibodies by new mutation approach, defining time-dependent parameters to stabilize the size of memory M, and global research replaced by modified local research to find robust antibodies. For the performance validation, it focused more on the structural properties of the immune network obtained and the impact of initialization strategies on document clustering. And some meaningful results are observed.

The contribution of the chapter is that it proposes some effective strategies to modify aiNet to be adaptive to the problem of document clustering, and it is combined with many technologies, such as document map, visualization, to solve the problem. And the modified aiNet for document clustering is computational savings as well as the resulting slow evolution of clusters compared to the alternative “from scratch”.

In chapter 9, an important topic of pattern recognition-feature selection, which is also an important step in data mining, is discussed. Clonal selection algorithm, which is an important and representative algorithm, is used to solve this problem. In order to solve feature selection problem, an antibody represents a given feature subset and a binary bit denoted the presence or absence of the feature at the corresponding location. The affinity function takes both the separability and the dimension of feature subset into consideration. New clonal operators, including clonal proliferation, affinity maturation, clonal selection are implemented. The algorithm is tested by three different ways. The first one is datasets from UCI repository. The second one is brodatz textures classification, the last one is SAR image classification. All of these test results prove the good performance in solving the problem of feature selection.

The contribution of the chapter lies in that the clonal selection algorithm is used for feature selection and tested on real problem. It testifies that clonal selection algorithm, with the characteristics of local optimization and global searching, is a good choice for finding an optimal feature subset for classifier. Of course, this algorithm had been developed into many different versions and used for different tasks.

As we have seen, most of the algorithms of AIS solving engineering problems in the above chapters are based on a few immune algorithm developed earlier, such as aiNet and clonal selection algorithm. But AIS should be limited here. In the following four chapters, we can some different immune inspired methods to solve more broad problems.

In Chapter 10, the researchers propose immune based bio-network architecture. It is inspired by the resemble features between the immune systems and future Internet, and uses some key principles and mechanisms of the immune systems to design a bio-network architecture to address the challenges of future Internet. In such architecture, network resources are represented by various bio-entities, while complex services and application can be emerged from the interactions among bio-entities. The authors designs a bio-network simulation platform with the capability of service emergence, evolution etc. It can be used to simulate some complex services and applications for Internet or distributed network. Two immune network computation models, mutual-coupled immune network model and immune symmetrical network model, are developed to generate the emergent services through computer simulation experiments on the platform. They are used to simulate web service simulator and peer to peer service model respectively. The feasibility of the two models is demonstrated through computer simulation experiments on the platform.

The main contribution of the chapter lies in that the researchers design two new immune network models inspired by immunology theories and use them to address the challenges from Internet. It let us believe in that classical immune network theory is not the only choice for inspiring or solving engineering problems. There should be more new ideas inspired by immune system.

Chapter 11 is more interesting than the chapters above because of the problem in it. It is not solved by existing immune algorithms. But a new algorithm based on mathematical models of immuncomputing. In the chapter, the key model is the formal immune network (FIN) including apoptosis (programmed cell death) and immunization both controlled by cytokines (messenger proteins). It is used for spatio-temporal forecast (STF). For the task, FIN can be formed from raw signal using discrete tree transform (DTT), singular value decomposition (SVD). The performance of the model is demonstrated on real application, that is, data of space monitoring of the Caspian, Black, and Barents Sea. And it gets surprising results by using this model.

The contribution of the chapter is great. One hand, the work in the chapter makes AIS application extend to the field of forecast in practice. On the other hand, what is more important, the work is based on a relative complete mathematic theory of AIS, named immunocomputing, which was proposed by the author. The core of theory is the model of FIN. Although it is not new to researchers in the field of AIS, it is deserved to be paid attention to. As we know, one of the disadvantages of AIS is that it has no general theory itself. Immunocomputing brings promise to us.

In chapter 12, the mechanisms of immune system are used for controller, which is important component in control system. It is not surprised that immune based controller can be developed out because similar ideas and technologies inspired by biology exist for a long time, for example neural network.

In the chapter, three kinds of artificial immune controller are researched. The first one is based on the principle of T cells and B cells immunity. The second one is based on a simple double-cell immune dynamics model. And the third one is based on varela immune network model, which is a famous model in the field of theory immunology.

All of them are verified by simulations. At last, the authors propose a general structure of artificial immune controller based on the comparison results of the three controllers.

AIS had been used in field of automation control for a long time. But the earlier researches mainly focused on using immune algorithm to optimize classical PID controller. So the contribution of the chapter lies in that the mechanisms and theories of immunology are used to the design new type of controllers. Not only does the research bring fresh thought to the field of control theory, but also broad mind and new research content to the field of AIS.

In chapter 13, a kind of immune algorithm--immune programming is applied in image segmentation. It combines immune mechanism with that of evolution to form a global optimization algorithm. Two types of IP algorithms, the immune image threshold segmentation algorithm based on the maximum information entropy theory and the immune image segmentation algorithm based on the expense function minimization, are proposed in the chapter. All of them are tested on image segmentation. And the results show that IP can retain the image details perfectly.

The main contribution of the chapter is that it proposes two types of IP algorithm to solve the problem of image processing. It proves that IP can be used as a new tool for image processing.

The research content of chapter 14 is different from that of any chapter above. All of the researches above are realized by computer programming or simulation. They can be seen as software artificial immune systems in general meaning. While in this chapter, a kind of hardware immune system for MC8051 IP Core is developed out.

A framework of a hardware immune system is designed and used for the error detection of MC8051 IP core.

The principle of negative selection, which is an important mechanism keeping health of body, is realized in the design. The normal state of MC8051 is collected as self set and detectors are generated by negative selection algorithm. After this phase, the error of the system can be detected.

The system designed in this chapter is tested on an FPGA development board and the detail hardware system including circuits is shown. The experiment results are given as waveforms of the implemented circuit.

Although a few kinds of hardware immune systems are designed in recent years, more attention should paid to them. Immunotronics had been an important concept and research topic in the field of AIS. The contribution of this chapter is that the hardware immune system is used detecting errors of hardware system in practice. It is still a system with simple functions, but it will be improved in further and play more important role in similar task.

As we can see, the researches in chapters above are mainly for real engineering applications, while chapter 15 proposes a general platform of AIS. A Static Web Immune System(SWIS) is presented in it. It can also be looked as an important application of artificial immune system. According to the chapter, on the static web system, a normal model is proposed with the space property and the time property of each component, in order to identify the normal state of the system that the artificial immune system protects. There are three tiers, including the innate immune tier, the adaptive immune tier and the parallel immune tier, which are inspired by immune system. Such a system can be used as a general tool or platform to execute the task of detecting abnormal states of systems. It is useful for both theories and applications.

The contribution of the chapter is that it presents a general platform inspired by immune system and can be used for theories and applications researches. Such work can promote the development of AIS.

AIS is only one of a whole array of NC available. The second part is mainly on the other methods of natural computing.

Chapter 16 focuses on the mathematical foundation of evolutionary computation based on an evolutionary automata approach. Three classes of evolutionary automata, including evolutionary finite automata, evolutionary Turing machines and evolutionary inductive Turing machines, are studied. And they are proved to have universal automata—evolutionary automata(EA). An important property of evolutionary automata is their expressiveness. Based on the models, universal evolutionary algorithms and automata are proposed. All four basic kinds of evolutionary computation, including GA,ES, EP and GP, can be modeled by it.

The contribution of the chapter is that it proposes a new theory of evolutionary computation with sound mathematics foundation, which is meaningful to this field. And it also provides means to study many important evolutionary algorithms because the EA approach is very general and allows one to capture all currently known areas of evolutionary computation. The potential ability of this approach is so big that it will allow one to accommodate future areas of evolutionary computation as well, including algorithms of AIS.

Optimal motion planning is critical for the successful operation of an autonomous mobile robot. In chapter 17, it proposes a concrete application of evolutionary algorithm (EA)--genetic algorithm-- for this problem. It is an evolutionary path planning for robot navigation under varying terrain conditions. It has the advantages of dealing with diverse terrain conditions when determining the optimal path. The key point of the method is that it combines GA with fuzzy logic. The later is used to model terrain conditions. The simulation results show that this method is robust and makes the robot adapt to environment where conditions are dynamic.

The contribution of the chapter is that the proposed genetic algorithm can make the autonomous objects deal with dynamic environment robustly. And in fact, it is a kind of hybrid algorithm for using fuzzy linguistic variables.

The two chapters discussed above are mainly on evolutionary algorithm. The following chapters turns to another important method of BIC-Ant Colony Optimization(ACO).

In chapter 18, ACO is used for content based image retrieval. Similar to chapter 17, it is also a hybrid method, which combines ACO with fuzzy logic. It has two stages to deal with problem. In this first stage of the algorithm, a comparison is performed using all the bins of the three features, that is, the whole population of the ants, and an initial ranking of the images takes place. The second stage is initiated after classifying image database. The experiment results prove that it is a more efficient means compared to popular and contemporary methods. In order to speed up the process, hardware implementation analysis of the whole system is designed and analyzed. The increase in speed is phenomenal.

The main contribution of the chapter is that it deals with image processing by ACO and fuzzy logic and get better results. What is more important and meaningful, hardware system is designed for speeding up processing. Similar work is few to see in the applications of NC.

In chapter 19, Ant Colony is applied in the field of data mining. It focuses on the problem of medical data clustering. After introducing some basic concepts of data mining and ant colony, it presents a clustering method based on ant colony for processing biomedical data- electrocardiogram. The method is called ACO DTree, because ACO works with classifier trees—decision tree(DT) like structure. Creation of the trees is driven by a pheromone matrix, which uses the ACO paradigm, while DT is evolved by PSO and the fittest one is selected as classifier. The data are divided into subgroups having similar properties and the classes are different. It thus finishes the task of clustering data. This method is compared with the other nature inspired methods and some classical methods. It is comparable to them.

The main contribution of the chapter is that it proposes a hybrid method of NC to solve clustering problem. This chapter and the chapters above inspire us that single method of NC may not be enough to deal with some complex problem. Hybrid one may be a good choice according to different problems.

In chapter 20, a well known and interesting NC method is introduced and used for optimization. It is a hybrid artificial life optimization algorithm (ALRT) based on emergent colonization. In this method, the emergent colony is a fundamental mechanism for searching the optimum solution. It can be formed by the metabolism, movement and reproduction among artificial organisms appearing at the optimum locations in the artificial world. In this case, the optimum locations mean the optimum solutions of the optimized problem. Thus, the ALRT can search the location of emergent colonies and achieve more accurate global optimum. The experiment results are compared with those of genetic algorithm and successive quadratic programming and show the optimizing ability of ALRT.

The main contribution of this chapter is that it uses hybrid AL method to solve optimization problem. Its idea is fresh because AL is known to be an important research field of NC, not a optimizing method.

If it is not enough, the ideas of following two chapters give us more surprises.

In chapter 21, a new method of optimization—so called Social Impact Theory based Optimizer(SITO)is presented. It is a kind of optimizer based on the social impact theory.

The main source of inspiration for SITO comes from the models of social interactions. The individuals or agents participating social activity forms an artificial society. The individuals are scattered over a grid defining their environment. Each individual changes their properties based on some rules involving the individual’s neighborhood. Each individual is represented by four parameters: the individual's attitude, two indicators of strength (persuasiveness and supportiveness) and the individual's location in the social structure. It is used to binary optimization problems. The experiment results show that the new optimizing technology is valid.

The main contribution of the chapter is that it presents a social-psychology-inspired, stochastic and population based method, SITO algorithm, which can be used for optimization of generally non-linear, multimodal and analytically, unoptimizable binary functions. It brings fresh technology for the field of optimization and also the field of NC.

In chapter 22, a machine learning technology- ethology-based approximate adaptive learning is proposed. Ethology-based form of machine learning is presented by the author before. Fundamental to the proposed approach to adaptive learning is the notion of perception. Near sets is a theory which can be used to solve the problem of ethology-based machine learning. It can be viewed as an extension of rough set. A near set approach approximating the sets of behaviours of organisms forms the basis for approximating adaptive learning. The principles of near set and adaptive learning are given in detail in the chapter.

Adaptive learning algorithms are used in the simulation of the behaviour of a single species of fish. The results show that short-term memory model provides an efficacious basis of for adaptive learning.

The contribution of this chapter is that a complete framework for an ethology-based study of approximate adaptive learning is proposed. And it acquires some interesting results from the simulation.

Parallel computing becomes more and more important for science and engineer because the problems faced by people become more and more complex. Nature has prepared such paradigms for us. In chapter 23, Nature Inspired Parallel Computing (NIPC) is proposed. The parallelism mainly exists in five dimensions including horizontal, vertical, application, time, and user. The four dimensions can be organized into different decomposition and communication schemes of NIPC.

Four schemes of NIPC are introduced in detail. Based on these schemes, the architecture of NIPC is constructed and composed of four layers, corresponding to the four schemes. It is also tested on three examples.

NIPC is tested on three examples, such as detecting illegal buildings in a city and so on The results show that it can speed up the generation and simulation of digital city.

The contribution of the chapter is that the working way of NIPC regards parallel computing just as a simulation of the parallel way in nature based on computing technology and electronic information technology. Such a new nature inspired parallel computing can promote the efficiency of parallel programming and parallel applications.

The last chapter is also a hybrid NC method-fuzzy chaotic neural networks, which integrates fuzzy logic, chaotic, neural network in one model. It tries to mimic the function of brain to design new kind of neural network. Four types of fuzzy chaotic neural networks are introduced, including chaotic recurrent fuzzy neural networks, cooperation fuzzy chaotic neural networks, fuzzy number chaotic neural networks and self-evolution fuzzy chaotic neural networks. Analysis to them shows that fuzzy number chaotic neuron not only has fuzzy number but also chaos characteristic. Such neurons can be used to construct fuzzy number chaotic neural networks. At last, a self-evolution fuzzy chaotic neural network is proposed according to the principle of self-evolution network. It unifies the fuzzy Hopfield neural network constitution method. The new model can work under two kind of active statuses and complete different function.

The main contribution of the chapter is that it establishes a series of new neural network models, which have some new characteristics and may be used to tackle more complex problem.

So far, we can see that AIS has some characteristics which are different from the other methods, for example, it has many inspiration sources, like the mechanisms of it, and theories from immunology. All of them can inspire researchers to develop new algorithms for different tasks. While the inspiration source of the other NC methods is single. And readers can compare the backgrounds and performances of different NC methods through the excellent researching work in the book, especially the applications of AIS and other methods, for example, in the .field of optimizations. In the aspect of applications, AIS can be used for many different fields, as we have seen in the chapters, but not limited to them, while most of the other NC methods can only be used in optimization, except neural networks. Note that the book doesn’t cover all kinds of NC methods, such as DNA computing, membrane computing.

So, we hope the book brings some new and fresh ideas to the readers.

For the editing of the book, I must acknowledge the following people:

All the authors join in the book, sorry for not listing them one by one here. You can see them in each chapter. And especially those who I have never seen before, most of them are from outside of China. All of them have a warm heart, serious attitude and are patient to join the hard work for one and a half year.

And all the authors from China, thanks for their supporting very much!

I must thank all IGI Press staff for their great work in publishing the book, particular thanks go to Rebecca.

And I must say thank to my teacher, Prof. Yao Sun, who always supports me in so many years since I become a teacher and always encourage me to challenge myself.

And also Prof, Lipo Wang, Prof Licheng Jiao help me so much in editing the book.

At last, thank to my parents, my wife and my lovely daughter!

Hongwei Mo