Soft computing and intelligent systems design: theory, tools and applications, F. O. Karry and C. De Silva, Pearson, Addison‐Wesley, New York, NY, 2004
D. Subbaram Naidu · International Journal of Robust and Nonlinear Control · 2006
Intelligence is defined, according to Webster's Dictionary, as ‘the capacity to apprehend facts and propositions and their relations and to reason about them’. In terms of hierarchy, the intelligence occupies next to genius ‘person with very high intelligent quotient’ as shown in Figure 1. At the bottom of the ladder is the data which when formatted becomes information to be useful for any analysis. If one acquires a lot of information, he/she becomes a knowledgeable person. If knowledge is used with respect to facts and reason, it becomes intelligence. A highly intelligent person becomes a genius. The conventional artificial intelligence (AI) refers to mimicking human intelligent behaviour by expressing in language or symbolic forms 1–3. The earlier products of AI are expert systems or knowledge-based systems. With spectacular advances in computer hardware and software, and biological modelling, the modern AI, also called machine intelligence, computational intelligence, or soft computing (SC), comprises additionally methodologies based on neural networks (NNs), fuzzy logic (FL), evolutionary computation, probabilistic reasoning (PR), simulated annealing (SA), chaotic systems and so on. The SC, unlike the traditional hard computing aiming for precision and certainty, focuses on accommodating pervasive imprecision and uncertainty of the real world. The NNs provide learning, identification and adaptation 4; the FL deals with imprecision, approximate reasoning, and rule-based systems 5. Evolutionary computing (EC), also called evolutionary algorithms, is based on biological evolutionary processes of Darwian theory 6 that ‘allow populations of organisms to adapt to their surrounding environment, genetic inheritance and survival of the fittest’ 7. The EC 8 comprises of evolutionary programming (EP), genetic algorithms (GAs), genetic programming (GP), and classifier systems (CS). The GAs and SA aim at systemized random search and derivative-free optimization 9,10. The GP is used in the automatic production of computer programs 11. The CS, consisting of message and rule systems, credit assignment system and GA, are based on machine learning which deals with building computer programs that change depending upon the environment or experience 12. Often, GP and CS are considered as special cases of GA and not as special cases of EC 13. The EP is an approach to AI making use of finite states automata 14. The PR provides decision analysis and management of uncertainty. These various methodologies are synergetic and complementary rather than competitive and hence are frequently used in combination leading to some kind of hybrid systems. One such very useful and promising integrating technology emanates from ‘neuro-fuzzy’ or ‘fuzzy-neuro’ systems, although other forms such as fuzzy-genetic, neuro-genetic, or neurofuzzy-genetic are continuously evolving 15. Hierarchy of Data to Genius (© D. S. Naidu, 2005). There are specific books that deal with SC and its applications in general 16–19 and to specific fields such as electrical power systems and drives 20 and telecommunications 21. The present book by Karry and De Silva 22 joins the category of books that deal with NN, FL, and their combinations such as 23–26, whereas there are other books that deal with NN, FL, GA, SA and their combinations 7, 13, 15, 27–30. Part I: Fuzzy Logic and Fuzzy Control Chapter 1: Introduction to Intelligent Systems and Soft Computing Chapter 2: Fundamentals of Fuzzy Logic Systems Chapter 3: Fuzzy Logic Control Part II: Connections in Modeling and Neural Networks Chapter 4: Fundamentals of Neural Networks Chapter 5: Major Classes of Neural Networks Chapter 6: Dynamic Neural Networks and their Applications to Control and Chaos Prediction Chapter 7: Neuro-Fuzzy Systems Part III: Evolutionary and Soft Computing Chapter 8: Evolutionary Computing Part IV: Applications and Case Studies Chapter 9: Soft Computing for Soft Machine Design Chapter 10: Tools of Soft Computing in Real-World Applications A brief summary of these four parts of the book follows. Part I focusing on FL and FL control starts with an introduction to intelligent systems and SC and focuses on fundamentals of FL and its operations, fuzzification, defuzzification, fuzzy control architectures and robustness and stability related to fuzzy dynamical systems. The field of NNs is the focus of Part II discussing basics of NNs, classes of NN (such as radial basis function networks, Kohonen's networks, Hopfield network), recurrent NNs, applications to identification and control, and chaos. Also focused in this part is the important topic of neuro-fuzzy architectures. Under Part III, the authors discuss the topics under EC such as GAs, fusion of GAs with NNs and with FL. The important topics of applications and case studies is covered under the final Part IV. Some of the interesting applications include controller tuning, supervisory control of a fish processing machine, direct-current (DC) motor control, CDMA cellular system, and asynchronous transfer mode (ATM). The field of SC is rapidly changing and this book captures recent developments integrating theory, tools or techniques and applications. Another distinguishing feature of this book is the style of presentation, it is simple, concise and easy to read and understand, and the book is written ‘for the benefit of the student’. There is lot of thought that went into preparing the examples, end-of-chapter problems and references, exercises, and projects. Another useful feature of this book is the case studies and application spectrum discussed exclusively in separate chapters. Another regular feature that is incorporated now-a-days in most of the books is the integration of some kind of simulation software and this book used the academia-standard MATLAB®. The book provides an extensive bibliography at the end of each chapter for further study. Finally, this reviewer, having used the book by Jang et al. 28, is tempted to consider using the present book 22 accompanied by a Solutions Manual, for graduate course on Intelligent Control Systems, at Idaho State University. The book can be used without much background material for senior and/or graduate students interested in knowing fundamental aspects of the fields of NNs, and FL within the broader discipline of intelligent control systems. On the whole, the present book is a welcome addition to the developing field of SC.