Computational intelligence systems and applications. Neuro-fuzzy and fuzzy neural synergisms (Q5950775)

The objective of this research monograph is to present the main directions of synthesizing fuzzy sets, artificial neural networks and evolutionary computation methods for the design of Computational Intelligence (CI) systems. A comparative analysis of particular methodologies which are typical for intelligent systems shows a high degree of synergy of fuzzy systems and artificial neural networks. CI systems are used in applications in the field of data mining and decision support systems. The book is recommended to researchers and practitioners in the fields of Artificial Intelligence and Computational Intelligence, and for students of application-oriented Computer Science and special fields of Information Systems. The book consists of 10 chapters: Chapters 2, 3 and 4 present basic principles from fuzzy sets and fuzzy logic, artificial neural network and evolutionary computation methods. Ch. 5 shows relations between CI systems and traditional AI systems and characterizes the combinations of the three methodologies. Selection of the most promising direction of synthesizing these techniques is possible by determining attributes typical of CI systems. Ch. 6 presents the approach of neuro-fuzzy-genetic systems for synthesizing rule-based knowledge from data. A general concept of the proposed neuro-fuzzy system in learning mode, and a procedure in five phases is suggested. After the learning process the system provides a set of fuzzy rules representing the knowledge having been extracted from learning data. Ch. 7 presents the formulation and solution of the identification of complex dynamic systems with the use of the rule-based system of Ch. 6. Two applications dominate in this Chapter: A rule-based model from data descibing the dynamic system of an industrial gas furnace, and a neuro-fuzzy controller from data for a nonlinear control problem of a simulated backing up of a truck. Ch. 8 shows a neuro-fuzzy-genetic rule-based classifier which is applied to designing three decision support systems. The conceptual scheme, the implementation, and the inference mode of the classifier is presented. Ch. 9 shows a fuzzy neural network for system modelling and shows a numerical example for a fuzzy neural controller. Ch. 10 presents a special case of the classifier of Ch. 9, demonstrating its learning and inference modes in the framework of a diagnosis application.