Zgurovsky M.Z., Zaychenko Y.P. The Fundamentals of Computational Intelligence: System Approach

Springer, 2017. — 389 p.This monograph is dedicated to systematic presentation of main trends, approaches, and technologies of computational intelligence (CI). The introduction includes brief review of CI history, the authors’ interpretation of CI, the analysis of main CI components: technologies, models, methods, and applications. The interconnections among these components are considered and relations between CI and soft computing are indicated.
Significant attention in the book is paid to analysis of the first CI technology — neural networks. The classical neural network backpropagation (NN BP) is described, the main training algorithms are considered. The important class of neural networks — with radial basic functions is described and its properties are compared with NNBP. The class of neural networks with backfeed — Hopfield and Hamming are described and their properties and methods of weights adjustment are considered. The results of experimental investigations of these networks in the problem of images recognition under high level of noise are presented and compared. NN with self-organization by Kohonen are considered, its architecture and properties are described and various algorithms of self-organization are analyzed. The application of Kohonen neural networks in the problems of automatic classification and multidimensional visualization is considered.
Great attention in monograph is paid to novel important CI technology — fuzzy logic (FL) systems and fuzzy neural networks (FNN). The general description of fuzzy logic systems is provided, main stages of fuzzy inference process and fuzzy logic algorithms are described. The comparative analysis of fuzzy logic systems properties is presented, their advantages and drawbacks are analyzed. On this base the integration of two CI technologies — NN and FL was performed and as a result the new CI technology was created — fuzzy neural networks. Different FNN are described and their training algorithms are considered and compared.
New class of FNN — cascade neo-fuzzy neural networks (CNFNN) are considered, its architecture, properties, and training algorithms are analyzed. The applications of FNN to the forecast in economy and at stock markets are presented. The most efficient algorithms of fuzzy inference for the problem of forecasting in economy and financial sphere are determined.
The distinguishing features of this monograph are a great number of practical examples of CI technologies and methods and applications for solution of real problems in economy and financial sphere, in particular forecasting, classification, pattern recognition, portfolio optimization, bankruptcy risk prediction of corporations and banks under uncertainty which were developed by the authors and are published in the book for the first time. Just system analysis of presented experimental and practical results enables to estimate the efficiency of the presented methods and technologies of computational intelligence.
All CI methods and algorithms are considered from the general system approach and the system analysis of their properties, advantages, and drawbacks is performed that enables practicians to choose the most adequate method for their own problems solution.
The proposed monograph is oriented first of all to the persons who aspire to make acquaintance with possibilities of current computer intelligence technologies and methods and to implement them in practice. It may also serve as inquiry book on contemporary technologies and methods of CI, it will be useful for students of corresponding specialties.Neural Networks
Neural Networks with Feedback and Self-organization
Fuzzy Inference Systems and Fuzzy Neural Networks
Application of Fuzzy Logic Systems and Fuzzy Neural Networks in Forecasting Problems in Macroeconomy and Finance
Fuzzy Neural Networks in Classification Problems
Inductive Modeling Method (GMDH) in Problems of Intellectual Data Analysis and Forecasting
The Cluster Analysis in Intellectual Systems
Genetic Algorithms and Evolutionary Programing
Problem of Fuzzy Portfolio Optimization Under Uncertainty and Its Solution with Application of Computational Intelligence Methods