Fuzzy reasoning in decision making and optimization (Q5950268)

In many real-life problems, we have only uncertain expert information about some parameters of the situation, information presented by words of natural language rather than by numbers. A natural way to formalize such situations is to use a formalism that was specifically designed for translating such expert knowledge into precise mathematical terms -- formalism of fuzzy logic. There have been many successful applications of fuzzy techniques in real-life problems, but these techniques are not as widely used and not as successful as they can be; one of the main reasons for the reluctance to use these techniques is that at present, there are a lot of different fuzzy techniques based on different heuristic ideas: some ideas are used in optimization, other ideas are used in reasoning, etc. As a result, when we have a complex real-life problem in which we have to combine several such techniques, the resulting combination of drastically different ideas does not look very convincing and sometimes, does not lead to good results. To remedy this situation, the authors provide a new unified methodological approach to fuzzy techniques. Specifically, they go back to the very first fuzzy techniques -- techniques of fuzzy logic and fuzzy reasoning, and then show that fuzzy reasoning can serve as a foundation of decision and optimization as well. They compare the decision and optimization techniques coming from fuzzy reasoning with the previously proposed methods and show that their methods lead to similar results. Thus, if we replace the existing ``quilt'' of techniques with the author's more consistent methods, we get a more clear and more convincing tool and still retain all the previous successful applications of fuzzy optimization and fuzzy decision making. The authors do not simply duplicate the existing decision-making and optimization techniques in the new terms, they go beyond that in providing techniques for multi-criteria optimization that take into consideration inter-dependence between different criteria \(f_1,\ldots,f_n\). If we have \(n\) dependent vectors, then we can form an independent basis by forming appropriate linear combinations of these vectors. Similarly, the author propose to first find the independent combination \(f_i'(x)=F_i(f_1(x),\ldots,x_n(x))\) (possible non-linear), and then use the standard multi-criterion techniques to combine these new objectives \(f_i'(x)\). That the authors' methods really work is shown on the example of real-life applications in forest products industry. They conclude the book with a visionary description of future trends in fuzzy reasoning and decision making, trends related to the ever increasing use of interacting autonomous intelligent agents to solve real-life problems. The book starts with a nice, clear, and rigorous introduction to fuzzy logic, fuzzy arithmetic, and fuzzy reasoning. It can therefore be recommended both for researchers in fuzzy techniques who are interested in new methods and to researchers and students who may not be that familiar with fuzzy techniques; for such readers, the first part of this book provides a textbook-quality introduction.