Introduction to the fuzzy expert system flops and state of the art fuzzy expert system theory

"Fuzzy Expert Systems and Fuzzy Reasoning" is a beautiful and surprisingly small book (effectively 274 pages, the long appendix consists mostly of the flops sample programs). You will find an introduction to rule based systems, to fuzzy logic and at the core of this book to inference in fuzzy expert systems.


Inference in fuzzy expert systems is quite different to the main application of fuzzy logic: fuzzy control systems. Of importance is the correct generalisation of the classic and operator. The authors claim here a one to one relationship to the correlation of the involved terms. For total correlation use the minimum of the involved terms (classical way of Zadeh), for uncorrelated entries multiply the involved terms (like in probability theory) and for total anti correlation use the bounded difference operator (P and Q = max(0, P + Q - 1). Siler and Buckley give a whole family of and operators for any given correlation of P and Q. This is state of the art and the authors state also controversial in the field.


Finally the reader gets a thorough introduction into the fuzzy expert system shell flops a fuzzy logic version of ops5 with an ample amount of sample programs. Especially interesting were the classification examples and the thorough discussion of fuzzification and for fuzzified numbers the defuzzification. For expert systems there is a longer lifecycle and more care required than for the control system applications.


The book is written in a clear and pleasant style. With a little background in the covered topics you probably will find most of the text easy to follow in spite of its briefness. You can download an educational version of flops from the publishers ftp site. I enjoyed reading "Fuzzy Expert Systems and Fuzzy Reasoning".

Other Reviews

• Introduction to the fuzzy expert system flops and state of the art fuzzy expert system theory