You are on page 1of 1

1734 PROCEEDINGS

IEEE,OF THE VOL.


12, 6 5 , NO. DECEMBER 1977

Appendiv 1 is a glossary of terms. Appendix2 is abibliography, functions, especially when associatively, the existence of the identity
with more than 300 references, divided into 17 categories. Appendix element, and/or of a unique inverse transforms the internal law into a
3concerns the equivalence of simple and modular sequences, and fuzzymonoid or fuzzygroup. Thechapter concludes by showing
Appendix 4 iscalled “Multiplication of Direct SequenceSignals” how the convolutionofmembership functions ofcertainpdftypes
The author is certainly to be commended for his efforts. Unfortu- generates a fuzzy monoid isomorphic to the natural numbers.
nately,thebookhas several deficiencies which greatly diminish its Chapter V is divided into two main parts. First, the concept of fuzzy
value. The material was apparently assembled in great haste and with sets is extended byallowing the setinto which the membership function
little attention to detail, as evidenced by the abnormally large number takes value to have structures ranging from preorder, order, semilattice,
of typographical errors. Many topics are covered in only a very brief and lattices of various types to ring. Numerous examples are presented
and shallow fashion.This reviewer foundthe author’s writing style to show how the structure of the latter set and the properties of its
very confusing in many places, withunconventionalmathematical internal laws are transferred to the structure of the fuzzypower set
representationsgreatlydetracting fromthe work. For example, on and its induced internal laws The concepts of Hamming distance and
page 16 a square-wave is represented as cos Wm t ; on page 170 a square- of various operations (inclusion, complement, union, intersection) are
wave is represented as (wm)= (4/n) (cos wmt - [ 5 11 cos 3wmt + . . . ). generalized when the set (into which the membership functions are
The author apparently has a great deal of trouble with trigonometric mapped) is alattice, using, respectively, the Hasse diagram andthe
identities,withmanyincorrect statements given concerningthem. order induced by the lattice.
Perhaps more significant are the many errors concerning probability In the second part,theconcept ofcategory is reviewed and then
results, e.g., pages 32, 148, 204. Basic termsandconcepts such as made fuzzy by mapping the morphisms defined by the category into a
convolution,correlation, and covariance are often misused, further target lattice, thus generating a fuzzy power set from the category into
confusing many readers. Other examples could be given of the author’s the lattice. Two types ofcomposition of fuzzymorphisms are then
imprecise use of language and mathematics, which this reviewer found presented.
very disturbing. Emphasis is also given to link distances, to membership functions of
Because the need is so clear, it is disappointing that a whole-hearted similitude and dissimilitude relations, and to membership functions of
recommendation of this book as a textbook for the classroom or as a two types of the transitive closures (min-max and min-sum) of ressem-
reference book for thepracticing engineer cannot be given. blance relations Decomposition of fuzzy order or similitude relations
is also considered. Thechapter concludes by presenting classes of
Reprinted from IEEE Communications Society Magazine, May 1977.
membership functions for labels such as small and large on the universe
sets N,Z , R + , and R .
Introduction to theTheory of Fuzy Subsets-Vd 1: Fundamental
Theoretical Elements-A. Kaufmann (New York: Academic, 1975, Chapter 111 mainly considers fuzzy variables and functions of fuzzy
432 pp.). Reviewedby A. P. Bonaert,Metra-Sobemap, Brussels, variables. Fuzzy variables aremembership functions treated as varia-
Belgium. bles. Among the class of functions of fuzzy variables, analytical func-
tions (Le., polynomials of fuzzy variables and their complements with
This book is the English translation of the French original published binary operators min and max) are singled out since they themselves
three years ago and constitutes the fust volume of a collection of five. can become new fuzzy variables.
The fust volume to be reviewed surveys mathematical essentials (es- Concepts of Boolean polynomialsare adaptedandextended to
pecially from an algebraic and structural point of view) of fuzzy sets. analyticalfuzzy functions such as the transformations of analytical
The remaining four volumes (of which two are already published in functions into functions with one binary operator (Sheffer, Peirce) or
French) are devoted to applications. in uniquecanonical expressions: minof max termsor max of min
Chapter I is an introduction to a number of concepts. Fuzzy subsets terms Series-parallel networks connecting elements representing fuzzy
(the prefiv “sub” is henceforth omitted) are defined as a mapping of variables are used to derive the canonicalexpressions of analytical
the elements of a universe set into the closed interval [0, I ] . This functionsor to obtainthe min-max expression fromthe max-min
mapping, called the membership function, is a generalization of the expression.
characteristic
function. The operations/concepts of complement, The Marinos algorithm is used to write systematically the conditions
inclusion, union,intersection, algebraic productand sumas well as under which acanonical form w i be a giveninterval.
l The concept
Hamming and Euclideandistances are extended to fuzzy sets. The of truth tables is generalized to one of value tables. Two methods (one
concepts of the nonfuzzy setnearest to a given fuzzy set and nonfuzzy analytical, the other using value tables) are presented for determining
set of given level are introduced. It is shown how a fuzzy set can be the intervals into which an analytical function will lie when its fuzzy
decomposed using nonfuzzy sets of various levels. Then the concept variables are restricted to be in given intervals. A dual problem
of power set is extended to fuzzy power sets (Le., to all the mappings considered is, given a target interval and a set of fuzzy variables lying
of the elements of the universe into [ 0, 11). in known intervals, whatconditions arerequired for a given fuzzy
Chapter I1 starts by studying the mappings of the elements of the analytical function to map the fuzzy variables from their domains into
Cartesian products of sets into [0, 11, defining in this fashion fuzzy the target interval? The conditions are synthesized through multipliers,
graphs, fuzzy mappings, and fuzzy relations. Operatorsandother threshold, and logic circuits.
concepts presented in Chapter I are adapted to fuzzy relations Various Theauthor
has succeeded in writing an extremely clear and
types of composition of fuzzyrelations are introduced (max-min, pedagogical book, while covering the fust comprehensive survey of the
max-product, etc.). Fuzzy sets induced by mapping of a universe set field (one may nevertheless regret thatthree years wereneeded to
into another and by the membership function of the former are pre- obtain the English translation of the original French text). The book
sented. Conditional fuzzy sets are also considered. is self-contained in the sense that summarized revisions of concepts of
Then the concept of reflexivity, symmetry,and theiropposite, as modem algebra are given before they are used. Numerousexamples
well astransitivity are extended to fuzzy relationsThoseconcepts are provided without neglecting (except in Chapter V) formal proofs.
are central to the second part of the chapter, since fromthem are A list of references is appended, although it is not quoted as often as
defined various types offuzzy relations: pre-order, order, ordinal, it should be in the text.
resemblance, dissemblance, similitude, and dissimilitude. The transi- Readerswho should have this bookon their shelves include prac-
tive closure of a binary relation is linked to the concept of strongest tically everyone interested in system theory (from a practical as well
path. Variousproperties are derived on this closure, in particular, as a theoretical-algebraic point of view). The book would be a very
for pre-ordered relations A pre-ordered relation having classes of good reference at the graduate level, although it is also a perfect self-
similarity is shown to induce anordered relation that is related to teaching text.
the ordinal function of the nonfuzzygraph of this latter relation.
Reprinted from IEEE Transactions on Systems, Man, and Cybernetics,
The formal differences and analogies between fuzzy set theory and June 1977,
probability theory are summarized. Thechapter concludeswith two
applications of fuzzy functions to fuzzypropositions and reliable Applied ComputationTheory Analysis, Design, Modeling-Raymond
structures T. Yeh, Ed. (Englewood Cliffs, NJ: Prentice Hall Inc., 1976, xxi 624
Chapter IV introduces the concepts of internal composition laws pp., $22.50). Reviewed by Robert M. Keller of the Department of
(a binary operator mapping two elements of a fuzzy power set into Computer Science, University of Utah, Salt Lake City, UT .
anotherelementofthe same power set) andexternal cornposition
laws (the three elements can belong to threedistinct fuzzypower Computation theory (also known as automata theory) is that body of
sets). Attention is given to the internal composition law of a fuzzy mathematical results which deals with the formalization of computa-
power set induced by the internal composition law of the membership tional processes. The theory, which is perhaps not asunified as the

You might also like