Fuzzy Logic-A Modern Perspective
Fuzzy Logic-A Modern Perspective
Fuzzy Logic-A Modern Perspective
1, JANUARY/FEBRUARY 1999
153
F
1 INTRODUCTION
described abovethe cost-precision trade-off. Indeed, providing a cost-effective solution to a wide range of real
world problems is the primary reason that fuzzy logic has
found so many successful applications in industry to date.
Understanding this driving force of the success of fuzzy
logic will prevent us from falling into the trap of debating
whether fuzzy logic can accomplish what X cannot accomplish where X is an alternative technology such as probability theory, control theory, etc. Such a debate is usually
not fruitful because it ignores one important issuecost. A
better question to ask is What is the difference between the
cost of a fuzzy logic approach and the cost of an approach
based on X to accomplish a certain task?
Authorized licensed use limited to: Pontificia Universidad Javeriana. Downloaded on February 19,2010 at 13:00:44 EST from IEEE Xplore. Restrictions apply.
154
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, VOL. 11, NO. 1, JANUARY/FEBRUARY 1999
Authorized licensed use limited to: Pontificia Universidad Javeriana. Downloaded on February 19,2010 at 13:00:44 EST from IEEE Xplore. Restrictions apply.
evolutionary computing, which includes genetic algorithms, evolutionary strategies, and evolutionary programming. Genetic algorithms (GA) and evolutionary
strategies are optimization techniques that attempt to avoid
being easily trapped in local minima by simultaneously
exploring multiple points in the search space and by generating new points based on the Darwinian theory of evolutionsurvival of the fittest. The popularity of GA in the
1990s inspired the use of GA for optimizing parameters in
fuzzy systems [13]. Various synergistic combinations of
neural networks, genetic algorithms, and fuzzy logic help
people to view them as complementary. To distinguish
these paradigms from the conventional methodologies
based on precise formulations, Zadeh introduced the term
soft computing in the early 1990s [41].
155
Authorized licensed use limited to: Pontificia Universidad Javeriana. Downloaded on February 19,2010 at 13:00:44 EST from IEEE Xplore. Restrictions apply.
156
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, VOL. 11, NO. 1, JANUARY/FEBRUARY 1999
( A B) ( x) = A ( x) B ( x)
A B (x) = A (x) + B (x) A (x) B (x)
(1)
(2)
(4)
by a function that maps ordered pairs in X Y to their degree in the relation, which is a number between 0 and 1, i.e.,
R: X Y [0, 1]. More generally, a fuzzy n-ary relation R
in, x1, x2, , xn, whose domains are X1, X2, , Xn, respectively, is defined by a function that maps an n-tuple <x1, x2,
, xn> in X1 X2 Xn to a number in the interval, i.e.,
R: X1 X2 Xn [0, 1]. Just as a classical relation can
be viewed as a set, a fuzzy relation can be viewed as a
fuzzy subset. From this perspective, the mapping above is
equivalent to the membership function of a multidimensional fuzzy set.
If the possible values of x and y are discrete, we can express a fuzzy relation in a matrix form. For example, suppose we wish to express a fuzzy relation Petite in terms of
the height and the weight of a female. Suppose the range of
the height and the weight of interest to us are {5, 51, 52,
53, 54, 55, 56}, denoted h, and {90, 95, 100, 105, 110,
115}, (in pounds) denoted w, respectively. We can express
the fuzzy relation in a matrix form as shown below:
90
5
51"
52"
53"
54"
55"
56"
11
!
1
1
1
1
1 0.5 0.2
1
1
1
1 0.9 0.3 0.1
1
1
1
1
1 0.7 0.1 0
1
1
1
1 0.5 0.3 0
0
0.8 0.6 0.4 0.2 0
0
0
0
0.6 0.4 0.2 0
0
0
0
0
0
0
0
0
0
0
0
0
"#
##
##
#$
Authorized licensed use limited to: Pontificia Universidad Javeriana. Downloaded on February 19,2010 at 13:00:44 EST from IEEE Xplore. Restrictions apply.
xw j [ Possible weight( x , w j )
( Possible height( x , hi ) Petite( hi , w j ))]
157
(5)
hi
(6)
where and denote fuzzy conjunction and fuzzy disjunction, respectively. We define the compositional rule of
inference more formally below.
DEFINITION 3. Let X and Y be the universes of discourse for variables x and y, respectively, and Xi and Yj be elements of X
and Y. Let R be a fuzzy relation that maps X Y to [0, 1]
and the possibility distribution of X is known to be PX (xi).
The compositional rule of inference infers the possibility
distribution of Y as follows:
Y ( y j ) = ( X ( xi ) R ( xi , y j ))
xi
(7)
(8)
xi
(9)
xi
Authorized licensed use limited to: Pontificia Universidad Javeriana. Downloaded on February 19,2010 at 13:00:44 EST from IEEE Xplore. Restrictions apply.
158
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, VOL. 11, NO. 1, JANUARY/FEBRUARY 1999
(10)
Such a set-to-set implication specifies a set of possible implications such as x = b y = s, and a set of impossible implications such as x = b y = r. Notice that if the antecedent is
known to be false, the implication is true regardless of ys
value. Therefore, the following implications is possible: x =
a y = r. Hence, we can represent the meaning of the setto-set implication using the following matrix:
r s t u v
a
b
c
R( xi , y j ) = d
e
f
01
!
0
0
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0
0
0
1
1
1
0
0
0
1
1
"#
##
##
$
( x is A) ( y is B)
(15)
(16)
(11)
xi and yj and determine whether the resulting (i.e., instantiated) implication is true or false, i.e.,
j
if
if
%&1
'0
%1
R( x , y ) = &
'0
R( xi , yi ) =
(13)
(18)
r (( x = xi ) ( y = y j )) = t(( xi is A) ( y j is B))
k
= A ( xi ) B ( y j )
(19)
Such a set-to-set mapping specifies a set of possible association between the values of x and y. Such an association
can be naturally represented by a relation R( xi , y j ) that
is B in (17) is defined in terms of the truth value of the antecedent xi is A and the truth value of the consequent yi
is B. For the convenience of our discussion, we will refer to
describes whether input-output pair is a possible association based on the mapping rule. Hence, the relation for the
mapping rule can be represented as follows:
Authorized licensed use limited to: Pontificia Universidad Javeriana. Downloaded on February 19,2010 at 13:00:44 EST from IEEE Xplore. Restrictions apply.
159
(a)
(b)
Fig. 2. A pictorial view of the possibility relation for: (a) an implication rule; (b) a corresponding mapping rule.
y is V (unknown)
Notice that y is V (assigning the entire universe of discourse V to y) represents y is unknown. Other intuitive
criteria are summarized in Table 1.
Fuzzy implications can be classified into three families. Each family extends a particular logic formulation of
implication in propositional logic. Even though these formulations are equivalent in classical logic, they are not
equivalent in fuzzy logic because the law of excluded middle no longer holds in fuzzy logic as we explained in Section 3. Fuzzy implications within a family differ on their
choice of the fuzzy conjunction and fuzzy disjunction operators. We briefly describe each family below.
The first family of fuzzy implication is obtained by generalizing material implications in two-valued logic to fuzzy
logic. A material implication p q is defined as p q.
Generalizing this to fuzzy logic gives us t(p q) = t(p q).
More specifically, fuzzy implications in this family can be
generically defined as:
t( xi is A y j is B) = t( ( xi is A) ( y j is B))
= (1 A ( xi )) B ( y j )
(20)
(21)
TABLE 1
INTUITIVE CRITERIA FOR REASONING
INVOLVING FUZZY IMPLICATION x IS A y
IS
Authorized licensed use limited to: Pontificia Universidad Javeriana. Downloaded on February 19,2010 at 13:00:44 EST from IEEE Xplore. Restrictions apply.
160
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, VOL. 11, NO. 1, JANUARY/FEBRUARY 1999
t( xi is A y j is B) = t( ( xi is A) [( xi is A) ( y j is B)])
= (1 A ( xi )) ( A ( xi ) B ( y j ))
(22)
t( xi is A y j is B)
= sup{ [0, 1], t( xi is A) t( yi is B)}
A ( xi ) B ( y j )
A ( xi ) > B ( y j )
(25)
t( xi is A y j is B) =
%& 1
' ( y )
B
A ( xi ) B ( y j )
A ( xi ) > B ( y j )
(26)
t( xi is A y j
%K 1
is B) = & ( y )
K' (x )
B
A ( xi ) B ( y j )
A ( xi ) > B ( y j )
%&1
'0
EXAMPLE 1. Let U and V be two universes representing numeric ratings (from 1 to 10) of redness and ripeness of
tomatoes, respectively. We denote the variable of
these two ratings as x and y, respectively. Let Red be a
fuzzy subset of U defined as
t( xi is A y j is B) =
concepts and techniques in fuzzy logic such as the compositional rule of inference, fuzzy relations, and possibility
distributions. Without them, approximate reasoning would
not have been possible.
Table 2 summaries how the criteria introduced in Table 1
are satisfied by the five fuzzy implication functions based
on sup-min composition (except that the sup-product composition is applied to Goguens fuzzy implication). We applied the sup-product composition to Goguens fuzzy implication because Goguens multivalued logic system uses
product as its conjunction operator. In other words, the
choice of implication function and the choice of disjunction/conjunction operator in the compositional rule of inference are not unrelated.
(27)
These three fuzzy implication functions came from, respectively, the standard sequence many-valued logic system (denoted Sn in the literature), a many-valued logic
system proposed by K. Godel, and a many-valued logic
system J.A. Goguen introduced in 1969. Fig 3 shows
graphically the function surface of the five fuzzy implication functions we discussed.
Even though implication functions in multivalued logic
systems can be used for constructing fuzzy implication relations, approximate reasoning in fuzzy logic is fundamentally different from logic inference in multivalued logic
approximate reasoning infers possible values of a variable,
whereas multivalued logic infers the truth values of propositions. The connection between the two was established by
(17). Even if we choose to use a fuzzy implication function
originated in a multivalued logic system (e.g., standard
sequence, Godelian implication, or Goguens implication),
approximate reasoning still benefits from other important
1 1 3
1]
4 2 4
(29)
Authorized licensed use limited to: Pontificia Universidad Javeriana. Downloaded on February 19,2010 at 13:00:44 EST from IEEE Xplore. Restrictions apply.
161
Fig. 3. Five fuzzy implication functions: (a) Zadehs arithmetic fuzzy implication; (b) Zadehs maxmin fuzzy implication; (c) standard sequence
fuzzy implication; (d) Godelian fuzzy implication; and (e) Goguen fuzzy implication.
TABLE 2
SATISFACTION OF FUZZY INFERENCE CRITERIA BY FUZZY IMPLICATION FUNCTIONS I (x, y )
162
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, VOL. 11, NO. 1, JANUARY/FEBRUARY 1999
(30)
A B ( u , v ) = A ( u) B ( v )
An expression of the form A B where A and B are
words (fuzzy sets) is referred as a Cartesian granule [42].
Fig. 4 depicts a fuzzy graph consisting of three fuzzy mapping rules:
f:
(31)
where
represents the fuzzy graph of a given fuzzy
model, o denotes the compositional rule of inference.
(32)
where Ai, Bi, and Ci are fuzzy subsets of U, V, and W, respectively. The models fuzzy graph f is expressed as
f =
i =1
Ri =
( Ai Bi ) Ci
i =1
(33)
( A B ) o f = ( A B ) o Ri
i =1
(34)
1in
Authorized licensed use limited to: Pontificia Universidad Javeriana. Downloaded on February 19,2010 at 13:00:44 EST from IEEE Xplore. Restrictions apply.
163
C ( z) = max( i Ci ( z))
(35)
i =1
where
i = sup ( A ( x ) Ai ( x )) sup ( B ( y ) Bi ( y ))
x
z = Centroid
n
i =1
Ai ( x0 )
B ( y0 ) Ci
i
(36)
Authorized licensed use limited to: Pontificia Universidad Javeriana. Downloaded on February 19,2010 at 13:00:44 EST from IEEE Xplore. Restrictions apply.
164
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, VOL. 11, NO. 1, JANUARY/FEBRUARY 1999
Fig. 5. Various methods for partitioning the input space: (a) gridpartition (static); (b) grid partition (adaptive); (c) tree partition; (d) scatter partition.
8 SUMMARY
In this paper, we have summarized major concepts and
techniques in fuzzy logic. We have also presented a modern
perspective about two types of fuzzy rules: 1) fuzzy implication rules, and 2) fuzzy mapping rules. The latter have been
widely used in fuzzy logic control and other industrial applications. This new perspective not only clarifies the formal foundation of these rules, but also sheds lights on how
to deal with various challenges in identifying and learning
fuzzy rule-based models for high dimensional problems.
ACKNOWLEDGMENTS
We thank Lotfi A. Zadeh, Henri Prade, and Didies Dubois
for fruitful technical exchanges regarding fuzzy logic and
artificial intelligence. We also thank Reza Langari and Liang
Wang for their contributions to our research and teaching
related to fuzzy logic. This research was partially supported
by National Science Foundation Young Investigator Award
No. IRI 9257293.
ReFERENCE
[1] B. Babuska, M. Setnes, U. Kaymak, and H.R. van Nauta Lemke,
Rule Base Simplification with Similarity Measures, Proc. Fifth
IEEE Intl Conf. Fuzzy Systems, pp. 1,642-1,647, New Orleans, Sept.
1996.
[2] J.C. Bezdek, Fuzzy Mathematics in Pattern Classification, PhD
thesis, Center for Applied Mathematics, Cornell Univ., Ithaca,
New York, 1973.
[3] J. Bezdek, Fuzziness vs. ProbabilityAgain!? IEEE Trans. Fuzzy
Systems, vol. 2, no. 1, pp. 1-3, Feb. 1994.
[4] C.T. Chen, Y.J. Chen, and C.C. Teng, Simplification of FuzzyNeural Systems Using Similarity Analysis, IEEE Trans. Systems,
Man, and Cybernetics, vol. 26, pp. 344-354, 1996.
Authorized licensed use limited to: Pontificia Universidad Javeriana. Downloaded on February 19,2010 at 13:00:44 EST from IEEE Xplore. Restrictions apply.
165
[30] J. Yen and R. Langari, Fuzzy Logic: Intelligence, Control, and Information. Prentice Hall, 1999.
[31] J. Yen, R. Langari, and L.A. Zadeh, Industrial Applications of Fuzzy
Logic and Intelligent Systems. IEEE CS Press, 1995.
[32] J. Yen and L. Wang, Application of Statistical Information Criteria for Optimal Fuzzy Model Construction, IEEE Trans. Fuzzy
Systems, vol. 6, no. 3, pp. 362-372, Aug. 1998.
[33] J. Yen, L. Wang, and W. Gillespie, Improving the Interpretability
of TSK Fuzzy Models by Combining Global Learning and Local
Learning, IEEE Trans. Fuzzy Systems, vol. 6, no. 4, pp. 530-537,
Nov. 1998.
[34] J. Yen and L. Wang, Simplifying Fuzzy Rule-Based Models Using
Orthogonal Transformation Methods, IEEE Trans. Systems, Man,
and Cybernetics, vol. 29: Par B, no. 1, Feb. 1999.
[35] L.A. Zadeh, Fuzzy Sets, Information and Control, vol. 8, 1965.
[36] L.A. Zadeh, Probability Measures and Fuzzy Events, J. Math.
Analysis and Applications, vol. 23, no. 2, pp. 421-427, 1968.
[37] L.A. Zadeh, Toward a Theory of Fuzzy Systems, Aspects of Network and System Theory. pp. 469-490. New York: Rinehart and
Winston, 1971.
[38] L.A. Zadeh, Outline of a New Approach to the Analysis of
Complex Systems and Decision Processes, IEEE Trans. Systems,
Man, and Cybernetics, vol. 3, 1973.
[39] L.A. Zadeh, The Concept of a Linguistic Variable and Its Application to Approximate Reasoning-I, II, III, Information Sciences,
vol. 8, pp. 199-249, pp. 301-357; vol. 9, pp. 43-80, 1975.
[40] L.A. Zadeh, Possibility Theory and Soft Data Analysis, Cobb, L.
and R.M. Thrall, eds., Math. Frontiers of the Social and Policy Sciences, pp. 69-129, Boulder, Colo.: Westview Press, 1981.
[41] L.A. Zadeh, Fuzzy Logic, Neural Networks, and Soft Computing, Comm. ACM, vol. 37, no. 3, pp. 77-84, Mar. 1994.
[42] L.A Zadeh, Fuzzy Logic = Computing with Words, IEEE Trans.
on Fuzzy Systems, vol. 4, no. 2, 1996.
Authorized licensed use limited to: Pontificia Universidad Javeriana. Downloaded on February 19,2010 at 13:00:44 EST from IEEE Xplore. Restrictions apply.