1

A Brief Tutorial on Interval Type-2 Fuzzy Sets and Systems

Dongrui Wu

Abstract This tutorial illustrates the basic ideas of interval type-2 (IT2) fuzzy sets and systems, and provides a Matlab implementation of IT2 fuzzy logic system (FLS). One obstacle in learning IT2 fuzzy logic is its complex notations. In this tutorial we try to avoid these notations and give the reader some intuitive understanding of IT2 FLSs.

I. T YPE -1 F UZZY S ETS (T1 FS S ) Type-1 fuzzy set (T1 FS) theory was rst introduced by Zadeh [56] in 1965 and has been successfully applied in many areas, including modeling and control [6], [8], [24], [43], [54], data mining [4], [21], [37], [53], [58], time-series prediction [27], [29], [41], etc. An example of a T1 FS, X , is shown in Fig. 1(a). When only integer numbers are considered in the x domain, the T1 FS can be represented as {0/2, 0.5/3, 1/4, 1/5, 0.67/6, 0.33/7, 0/8}, where 0/2 means that number 2 has a membership degree of 0 in the T1 FS X , 0.5/3 means number 3 has a membership degree of 0.5 in the T1 FS X , etc. In contrast, for a crisp set, the membership degree of each element in it can be either 0 or 1; there is no value (e.g., 0.5) in between.

(a) Fig. 1. Examples of a T1 FS (a) and an IT2 FS (b).

(b)

The membership function (MF), X (x), of a T1 FS can either be chosen based on the users opinion (hence, the MFs from two individuals could be quite different depending upon their experiences, perspectives, cultures, etc.), or, it can be designed using optimization procedures [22], [23], [42]. II. I NTERVAL T YPE -2 F UZZY S ETS (IT2 FS S ) Despite having a name which carries the connotation of uncertainty, research has shown that there are limitations in the ability of T1 FSs to model and minimize the effect of uncertainties [19], [20], [33], [50]. This is because a
Dongrui Wu was with the Signal and Image Processing Institute, Ming Hsieh Department of Electrical Engineering, University of Southern California, Los Angeles, CA 90089-2564, USA. He is now with the Industrial Articial Intelligence Lab, GE Global Research, Niskayuna, NY 12309 USA (phone: 213-595-3269; email: [email protected]). This tutorial can be distributed freely.

0) (

& ' %

  

$ ! # " 

 " '1

   

T1 FS is certain in the sense that its membership grades are crisp values. Recently, type-2 FSs [57], characterized by MFs that are themselves fuzzy, have been attracting interests. Interval type-2 (IT2) FSs1 [33], a special case of type-2 FSs, are currently the most widely used for their reduced computational cost. An example of an IT2 FS, X , is shown in Fig. 1(b). Observe that unlike a T1 FS, whose membership for each
x is a number, the membership of an IT2 FS is an interval. For example, the membership of number 3 is [0.25, 1], and the membership of number 5 is [0.75, 1]. Observe also that an IT2 FS is bounded from the above and below by two T1 FSs, X and X , which are called upper MF (UMF) and lower MF (LMF), respectively. The area between X and X is the footprint of uncertainty (FOU).

IT2 FSs are particularly useful when it is difcult to determine the exact MF, or in modeling the diverse opinions from different individuals. The MFs can be constructed from surveys [30], [35], [45] or using optimization algorithms [50], [51].

III. I NTERVAL T YPE -2 F UZZY L OGIC S YSTEM (IT2 FLS) Fig. 2 shows the schematic diagram of an IT2 FLS. It is similar to its T1 counterpart, the major difference being that at least one of the FSs in the rule base is an IT2 FS. Hence, the outputs of the inference engine are IT2 FSs, and a type-reducer is needed to convert them into a T1 FS before defuzzication can be carried out.

&ULVS LQSXWV

)X]]LILHU

5XOHEDVH ,QIHUHQFH (QJLQH

'HIX]]LILHU
7 )6

&ULVS RXWSXW

,7 )6V
Fig. 2. An IT2 FLS.

,7 )6V

7\SHUHGXFHU

In practice the computations in an IT2 FLS can be signicantly simplied. Consider the rulebase of an IT2 FLS consisting of N rules assuming the following form:
Rn :
n n IF x1 is X1 and and xI is XI , THEN y is Y n

n = 1, 2, ..., N

where Xin (i = 1, . . . , I) are IT2 FSs, and Y n = [y n , y n ] is an interval, which can be understood as the centroid [25], [33] of a consequent IT2 FS2 , or the simplest TSK model, for its simplicity. In many applications we use y n = y n , i.e., each rule consequent is a crisp number. Assume the input vector is x = (x , x , ..., x ). Typical computations in an IT2 FLS involve the following steps: 1 2 I 1) Compute the membership of x on each Xin , [X n (x ), X n (x )], i = 1, 2, ..., I , n = 1, 2, ..., N . i i i i i 2) Compute the ring interval of the nth rule, F n (x ):
F n (x ) = [X n (x ) X n (x ), X n (x ) X n (x )] [f n , f ], 1 I 1 I 1 I 1 I
n

n = 1, ..., N

(1)

Note that the minimum, instead of the product, can be used in (1).
IT2 FSs have also been called interval-valued fuzzy sets in the literature [10], [11], [18], [39], [40]. They can also be mapped into intuitionistic fuzzy sets [1][3]. Deschrijver and Kerre [14] have a comprehensive study on the relationships some extensions of T1 FSs, including interval-valued FSs, intuitionistic FSs, interval-valued intuitionistic FSs [1], and L-FSs [17]. 2 The rule consequents can be IT2 FSs; however, when the popular center-of-sets type-reduction method [33] is used, these consequent IT2 FSs are replaced by their centroids in the computation; so, it is more convenient to represent the rule consequents as intervals directly.
1

3) Perform type-reduction to combine F n (x ) and the corresponding rule consequents. There are many such methods. The most commonly used one is the center-of-sets type-reducer3 [33]:
N

f nyn = [yl , yr ] fn

Ycos (x ) =
f n F n (x ) yn Y n

n=1 N n=1

(2)

It has been shown that [33], [35], [46]:

yl = yr = min
n n k n=1 f y n k n=1 f k n n n=1 f y k n n=1 f

+ + + +

k[1,N 1]

k[1,N 1]

max

N n n n=k+1 f y N n n=k+1 f n n N n=k+1 f y n N n=k+1 f

n n L n=1 f y n L n=1 f R n n n=1 f y R n n=1 f

+ + + +

N n n n=L+1 f y N n n=L+1 f n n N n=R+1 f y n N n=R+1 f

(3) (4)

where the switch points L and R are determined by

yL yR yl yr y L+1 y R+1

(5) (6)

and {y n } and {y n } have been sorted in ascending order, respectively. yl and yr can be computed using the Karnik-Mendel (KM) algorithms [33] as follows: KM Algorithm for Computing yl : a) Sort y n (n = 1, 2, . . . , N ) in increasing order and call the sorted y n by the same name, but now y2 y N . Match the weights F n (x ) with their respective y n and renumber them so that y1 their index corresponds to the renumbered y n . b) Initialize f n by setting
fn + f f = 2
n n

n = 1, 2, . . . , N

(7)

and then compute

y=
N n n n=1 y f N n n=1 f

(8)

c) Find switch point k (1

N 1) such that yk y y k+1

(9)

d) Set
f =
n

f , n k f n, n > k

(10)

and compute
y =
N n n n=1 y f N n n=1 f

(11)

e) Check if y = y . If yes, stop and set yl = y and L = k. If no, go to Step 6. f) Set y = y and go to Step 3.
Another popular type-reducer is the centroid type-reduction method [33], which combines all the rule-output IT2 FSs by their union and then nds its centroid as the type-reduced set. Note that this method requires the rule consequents to be IT2 FSs instead of intervals.
3

KM Algorithm for Computing yr : a) Sort y n (n = 1, 2, . . . , N ) in increasing order and call the sorted y n by the same name, but now y1 y2 y N . Match the weights F n (x ) with their respective y n and renumber them so that their index corresponds to the renumbered y n . b) Initialize f n by setting
fn + f f = 2
n n

n = 1, 2, . . . , N

(12)

and then compute

y=
N n n n=1 y f N n n=1 f

(13)

c) Find switch point k (1

N 1) such that yk y y k+1

(14)

d) Set
fn = f n, n k n f , n>k

(15)

and compute
y =
N n n n=1 y f N n n=1 f

(16)

e) Check if y = y . If yes, stop and set yr = y and R = k. If no, go to Step 6. f) Set y = y and go to Step 3. The main idea of the KM algorithm is to nd the switch points for yl and yr . Take yl for example. yl is the minimum of Ycos (x ). Since y n increases from the left to the right along the horizontal axis of Fig. 3(a), we should choose a large weight (upper membership grade) for y n on the left and a small weight (lower membership grade) for y n on the right. The KM algorithm nds the switch point L. For n L, the upper membership grades are used to calculate yl ; for n > L, the lower membership grades are used. This will ensure yl be the minimum. yl and yr can also be computed more efciently by many other algorithms [16], [32], [46], [47], [55]. A comparison of them is presented in [47], where the Matlab implementation of the most efcient algorithm is also given.

(a) Computing yl : switch from the upper ring level to (b) Computing yr : switch from the lower ring level to the lower ring level. the upper ring level. Fig. 3. Illustration of the switch points in computing yl and yr . The switch points can be found by the ve algorithms introduced in [47].

53

63

43

23

4) Compute the defuzzied output as:

yl + yr . (17) 2 Comment: There are many other ways to perform type-reduction and defuzzication, e.g., [5], [15], [36], [48], [52]. However, the KM method is the most widely used approach. So far IT2 FLSs have been used in almost all application areas of T1 FLSs, and better performances were reported from many experiments. The reason may be that IT2 FLSs can achieve input-output mappings which cannot be obtained from T1 FLSs [44], [49]. However, there is no theoretical results on when and how IT2 FLSs can outperform T1 FLSs. y=

A. Example of an IT2 FLS In this section, the mathematical operations in an IT2 FLS are illustrated using an example. Consider an IT2 FLS that has two inputs (x1 and x2 ) and one output (y ). Each input domain consists of two IT2 FSs, shown as the shaded areas in Fig. 4.
CC DC DC BA AA
X11 X12

% X X X 1.5

% X
AA BA

X 0.5 0.3 0

0.5

1.5

1.5

0.5

0.6

(a) Input MFs of x1 . Fig. 4. MFs of the IT2 FLS.

(b) Input MFs of x2 .

The rulebase has the following four rules:

R1 : R2 : R3 : R4 :

IF x1 is X11 and x2 is X21 , THEN y is Y 1 . IF x1 is X11 and x2 is X22 , THEN y is Y 2 . IF x1 is X12 and x2 is X21 , THEN y is Y 3 . IF x1 is X12 and x2 is X22 , THEN y is Y 4 .

The complete rulebase and the corresponding consequents are given in Table I.
TABLE I RULEBASE AND CONSEQUENTS OF THE IT2 FLS.

PP P x2 x1 PP P

X21 Y 1 = [y1 , y 1 ] = [1, 0.9] Y 3 = [y3 , y 3 ] = [0.4, 0.6]

X22 Y 2 = [y2 , y 2 ] = [0.6, 0.4] Y 4 = [y 4 , y4 ] = [0.9, 1]

Consider an input vector x = (x , x ) = (0.3, 0.6). The ring intervals of the four IT2 FSs are: 1 2
[X 11 (x ), X 11 (x )] = [0.4, 0.9] 1 1 [X 12 (x ), X 12 (x )] = [0.1, 0.6] 1 1 [X 21 (x ), X 21 (x )] = [0, 0.45] 2 2

DC

X 1

CC

CC

% X

% X

BA AA

1.5

(18) (19) (20)

[X 22 (x ), X 22 (x )] = [0.55, 1] 2 2

(21)

The ring intervals of the four rules are: Rule No.:

R1 :
1 1

Firing Interval
[f , f ] = [X 11 (x1 ) = [0.4 0, 0.9 0.45] = [0, 0.405]
2

X 21 (x )] 2

Consequent
[y 1 , y 1 ] = [1, 0.9]

X 21 (x ), X 11 (x ) 1 2

R2 : R3 : R4 :

[f 2 , f ] = [X 11 (x ) X 22 (x ), X 11 (x ) X 22 (x )] 1 2 1 2 = [0.4 0.55, 0.9 1] = [0.22, 0.9] [f 3 , f ] = [X 12 (x ) X 21 (x ), X 12 (x ) X 21 (x )] 1 2 1 2 = [0.1 0, 0.6 0.45] = [0, 0.27] [f 4 , f ] = [X 12 (x ) X 22 (x ), X 12 (x ) X 22 (x )] 1 2 1 2 = [0.1 0.55, 0.6 1] = [0.055, 0.6]
1 4 3

[y 2 , y 2 ] = [0.6, 0.4] [y 3 , y 3 ] = [0.4, 0.6] [y 4 , y 4 ] = [0.9, 1]

From the KM algorithms, we nd that L = 1 and R = 3. So,

yl = f y1 + f 2 y2 + f 3 y3 + f 4y4
1

f + f2 + f3 + f4 0.405 (1) + 0.22 (0.6) + 0 0.4 + 0.055 0.9 = 0.405 + 0.22 + 0 + 0.055 = 0.7169 f 1 y1 + f 2 y2 + f 3 y3 + f y4
4 4

yr =

f1 + f2 + f3 + f 0 (0.9) + 0.22 (0.4) + 0 0.6 + 0.6 1 = 0 + 0.22 + 0 + 0.6 = 0.6244

Finally, the crisp output of the IT2 FLS, y , is:

y= 0.7169 + 0.6244 yl + yr = = 0.0463. 2 2

B. Matlab Implementation A Matlab function, IT2FLS.m, is provided for computing the output of an IT2 FLS given its rulebase and inputs. It uses the most efcient algorithm in [47] to compute yl and yr . For details, please refer to the explanations in IT2FLS.m. The above example is implemented in example.m. Note that each IT2 FS is repre sented by a 9-point vector (p1 , ..., p9 ) as shown in Fig. 5. So, the IT2 FS X11 in Fig. 4(a) is represented as (1.5, 1.5, .5, 1.5, 1.5, 1.5, 1.5, .5, 1), and X12 is represented as (1.5, .5, 1.5, 1.5, .5, 1.5, 1.5, 1.5, 1).

u
1

% X

Fig. 5.

The 9-point representation of an IT2 FS.

S
p p p p p p p p

IV. AWARDS

AND

F URTHER R EADINGS

The following is a list of international awards given to works on type-2 fuzzy logic (there may be others): 1) 2) 3) 4) 5) 6) 7) 8) 9) 10) 11) 12) Best Paper Award, IEEE Trans. on Fuzzy Systems, 1999 (Karnik, Mendel and Liang) [26]. Best Paper Award, IEEE Trans. on Fuzzy Systems, 2004 (Hagras) [19]. Best Student Paper Award, IEEE Intl. Conf. on Fuzzy Systems, 2005 (Wu and Tan) [49]. First Prize, IEEE Region 10 Postgraduate Student Paper Contest, 2005 (Zeng) [59]. Best Student Paper Award, IEEE Intl. Conf. on Fuzzy Systems, 2006 (Lynch, Hagras and Callaghan) [31]. Best Poster Presentation Award, IEEE Intl. Conf. on Fuzzy Systems, 2006 (Tan and Kamal) [38]. Best Paper Award, IEEE Trans. on Fuzzy Systems, 2007 (Coupland and John) [13]. Best Student Paper Award, IEEE-NAFIPS (North American Fuzzy Information Processing Society) Conference, 2007 (Celikyilmaz and Turksen) [12]. Fuzzy Systems Pioneer Award, IEEE Computational Intelligence Society, 2008 (Mendel). First Prize, IEEE Region 9 Undergraduate Student Paper Contest, 2008 (Bulla and Melgarejo) [9]. First Prize, IEEE Region 10 Undergraduate Student Paper Contest, 2008 (Li) [28]. Best Student Paper Award, NAFIPS (North American Fuzzy Information Processing Society) Conference, 2010 (Biglarbegian) [7].

You are also encouraged to refer to the following publications for more information on (interval) type-2 fuzzy sets and systems:

Jerry M. Mendel, Uncertain Rule-Based Fuzzy Logic Systems: Introduction and New Directions, Upper Saddle River, NJ: Prentice-Hall, 2001. Oscar Castillo and Patricia Melin, Type-2 Fuzzy Logic: Theory and Applications, Springer-Verlag, Berlin, 2008. Nilesh N. Karnik, Jerry M. Mendel, and Qilian Liang, Type-2 fuzzy logic systems, IEEE Trans. on Fuzzy Systems, vol. 7, No. 6, pp. 643-658, 1999. Qilian Liang and Jerry M. Mendel, Interval type-2 fuzzy logic systems: theory and design, IEEE Trans. on Fuzzy Systems, vol. 8, No. 5, pp. 535-550, 2000. Jerry M. Mendel and Robert I. John, Type-2 fuzzy sets made simple, IEEE Trans. on Fuzzy Systems, vol. 10, No. 2, pp. 117-127, 2002. Jerry M. Mendel, Robert I. John and Feilong Liu, Interval type-2 fuzzy logic systems made simple, IEEE Trans. on Fuzzy Systems, vol. 14, No. 6, pp. 808-821, 2006. Hani Hagras,Type-2 FLCs: A new generation of fuzzy controllers, IEEE Computational Intelligence Magazine, vol. 2, pp. 30-43, 2007. Jerry M. Mendel,Type-2 fuzzy sets and systems: an overview, IEEE Computational Intelligence Magazine, vol. 2, pp. 20-29, 2007. Jerry M. Mendel,Advances in type-2 fuzzy sets and systems, Information Sciences, Vol. 177, pp. 84-110, 2007. Jerry M. Mendel, Type-2 fuzzy sets and systems: How to learn about them, IEEE SMC eNewsletter, Issue 27, June 2009. Janet Aisbett, John T. Rickard, and David Morgenthaler, Type-2 fuzzy sets as functions on spaces, IEEE Trans. on Fuzzy Systems, vol. 18, no. 4, pp. 841-844, 2010. ACKNOWLEDGEMENT

The author would like to thank Professor Jerry M. Mendel for his support and valuable suggestions.

A PPENDIX A S OME F ORMAL D EFINITIONS

FOR

T1

AND

IT2 FS S

Denition 1: A type-1 fuzzy set X is comprised of a domain DX of real numbers (also called the universe of discourse of X ) together with a membership function (MF) X : DX [0, 1], i.e.,
X=
DX

X (x)/x

(22)

Here denotes the collection of all points x DX with associated membership grade X (x). Denition 2: [33], [34] An IT2 FS X is characterized by its MF X (x, u), i.e.,
X=
xDX uJx [0,1]

X (x, u)/(x, u) 1/(x, u)

xDX uJx [0,1]

=
xDX

uJx [0,1]

where x, called the primary variable, has domain DX ; u [0, 1], called the secondary variable, has domain Jx [0, 1] at each x DX ; Jx is also called the support of the secondary MF, and is dened below in (25); and, the amplitude of X (x, u), called a secondary grade of X , equals 1 for x DX and u Jx [0, 1]. For general type-2 FSs [33] X (x, u) can be any number in [0, 1], and it varies as x and/or u vary. An example of an IT2 FS is shown in Fig. 6.
V

1/u

(23)

Fig. 6.

An interval type-2 fuzzy set.

Denition 3: Uncertainty about X is conveyed by the union of all its primary memberships, which is called the footprint of uncertainty (FOU) of X (see Fig. 6), i.e., F OU (X) =
xDX

Jx = {(x, u) : u Jx [0, 1]} .

The size of an FOU is directly related to the uncertainty that is conveyed by an IT2 FS. So, an FOU with more area is more uncertain than one with less area. Denition 4: The upper membership function (UMF) and lower membership function (LMF) of X are two T1 MFs X and X that bound the FOU (see Fig. 6). Note that the primary membership Jx is an interval, i.e.,
Jx = X (x), X (x)

`X b f a e dc g X qi p hW

X Y

(24)

(25)

Using (25), F OU (X) can also be expressed as F OU (X) =

xDX

X (x), X (x)

(26)

R EFERENCES
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] K. Atanassov, Intuitionistic Fuzzy Sets: Theory and Applications. Heidelberg/New York: Physica-Verlag,, 1999. K. Atanassov and G. Gargov, Interval valued intuitionistic fuzzy sets, Fuzzy Sets and Systems, vol. 31, pp. 343349, 1989. K. Atanassov, Intuitionistic fuzzy sets, Fuzzy Sets and Systems, vol. 20, pp. 8797, 1986. J. F. Baldwin, Knowledge from data using fuzzy methods, Pattern Recongnition Letter, vol. 17, pp. 593600, 1996. M. Begian, W. Melek, and J. Mendel, Stability analysis of type-2 fuzzy systems, in Proc. IEEE Intl Conf. on Fuzzy Systems, Hong Kong, June 2008, pp. 947953. J. Bezdek, Fuzzy modelswhat are they, and why? IEEE Trans. on Fuzzy Systems, vol. 1, no. 1, pp. 15, 1993. M. Biglarbegian, Robustness of interval type-2 fuzzy logic systems, in Proc. NAFIPS, Toronto, Canada, July 2010. J. Buckley and H. Ying, Expert fuzzy controller, Fuzzy Sets and Systems, vol. 43, pp. 127137, 1991. J. Bulla and M. Melgarejo, Implementing a simple microcontroller based interval type-2 fuzzy processor, in IEEE Region 9 Undergraduate Student Paper Contest, 2008. P. Burillo and H. Bustince, Entropy on intuitionistic fuzzy sets and on interval-valued fuzzy sets, Fuzzy Sets and Systems, vol. 78, pp. 305316, 1996. H. Bustince, Indicator of inclusion grade for interval-valued fuzzy sets. Application to approximate reasoning based on interval-valued fuzzy sets, International Journal of Approximate Reasoning, vol. 23, no. 3, pp. 137209, 2000. A. Celikyilmaz and I. Turksen, Enhanced type-2 fuzzy system models with improved fuzzy functions, in Proc. Annual Meeting of the North American Fuzzy Information Processing Society, San Diego, July 2007, pp. 140145. S. Coupland and R. I. John, Geometric type-1 and type-2 fuzzy logic systems, IEEE Trans. on Fuzzy Systems, vol. 15, no. 1, pp. 315, 2007. G. Deschrijver and E. E. Kerre, On the relationship between some extensions of fuzzy set theory, Fuzzy Sets and Systems, vol. 133, no. 2, pp. 227235, 2003. X. Du and H. Ying, Control performance comparison between a type-2 fuzzy controller and a comparable conventional mamdani fuzzy controller, in Proc. Annual Meeting of the North American Fuzzy Information Processing Society, San Diego, CA, June 2007, pp. 100105. K. Duran, H. Bernal, and M. Melgarejo, Improved iterative algorithm for computing the generalized centroid of an interval type-2 fuzzy set, in Proc. Annual Meeting of the North American Fuzzy Information Processing Society, New York, May 2008, pp. 15. J. Goguen, L-fuzzy sets, Journal of Mathematical Analysis and Applications, vol. 18, pp. 145174, 1967. M. B. Gorzalczany, A method of inference in approximate reasoning based on interval-valued fuzzy sets, Fuzzy Sets and Systems, vol. 21, pp. 117, 1987. H. Hagras, A hierarchical type-2 fuzzy logic control architecture for autonomous mobile robots, IEEE Trans. on Fuzzy Systems, vol. 12, pp. 524539, 2004. , Type-2 FLCs: A new generation of fuzzy controllers, IEEE Computational Intelligence Magazine, vol. 2, no. 1, pp. 3043, 2007. K. Hirota and W. Pedrycz, Fuzzy computing for data mining, Proc. IEEE, vol. 87, no. 9, pp. 15751600, 1999. S. Horikawa, T. Furahashi, and Y. Uchikawa, On fuzzy modeling using fuzzy neural networks with back-propagation algorithm, IEEE Trans. on Neural Networks, vol. 3, pp. 801806, 1992. J. R. Jang, Self-learning fuzzy controllers based on temporal back-propagation, IEEE Trans. on Neural Networks, vol. 3, pp. 714723, 1992. T. A. Johansen, Fuzzy model based control: Stability, robustness, and performance issues, IEEE Trans. on Fuzzy Systems, vol. 2, no. 3, pp. 221234, 1994. N. N. Karnik and J. M. Mendel, Centroid of a type-2 fuzzy set, Information Sciences, vol. 132, pp. 195220, 2001. N. N. Karnik, J. M. Mendel, and Q. Liang, Type-2 fuzzy logic systems, IEEE Trans. on Fuzzy Systems, vol. 7, pp. 643658, 1999. N. K. Kasabov and Q. Song, DENFIS: Dynamic evolving neural-fuzzy inference system and its application for time-series prediction, IEEE Trans. on Fuzzy Systems, vol. 10, no. 2, pp. 144154, 2002. B. Li, Modeling of decision making process in power market with type-2 fuzzy logic and game theory, in IEEE Region 10 Undergraduate Student Paper Contest, 2008. S. S. Liao, T. H. Tang, and W.-Y. Liu, Finding relevant sequences in time series containing crisp, interval, and fuzzy interval data, IEEE Trans. on Systems, Man, and CyberneticsB, vol. 34, no. 5, pp. 20712079, 2004.

[16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29]

10

[30] F. Liu and J. M. Mendel, Encoding words into interval type-2 fuzzy sets using an interval approach, IEEE Trans. on Fuzzy Systems, vol. 16, no. 6, pp. 15031521, 2008. [31] C. Lynch, H. Hagras, and V. Callaghan, Using uncertainty bounds in the design of an embedded real-time type-2 neuro-fuzzy speed controller for marine diesel engines, in Proc. IEEE Intl Conf. on Fuzzy Systems, Vancouver, Canada, July 2006, pp. 72177224. [32] M. Melgarejo, A fast recursive method to compute the generalized centroid of an interval type-2 fuzzy set, in Proc. Annual Meeting of the North American Fuzzy Information Processing Society, San Diego, CA, June 2007, pp. 190194. [33] J. M. Mendel, Uncertain Rule-Based Fuzzy Logic Systems: Introduction and New Directions. Upper Saddle River, NJ: Prentice-Hall, 2001. [34] J. M. Mendel and R. I. John, Type-2 fuzzy sets made simple, IEEE Trans. on Fuzzy Systems, vol. 10, no. 2, pp. 117127, 2002. [35] J. M. Mendel and D. Wu, Perceptual Computing: Aiding People in Making Subjective Judgments. Hoboken, NJ: Wiley-IEEE Press, 2010. [36] M. Nie and W. W. Tan, Towards an efcient type-reduction method for interval type-2 fuzzy logic systems, in Proc. IEEE Intl Conf. on Fuzzy Systems, Hong Kong, June 2008, pp. 14251432. [37] W. Pedrycz, Fuzzy set technology in knowledge discovery, Fuzzy Sets and Systems, vol. 98, pp. 279290, 1998. [38] W. W. Tan and D. H. Kamal, On-line learning rules for type-2 fuzzy controller, in Proc. IEEE Intl Conf. on Fuzzy Systems, Vancouver, Canada, July 2006, pp. 25302537. [39] I. B. T rksen, Interval valued fuzzy sets and fuzzy connectives, Journal of of Interval Computations, vol. 4, pp. 125142, 1993. u [40] , Interval-valued fuzzy sets based on normal forms, Fuzzy Sets and Systems, vol. 20, pp. 191210, 1986. [41] M. Versaci and F. C. Morabito, Fuzzy time series approach for disruption prediction in tokamak reactors, IEEE Trans. on Magnetics, vol. 39, no. 3, pp. 15031506, 2003. [42] L.-X. Wang and J. M. Mendel, Fuzzy basis functions, universal approximation, and orthogonal least-squares learning, IEEE Trans. on Neural Networks, vol. 3, pp. 807813, 1992. [43] L.-X. Wang, A Course in Fuzzy Systems and Control. Upper Saddle River, NJ: Prentice Hall, 1997. [44] D. Wu, An interval type-2 fuzzy logic system cannot be implemented by traditional type-1 fuzzy logic systems, in Proc. World Conference on Soft Computing, San Francisco, CA, May 2011. [45] D. Wu, S. Coupland, and J. M. Mendel, Enhanced interval approach for encoding words into interval type-2 fuzzy sets and its convergence analysis, IEEE Transactions on Fuzzy Systems, 2011, submitted. [46] D. Wu and J. M. Mendel, Enhanced Karnik-Mendel Algorithms, IEEE Trans. on Fuzzy Systems, vol. 17, no. 4, pp. 923934, 2009. [47] D. Wu and M. Nie, Comparison and practical implementation of type-reduction algorithms for type-2 fuzzy sets and systems, in Proc. IEEE Intl Conf. on Fuzzy Systems, Taipei, Taiwan, June 2011. [48] D. Wu and W. W. Tan, Computationally efcient type-reduction strategies for a type-2 fuzzy logic controller, in Proc. IEEE Intl Conf. on Fuzzy Systems, Reno, NV, May 2005, pp. 353358. [49] , Type-2 FLS modeling capability analysis, in Proc. IEEE Intl Conf. on Fuzzy Systems, Reno, NV, May 2005, pp. 242247. [50] , Genetic learning and performance evaluation of type-2 fuzzy logic controllers, Engineering Applications of Articial Intelligence, vol. 19, no. 8, pp. 829841, 2006. [51] , A simplied type-2 fuzzy controller for real-time control, ISA Transactions, vol. 15, no. 4, pp. 503516, 2006. [52] H. Wu and J. M. Mendel, Uncertainty bounds and their use in the design of interval type-2 fuzzy logic systems, IEEE Trans. on Fuzzy Systems, vol. 10, no. 5, pp. 622639, 2002. [53] R. Yager, Database discovery using fuzzy sets, International Journal of Intelligent Systems, vol. 11, pp. 691712, 1996. [54] R. Yager and D. Filev, Essentials of Fuzzy Modeling and Control. John Wiley & Son, 1994. [55] C.-Y. Yeh, W.-H. Jeng, and S.-J. Lee, An enhanced type-reduction algorithm for type-2 fuzzy sets, IEEE Trans. on Fuzzy Systems, 2011, in press. [56] L. A. Zadeh, Fuzzy sets, Information and Control, vol. 8, pp. 338353, 1965. [57] , The concept of a linguistic variable and its application to approximate reasoning-1, Information Sciences, vol. 8, pp. 199249, 1975. [58] , Toward a theory of fuzzy information granulation and its centrality in human reasoning and fuzzy logic, Fuzzy Sets and Systems, vol. 19, pp. 111127, 1997. [59] J. Zeng, Type-2 fuzzy hidden Markov models and their application to phoneme classication, in IEEE Region 10 Postgraduate Student Paper Contest, 2005.