Fuzzy Logic Essay Assignment
Name
Q1A.
In order to prove the property ( A ∪ B )α =A α ∪Bα ∧ ( A ∩ B )α = A α ∩ Bα , we need to address each of
the equality separately. For every x ( A ∩ B)α , there exist ( A ∩ B ) ( X ) ≥ α . This implies that
A ( x )∧B ( x ) are greater than or equal to α. Consequently, x ∈ A α ∩ B α and ( A ∩ B )α ⊆ Aα ∩B α . In
this case, x ∈ A α and x ∈ B α and whereby A(x )≥ α and B(x) ≥ α. Thus, min ¿ and ( A ∩ B)( x) ≥α .
This means that ( x ∈ A ∩B )α and ( A ∩ B )α ⊆ Aα ∩B α . Thus,( A ∩ B )α = Aα ∩B α .
For the second relation, we have, max ¿ for any ( A ∪ B )α. Consequently, x ∈ A α ∪ Bα implying
that ( A ∪ B )α ⊆ Aα ∪ B α. x ∈ A α ∪ Bα . In this case, x ∈ A α and x ∈ B α and whereby A(x )≥ α and
B(x) ≥ α. Thus, max ¿ and ( A ∪ B)( x )≥ α. ( A ∪ B )α ⊆ Aα ∪ B α . Thus,( A ∪ B )α =A α ∪Bα .
Q1B
A set consisting of elements whose membership values are equal to or greater than a given
threshold value α ∈ [ 0,1 ] . The q-cut of a fuzzy set can be expressed as Aα ={x ∈ X∨ A ( x)≥ α .
∈X ∨ A (x)≥ α ¿
The q-cut is considered to be strong when it identifies all elements of X whereby A+¿={x
α
. The supp (A) and the Core (A) are the limits of the q-sets and the strong q-cuts. When α =0 (
A+¿ ¿
0 ) for a strong q-cut, we arrive at the supp (A). When α =1 for a q-cut we have the core (A).
Q1C
A=\{ ( 0.2 , x 1 ) , ( 0.4 , x 2 ) , ( 0.6 , x 3) , ( 0.8 , x 4 ) ,(1 , x 5) \}
Then, the q-cut representation is as follows:
A0.2 ={x 1 , x 2 , x 3 , x 4 , x 5 }
A0.4 ={x 2 , x 3 , x 4 , x 5 }
A0.6 ={x 3 , x 4 , x5 }
A0.8 ={x 4 , x 5 }
A1={x 5 }
α
A fuzzy set can then be defined for each α based on the equation Aα = ∑ x as follows:
x∈A α
0.2 0.2 0.2 0.2 0.2
A0.2 ={ , , , , }
x1 x2 x3 x4, x5
0.4 0.4 0.4 0.4
A0.4 ={ , , , }
x2 x 3 x 4 , x 5
0.6 0.6 0.6
A0.6 ={ , , }
x3 x4 , x5
0.8 0.8
A0.8 ={ , }
x4, x5
1
A1={ }
x5
0.2 0.4 0.6 0.8 1
Thus, A= , , , ,
x1 x2 x 3 x 4 , x 5
Using the equation μ A = α ∈ [ 0,1 ] α ⋅ μ A (x ) and the fuzzy set A= {(
¿
α
( 0.2 , x 1) , ( 0.4 , x 2 ) , ( 0.6 , x3 ) , ( 0.8 , x 4 ) ,(1 , x5 )},
μ A ( x 1)= {0.2× 1,0.4× 0,0.6× 0,0.8×0,1 × 0 }=0.2 ¿
μ A ( x 1)= {0.2× 1,0.4× 1,0.6×0,0.8 ×0,1 × 0}=0.4 ¿
μ A ( x 1)= {0.2× 1,0.4× 1,0.6×1,0.8 ×0,1 ×0 }=0.6 ¿
μ A ( x 1)= {0.2× 1,0.4× 1,0.6×1,0.8 ×1,1 ×0 }=0.8 ¿
μ A ( x 1)= {0.2× 1,0.4× 1,0.6×1,0.8 ×1,1 ×1 }=1 ¿
¿
Thus, for a discrete fuzzy set, μ A = α ∈ [ 0,1 ] α ⋅ μ A (x )
¿
α
Q2A
Bellman and Zadeh (1970) proposed a fuzzy-based decision making model whereby goals and
constraints were expressed as fuzzy sets. In this case, decisions were based on the aggregation of
the sets. The model composed of a set of possible actions (A), a set of constraints (C j ¿¿ and a set
of goals Gi (i∈ N) . In this case, the goal is to determine a fuzzy set that satisfies both the goals
and constraints as defined by the following equation:
D ( a )=min ¿ where a ∈ A
Q2B
four available jobs (a, b, c, d) and four salaries (
f ( a )=30,000 , f ( b )=25,000 , f ( c )=20,000∧f ( d )=15,000 ¿ for each of the job category. There are
two constraints including job is interesting and within close driving distance. The constraints are
represented by:
C 1=\{ ( 0.4 / a ) , ( 0.6/b ) , ( 0.8 /c ) , ( 0.6/ d ) \} and C 2=\{ ( 0.1 / a ) , ( 0.9 /b ) , ( 0.7 /c ) , ( 1/d ) \}
For the fuzzy goal G, x is assumed to be a member of each of the salary function as follows:
G={ x ∈ { F ( a ) , F ( b ) , F ( c ) , F ( d ) } }
Applying the membership function for the fuzzy goal and substituting the salary functions,
G={0.875/a , 0.71875/b ; 0.5 /c ; 0.21875/d }
Based on Bellman and Zadeh’s fuzzy decision model, the decision is the aggregation of the
constraints and the goal given by:
D=G ∩C 1 ∩C 2
Given that D ( a )=min ¿, we can deduce that;
C 1 ∩C 2=min [ μc 1 , μc 2 ]={0.1/a , 0.6/b , 0.7/c , 0.6/d }
Then, D=min [ μ G ∩ [ μ c1 , μc 2 ] ]={0.1 /a ,0.6 /b , 0.5/c ,0.21875 /d }
Thus, b has the highest membership degree implying that b is the best job to select when
applying.
Q3
The IRIS data seen in the files "irisTrain.dat" and "irisTest.dat" can be used to generate a basic
Takagi-Sugeno fuzzy system automatically and then facilitating the training of the system users
by using the data in "irisTrain.dat". In training the ANFIS, one is required to combine the least-
squares and the back-propagation gradient descent algorithms. The two algorithms are applied to
investigate a function’s global maximum without knowing if it exists or where is located. The
algorithms works by first establishing a solution to a problem and then use it to look for an
alternative solution to make comparisons between the two to determine which is better than the
other and repeats the process until the user stops the process. So it is essential to set a number of
iterations for the algorithms. The ANFIS will use an automatic number of iterations of 10
implying the two algorithms will align the parameters of the FIS by 10 iterations which means
the data will be classified in each instance better. For instance, when applying the "irisTrain.dat",
the data is inputted in the FIS triggering the firing one rule for the FIS although the output is zero
instead of two, a value that is correct as it is the first example. Example two will then follow
implying the output value will also change. This is done until the tenth example is reached which
highlights a value of the training data that is well categorized as compared to the first attempt.
Using a large number of epoch to facilitate the classification of words is problematic as the FIS
will fail to classify new data for lack of randomness during the process of classification.
Therefore, ANFIS requires the setting up of other parameters including the rules for the
membership function among others to ensure the classification is correct. The ANFIS is best
used to develop fuzzy systems when there is a way to obtain measurements and their associated
classes and if the system to be built is complex.
Q4
Introduction
Good results have been obtained when applying the fuzzy logic type one to solve problems,
however fuzzy logic type two is the most preferred in solving linguistic uncertainties more
effectively (Mendel, 2007). The section explores the meaning, the need for the development of
fuzzy sets type two and finally the applications as well as issues associated with the practical
application of the said fuzzy sets.
Fuzzy sets type II
Since the membership function in the type II fuzzy sets is itself fuzzy, it then means that for each
element of the set, the membership grade is considered a type I fuzzy set (Hagras & Wagner,
2012; Azar, 2012). To represent the fuzzy sets type II in 2D, the footprint of uncertainty (FOU)
is always used. Fuzzy sets type II in this view are regarded to be fuzzier than type I sets (Wagner
& Hagras, 2010).
The need for development of fuzzy sets type II
Fuzzy sets type II are better than type one fuzzy sets since they are scientifically correct to model
words using such sets (Hagras & Wagner, 2012). As much as the type one fuzzy sets are certain,
the words are always uncertain during modeling. Therefore, there is need to develop type two
fuzzy sets since they can be used to model certainty of the words as opposed to type one sets
which are paradoxical and used to model uncertainty of the words.
Application of fuzzy logic type II and associated issues
The fuzzy sets are used for industrial control purposes, the control of ambient intelligent
surroundings, the control of the aircraft altitudes and control of the mobile robotics (Wu &
Mendel, 2011; John & Coupland, 2007). The fuzzy logic type II is used for control purposes
since the sensors used produce fuzzy results meaning they can be used to model the words
correctly to understand the output correctly. The use of fuzzy sets type II becomes fuzzier as
compared to the type one fuzzy sets implying they are computationally expensive (Castillo,
Melin, Kacprzyk & Pedrycz, 2007). When defuzzifying the control outputs, one needs an extra
step which is type reducing and is the one which makes the application of the sets
computationally expensive (John & Coupland, 2007).
Conclusion
Fuzzy type II logic is fuzzier as compared to type one fuzzy logic since the membership grade of
each Fuzzy type II is fuzzy type I itself. The increased fuzziness increases the computational
costs but increases it chances of being applied in modeling words.
References
Azar, A. T. (2012). Overview of type-2 fuzzy logic systems. International Journal of Fuzzy
System Applications, 2(4), 1-28.
Bellman, R. E., & Zadeh, L. A. (1970). Decision-making in a fuzzy environment. Management
science, 17(4), B-141.
Castillo, O., Melin, P., Kacprzyk, J., & Pedrycz, W. (2007, November). Type-2 fuzzy logic:
theory and applications. In Granular Computing, 2007. GRC 2007. IEEE International
Conference on (pp. 145-145). IEEE.
Hagras, H., & Wagner, C. (2012). Towards the wide spread use of type-2 fuzzy logic systems in
real world applications. IEEE Computational Intelligence Magazine, 7(3), 14-24.
John, R., & Coupland, S. (2007). Type-2 fuzzy logic: A historical view. IEEE computational
intelligence magazine, 2(1), 57-62.
Mendel, J. M. (2007). Type-2 fuzzy sets and systems: An overview [corrected reprint]. IEEE
computational intelligence magazine, 2(2), 20-29.
Wagner, C., & Hagras, H. (2010). Toward general type-2 fuzzy logic systems based on
zSlices. IEEE Transactions on Fuzzy Systems, 18(4), 637-660.
Wu, D., & Mendel, J. M. (2011). On the continuity of type-1 and interval type-2 fuzzy logic
systems. IEEE Transactions on Fuzzy Systems, 19(1), 179-192.
