100% found this document useful (1 vote)
138 views13 pages

Fuzzy Logic and Neural Networks - 4 - Solution

Uploaded by

Rakesh Pattanaik
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
100% found this document useful (1 vote)
138 views13 pages

Fuzzy Logic and Neural Networks - 4 - Solution

Uploaded by

Rakesh Pattanaik
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 13

NPTEL Online Certification Courses

Indian Institute of Technology Kharagpur

Course Name: Fuzzy Logic and Neural Networks


Assignment- Week 4
TYPE OF QUESTION: MCQ
Number of questions: 10 Total mark: 10 X 2 = 20
______________________________________________________________________________

To model input-output relationships of a process having two inputs and one output, let us use a
fuzzy reasoning tool with Mamdani approach. Let us also assume three linguistic terms, namely
Low (L), Medium (M) and High (H) to represent each of two inputs: I1, I2, and one output O. The
initial membership function distributions of the inputs and output are shown in Fig. A, which are
assumed to be symmetrical triangles.

Fig. A: Initial membership function distributions of the variables.


A binary-coded genetic algorithm (BCGA) will be used to evolve the optimized data base (DB)
and rule base (RB) of the fuzzy reasoning tool. The base-widths of the triangles are decided by
the b values. A typical GA-string is shown below, where five bits are used to represent each of
the b values and nine bits are utilized to represent the RB (where 1 and 0 represent the presence
and absence of the rules, respectively).
11001 10001 10101 100111101
b1 b2 b3 RB
NPTEL Online Certification Courses
Indian Institute of Technology Kharagpur

The RB of the fuzzy reasoning tool is shown below.


I2
L M H
L L L M
1 0 0
I1 M L M H
1 1 1
H M H H
1 0 1

Corresponding to the above GA-string, the modified RB is given below.


I2
L M H
L L
1
I1 M L M H
1 1 1
H M H
1 1

Take the ranges of b1, b2 and b3 as follows:


0.4 ≤ 𝑏1 ≤ 1.7
0.2 ≤ 𝑏2 ≤ 1.2
0.9 ≤ 𝑏3 ≤ 2.3
Let us consider a training scenario as follows: I1 = 1.5, I2 = 1.0, and target output T0 =
3.05.Answer the following five questions (Q1 to Q5)by considering the modified DB and RB,
according to the above GA-string. Use the Center of Area method for defuzzification.
NPTEL Online Certification Courses
Indian Institute of Technology Kharagpur

QUESTION 1:
I1 = 1.5 corresponds toL and M, respectively,with approximate membership values of
(a) 0.9333, 0.0667
(b) 0.2759, 0.7241
(c) 0.7241, 0.2759
(d) 0.0667, 0.9333
Correct Answer: c

Detailed Solution:
Consider the modified membership function distributions according to the GA-string:
0.4 ≤ 𝑏1 ≤ 1.7
0.2 ≤ 𝑏2 ≤ 1.2
0.9 ≤ 𝑏3 ≤ 2.3

Accordingly, modified data base for the input 𝐼1 would be as follows:


NPTEL Online Certification Courses
Indian Institute of Technology Kharagpur

𝐹𝑜𝑟 𝐼1

𝑆𝑜, 𝜇𝐿 = 𝟎. 𝟕𝟐𝟒𝟏 𝑎𝑛𝑑 𝜇𝑀 = 𝟎. 𝟐𝟕𝟓𝟗


______________________________________________________________

QUESTION 2:
I2 = 1.0 corresponds to M and H, respectively, with approximate membership values of
(a) 0.9333, 0.0667
(b) 0.2759, 0.7241
(c) 0.7241, 0.2759
(d) 0.0667, 0.9333
Correct Answer: a

Detailed Solution:
Modified data base for𝐼2
NPTEL Online Certification Courses
Indian Institute of Technology Kharagpur

𝑆𝑜, 𝜇𝑀 = 𝟎. 𝟗𝟑𝟑𝟑 𝑎𝑛𝑑 𝜇𝐻 = 𝟎. 𝟎𝟔𝟔𝟕


______________________________________________________________________________

QUESTION 3:
Q3. Firing strength of the first fired rule present in the rule base (i.e., if I1 is M AND I2 is M then
Output O is M) is approximately calculated as
(a) 0.9333
(b) 0.7241
(c) 0.0667
(d) 0.2759
Correct Answer: d

Detailed Solution:

Rules present according to the above GA-string are given below.


NPTEL Online Certification Courses
Indian Institute of Technology Kharagpur

I2
L M H
L L
1
I1 M L M H
1 1 1
H M H
1 1

Fired rules
L M L Absent
L H M Absent
M MM Present
M H H Present
NPTEL Online Certification Courses
Indian Institute of Technology Kharagpur

𝑆𝑜, 𝑡ℎ𝑒 𝐹𝑖𝑟𝑖𝑛𝑔 𝑆𝑡𝑟𝑒𝑛𝑔𝑡ℎ 𝑜𝑓 𝑡ℎ𝑖𝑠 𝑟𝑢𝑙𝑒 = 𝟎. 𝟐𝟕𝟓𝟗


_____________________________________________________________________________

QUESTION 4:
Firing strength of the second fired rule present in the rule base (i.e., if I1 is M AND I2is H then output O is
H) is approximately calculated as

(a) 0.9333
(b) 0.7241
(c) 0.0667
(d) 0.2759
Correct Answer: c
Detailed Solution:
NPTEL Online Certification Courses
Indian Institute of Technology Kharagpur

𝑆𝑜, 𝑡ℎ𝑒 𝐹𝑖𝑟𝑖𝑛𝑔 𝑆𝑡𝑟𝑒𝑛𝑔𝑡ℎ 𝑜𝑓 𝑡ℎ𝑖𝑠 𝑟𝑢𝑙𝑒 = 𝟎. 𝟎𝟔𝟔𝟕

QUESTION 5:
Crisp output corresponding to the set of inputs (I1=1.5, I2 =1.0) is found to be approximately equal to

(a) 2.96
(b) 5.95
(c) 8.25
(d) 3.87
Correct Answer: a

Detailed Solution:

𝐴𝑖 𝑎𝑛𝑑 𝐶𝑖 𝑑𝑒𝑛𝑜𝑡𝑒𝑠 𝑎𝑟𝑒𝑎 𝑎𝑛𝑑 𝑐𝑒𝑛𝑡𝑒𝑟 𝑜𝑓 𝑎𝑟𝑒𝑎 𝑜𝑓 𝑡ℎ𝑒 𝑠ℎ𝑎𝑑𝑒𝑑 𝑟𝑒𝑔𝑖𝑜𝑛 𝑓𝑜𝑟 𝑖 𝑡ℎ 𝑓𝑖𝑟𝑒𝑑 𝑟𝑢𝑙𝑒 ∶
NPTEL Online Certification Courses
Indian Institute of Technology Kharagpur
NPTEL Online Certification Courses
Indian Institute of Technology Kharagpur
NPTEL Online Certification Courses
Indian Institute of Technology Kharagpur

The Crisp Output would be 2.96

___________________________________________________________________________

QUESTION 6:
Select the correct order of operations used in the working cycle of a Genetic Algorithm (GA)
after the population initialization of the random solutions.
(a) Reproduction, Assigning Fitness, Mutation, Crossover
(b) Assigning Fitness, Reproduction, Crossover, Mutation
(c) Assigning Fitness, Reproduction, Mutation, Crossover
(d) Crossover, Assigning Fitness, Reproduction, Mutation

Correct Answer: b

Detailed Solution:

_________________________________________________________________________
NPTEL Online Certification Courses
Indian Institute of Technology Kharagpur

QUESTION 7:
An artificial neuron (𝑗 𝑡ℎ ) receives 𝑛 inputs 𝑥1 , 𝑥2, 𝑥3 , … … . . , 𝑥𝑛 with the weights
𝑤1𝑗 , 𝑤2𝑗 , 𝑤3𝑗 , … … … … , 𝑤𝑛𝑗 and bias b is attached to this neuron. Find the correct way to compute
theweighted sum of inputs to be passed on to an activation function to determine the output.

(a) ∑𝑛𝑖=1 𝑤𝑖𝑗 𝑏 + 𝑥𝑖


(b) ∑𝑛𝑖=1 𝑥𝑖 𝑤𝑖𝑗 + 𝑏
(c) ∑𝑛𝑖=1 𝑥𝑖 𝑏 + 𝑤𝑖𝑗
(d) ∑𝑛𝑖=1 𝑤𝑖𝑗 𝑥𝑖 𝑏

Correct Answer: b

______________________________________________________________________________

QUESTION 8:
Identify the nature of this activation function.
1
𝑂= , where𝑎 represents the coefficient of transfer function and 𝑢 denotes the input.
1+𝑒 −𝑎𝑢
(a) Hard Limit Transfer Function
(b) Linear Transfer Function
(c) Tan-Sigmoid Transfer Function
(d) Log-Sigmoid Transfer Function

Correct Answer: d

Detailed Solution:
______________________________________________________________________________

QUESTION 9:
Synapse of a biological neuron is copied in the artificial neuron in the form of
(a) connecting weights
(b) summing junction
(c) bias value
(d) activation/transfer function

Correct Answer: d

Detailed Solution:
NPTEL Online Certification Courses
Indian Institute of Technology Kharagpur

___________________________________________________________________

QUESTION 10:
The value of Tan-sigmoid transfer function used in Neural Networks varies in the range of
(a) (0.0, +1.0)
(b) (-1.0, +1.0)
(c) (-2.0, +2.0)
(d) (-3.0, +3.0)

Correct Answer: b

Detailed Solution:

______________________________________________________________________________

************END*******

You might also like