SASTRA DEEMED UNIVERSITY

UGC Act, 1956)

(A University under section 3 of the

End Semester Examinations

July 2022
Course Code: EIE106

NETWORKS
Course: FUZZY LOGIC & NEURAL

Duration: 3 hours
Question Paper No. :U0699

Max. Marks:100

PART-A
Answer all the questions 4x 20 80 Marks

the weights required to

1. (a) Using the Perceptron learning rule, find and
perform the classifications: vector 'x' (1 1), (-1 1) (1 -1)
are
is not
the members of the class (with a target value is 1); (-1 -1)
members of the class (with target value is -1). Use learning
rate
each
is 1; the threshold is 0.2 and starting weights are 0. Using
of the training 'x' vectors as input, test the response of the
net.
(10)
(b) Explain the Backpropagation algorithm with the architecture
diagram. (10)

(OR)

22. (a) Construct the Hebb neural network to train the two inputs AND,
OR gates and find the weights of it, assume the threshold is
equal to 0.2, include the bias. Use bipolar inputs and targets.
(15)
(b) Compare the various activation functions used in ANN. (5)
to
3 Consider a BAM network to map two simple letters (5x3 pattern)
the following target codes
 Target-1,1) Target=(1,1)

Find the weight matrix with input E and H, Test the response of the
net, and also test the response of the net with noisy patterm (1, 0)

(OR)
classes
4. Construct and test the LVQ with four vectors assigned to two
iteration
given below. Assume®.l and perform single
Vectors Class
(1100) 1
(0001) 2
(0011) 2
(1000)
(0110) 2

5. Two fuzzy relations are given by,

R o.7 .5s= o.9 0.6 0.21

0.8 Lo.1 .7 0.7 0.5
0.5
Find ROS using Max-min and Max-product compositions.

(OR)
certain changes in
6. For flight simulator data the determination of the basis of hard
conditions of the aircraft is made on
operating the
A fuzzy setto represent
breakpoints in the Match region. a second
condition of "near*' has a Match number of 0.74. Further,
*in the region of
fuzzy set is defined to represent the condition of Match
has a Match number of 0.74. In typical simulation data, a
number of 0.74 is a hard breakpoint.

2
 A Near match 0 74- 073 06 00
0715 0.74 0.745 0,75

B- n the regionof 0,4, 08 061

06
B-n matuh 00 74
of match 0.73 0,735 0.74 0.745 0.75)
Find out the following
a)AUB b)A B c)A d)B e) A B f)AnB g) AUB

7 Describe the defuzzification methods with necessary expressions.

(OR)
8. For the following fuzzy sets A and B defined on X
H(XI) xl 2xx4 x5 x6
A 0.1 0.6 0.80.9 0.7 0.1
B 0.9
0.70.5 0.2 0.1 0
Bxpress the following A-cutsets using Zadeh's notation:
a)A a7 b)B, c)(AUB)7 d)(A nB)os e)(AUA)o7
PART-B
Answer the following 1x 20 20 Marks

9. Diabetes mellitus or simply diabetes is a disease caused due to an

increased level of blood glucose. Various available traditional
methods for diagnosing diabetes are based on physical and chemical
tests. These methods can have errors due to different uncertainties.
Neural-based predictors will produce a better result based on tne

to predict
the
ne

previous experience, because of their ability

uncertainty. Diabetes Analyzer tor the
a Prediction and Diagnosis of the inputs and
ESgn for both
database. Assume binary
values (10)
5VCD
targets.

3
 Any onee
Diabetic/
Person Diet Exercise non
of the
diabetic
Parents
has
diabetic
Diabetic
active No
A Correct ratio between
fiber and
carbohydrates Yes
Diabetic
B Correct ratio between inactive
fiber and
carbohydrates Non-Diabetic
inactive No
C Eat too many grams
of carbohydrates Yes
Diabetic
active
D Corect ratio between
fiber and
carbohydrates Yes Non-
inactive
E Eat too many grams Diabetic
of carbohydrates the
inference system (FIS) that the FIS considers
(6) Desiign a fuzzy and food quality as inputs and
takes service
domain, which (10)
using the following rules.
computes tip
a percentage
is
food is rancid, then the tip
If the service is poor or the
cheap
If the service is good, then the tip is average.
. If the service is excellent or the food is delicious,
then the tip
1S generous.