Roll No. B.E/ B.Tech (Fulltime) Degreeend Semesterexaminations, April/May2013

Roll No.
B . E / B . T e c h ( Full T i m e ) D E G R E E E N D S E M E S T E R E X A M I N A T I O N S , April / May
2013
B.Tech IT
VIII Semester
IT9035 Soft Computing
(Regulation ...2008
Max. M a r k s 100
A n s w e r A L L Questions
T i m e : 3 Hours
P A R T - A (10 x 2 = 20 M a r k s )
1
2-
I
I
What is the difference between the classic proposition and fuzzy proposition.
Let A be a fuzzy set defined by:
A = 0 . 5 / x + 0 . - 4 / x + 0 . 7 / x + 0 . 8 / x + l/.x
1
. Find a-cuts and
strong a-cuts of A
3-
Give an example from daily life of any two fuzzy propositions and express in canonical form.
4.
For the fuzzy representation below, find

1
0.7
0.3
0.2
0.5
Af,=
M, M
0.6
0.6
0.6
0.1
and explain the significance in the real world.
5.
Differentiate supervised and unsupervised network, giving an example for each.
6.
Write the equation for perceptron and explain the significance of this equation.
7.
Write the four activation functions of neurons.
8.
You are given the task of identifying human gestures computationally. W h i c h neuron learning
model will you use? W h y ?
9.
Write any one equation of the kernel function and explain the terms.
10
Write down atleast one unique application area for each of the following learning process (i)
Fuzzy logic (or) fuzzy control process (ii) Neural Network (iii) Genetic Algorithm and <iv) Support
Vector machines.
P a r t - B ( 5 x 16 = 80 m a r k s )
11.
i)
Four
steps of
Hebbian
learning
implemented starting with
of a single-neuron
w'=[1-1]'
network
for learning constant
have
been
c-l
using
inputs as follows:
1
-2
ii)
x =
2
0
x
\ 1
1
1
Draw the neuron model for this problem and find t h e final weights after four
steps for bipolar continuous f{net), \= 1 (8)
Describe a neural network that could be used to control the steering o f a robotic
car o n a race track. H o w c a n it be tracked? (8)
i)
With an application of your choice explain t h e various stages of Fuzzy

Controller. Include the blockdiagram, fuzzy sets, membership functions that
are being decided upon, Fuzzy rule base, the type of inference that is being
carried out, and the defuzzification process . (10)
ii)
With a n example of your choice, explain t h e basis of calculating fuzzy
relations and projections. (6)
OR
i) Let A, B be two fuzzy n u m b e r s whose membership functions are given by
A ( x ) = (x+2)/2 for - 2 < x < = 0 , (2-x)/2 for 0 < x < 2 , 0 otherwise
B ( x ) = ( x - 2 ) / 2 for 2 < x < 4 , (6-x)/2 for"<x<=6, 0-otherwise.
Calculate the f u z z y numbers A + B , A - 8 , 6 - A, A * 8 , A / 6 , Min(A,fi) a n d
Max(A.B). (10)
ii) Let the sets o f values of variables X and Y b e X = { x l , x2 , x 3 } a n d
Y = { y l , y 2 }, respectively. If A = 0 . 6 / x l + i / x 2 + 0.9/x3 and A - 0 . 5 / x l +
0.9/x2 + l / x 3 , find the conclusion B' b y using m o d u s p o n e n s if 6 = 1 / y l +
0 . 4 / y 2 . (6)
i)
ii)
i)
ii)
With an example, explain the procedure of a single discrete perc-eptron in

classifying a linearly separable problem (6)
With an appropriate diagram, describe the classification of logical A N D , OR
and X O R (10)
OR
Illustrate h o w a linear classifier works geometrically. What difference does
it make, w h e n the s a m e problem is applied to a neural network, w h y ? H o w
are non linearly separable patterns handled in neural networks? ( 6 )
Prototype points are given as:
x = [5l]',x =[l3],
1
x =[32],x =[54]'-.Class
x =[00]',x =[-l-3]',x =[-23j',x =[-30] :C/a.w2

f
a)
Determine
i f
t h e t w o c l a s s e s of patterns are linearly
separable. (2)
b) Determine the center of gravity for patterns of each class, and find and
draw the decision surface in pattern space (4)
c) Using the equation from the decision surface, find the weight vectors and
design a dichotomizer. H o w would it recognize the following input patterns:
JC = [4
i)
ii)
2]'
JC=[0
5]'
JC=[36/13
Of
( )
4
W h y do w e need to calculate the gradient in the back propagation

algorithm? In a feed forward neural network trained by back propagation,
what is back propagated and from where to where? Explain with the help of
the diagram mathematically. W h a t role d o e s the steepness function in this
learning process? (-8)
(a) A r e there a n y relation between the number of neurons and the type of
classification? Explain. (4)
(b) W h a t are t h e advantages a n d disadvantages of neural network
classification s c h e m e s ? Which type of problems neural network is N O T
suited for? (4)
i) A two layer neural network has two neurons in each layer, three inputs,
including an augmentation input and two outputs. With back propogation
algorithm, how will you find weights, slopes, error signals, outputs and the
updated weight. (8)
ii) How d o e s neurons act as auto associative m e m o r y ? Explain the
functioning with a three neuron model. (8)
15:
a)
b)
i)
ii)
i ) Describe the terms, crossover rate, mutation, reproduction, Roulette

wheel selection and Fitness function in Genetic A l g o r i t h m .
(6)
ii) You are given digits 0 to 9, and operators +, -, * and /. Using Genetic
algorithm h o w will you represent the target positive integer.
Design the whole optimization process and the type of optimization function
used
(10)
OR
Explain the basis of classification in S V M (6)
H o w does SVM classify non-linearly separable classes? Explain it through
the kernel functions. W h e r e will you apply this classification? Explain with
a n example (10)

Roll No. B.E/ B.Tech (Fulltime) Degreeend Semesterexaminations, April/May2013

Uploaded by

Copyright:

Available Formats

Roll No. B.E/ B.Tech (Fulltime) Degreeend Semesterexaminations, April/May2013

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Roll No. B.E/ B.Tech (Fulltime) Degreeend Semesterexaminations, April/May2013

Uploaded by

Copyright:

Available Formats

Roll No.

B . E / B . T e c h ( Full T i m e ) D E G R E E E N D S E M E S T E R E X A M I N A T I O N S , April / May

. Find a-cuts and

For the fuzzy representation below, find

and explain the significance in the real world.

Differentiate supervised and unsupervised network, giving an example for each.

Write the four activation functions of neurons.

implemented starting with

for learning constant

With an application of your choice explain t h e various stages of Fuzzy

With an example, explain the procedure of a single discrete perc-eptron in

x =[00]',x =[-l-3]',x =[-23j',x =[-30] :C/a.w2

W h y do w e need to calculate the gradient in the back propagation

i ) Describe the terms, crossover rate, mutation, reproduction, Roulette

You might also like