B.E./B.Tech.

DEGREE EXAMINATION, NOVEMBER/DECEMBER 2016

Seventh Semester
EE 6006 – APPLIED SOFT COMPUTING

Answer All question

PART A – (10 x 2 = 20 marks)

1. Compare AN and biological Neuron.

2. What are the characteristics of neural networks and the elements that constitute the neural
networks?
3. Draw the typical architecture of competitive layer of neural network.
4. Distinguish between chance and fuzziness.
5. Explain the De Morgan’s principle of set operations.
6. Define fuzzification.
7. Sketch the closed-loop fuzzy control situation.
8. Give the advantages of fuzzy control situation.
9. How does the penalty change in the dynamic penalty method strategy?
10. List the various methods to select the best chromosomes.

PART B – (5 x 16 = 80 marks)

11. (a) Draw and explain the BPN. (16)

Or
(b) Write short notes on the following : (4 x 4)
(i) Signal processing
(ii) Pattern recognition
(iii) Speech recognition
(iv) Speech production.
12. (a) Describe the discrete time Hopfield network model of associative memories. (16)
Or
(b) List the applications of neural networks and discuss any two in detail. (16)

13. (a) Deliberate the methods of defuzzifying the fuzzy output functions given below:
(4 x4)
(i) Max membership principle
(ii) Centroid method
(iii) Weighted average method
(iv) Mean max membership.
Or
(b) What do you understand by classical sets and fuzzy sets? Elaborate on the
operations that can be performed on them. (16)
14. (a) Show how flexibility and accuracy can be implemented in aircraft landing control
problem with suitable diagrams. (16)
 Or
(b) (i) With a block flow diagram, describe the classical feedback control system.
Justify with an example. (8)
(ii) Write a brief note on classical PID control. (8)

15. (a) (i) Explain crossover and mutation operators for genetic algorithms.
(ii) Write the genetic algorithm applied to economic dispatch.
OR
(b) (i) Show how genetic algorithms are applied to unit commitment problems.
(ii) Compare binary and real coding GA.
 E.E./B.Tech. DEGREE EXAMINATION, NOVEMBER/DECEMBER 2017
Seventh Semester
EE 6006-APPLIED SOFT COMPUTING

(Regulations 2013)

Part A (10  2  20 Marks)

1. List the major applications in which Neural Networks have been used.
2. Recall the structure of a simple neuron and give its activation function for logistic
sigmoid function.
3. Differentiate in between supervised and unsupervised learning.
4. Describe the operations on fuzzy relations.
5. What is fuzzy logic based control system?
6. Mention some applications of fuzzy logic.
7. Discuss the assumptions in a fuzzy control system design.
8. Memorize the workflow of a simple fuzzy logic control system.
9. Define fuzzification.
10. Describe the term “Mutation” in genetic algorithm.
Part B (5 16  80 Marks)
11. a) Illustrate the working mechanism of single layered and multilayered neural network with
an architectural diagram. Explain the competitive layer with the mechanism of setting
weights over the network.
(OR)
b) Illustrate the algorithmic work flow of an un-supervised learning algorithm with its
inputs, outputs and measures for evaluation.

12. a) Demonstrate the working paradigm of Hopfield neural network with clear architectural
model. Compute the following for the training phase of the network with its periodic weight
updates.
o Binary input patterns
o Bipolar input patterns
Depict clearly about the energy function evaluation over a stables condition.
(OR)
b) Demonstrate the workflow model of neural networks for control of inverted pendulum.
Illustrate its applicability by a mathematical model with parameters such as mass of
pendulum, mass of cart, length of the bar, standard gravity and moment of inertia of the bar.
Draw the free body diagrams for the cart and the pendulum.

13. a) Five separate regions along the san Andreas fault in California have suffered damage
from a recent earthquake. For purposes of assessing payouts from insurance companies to
 building owners, the five regions must be classified as to their damage levels. Expression of
the damage in terms of relations will prove helpful. Surveys are conducted of the buildings
in each region. All the buildings in each region are described as being in one of three
damage states: no damages, medium damage and serious damage. Each region has each of
these three damage states expressed as a percentage (ratio) of the total number of buildings.
Hence, for this problem n=5 and m=3. The following table summarizes the findings of the
survey team:

Regions X1 X2 X3 X4 X5
X i1  ratio with no damage 0.3 0.2 0.1 0.7 0.4
X i 2  ratio with medium damage 0.6 0.4 0.6 0.2 0.6
X i 3  ratio with serious damage 0.1 0.4 0.3 0.1 0.0

Discover the fuzzy relations method used up for the above data. Compute the fuzzy
relationship and tolerance relation.
(OR)
b) One of the decisions your project team faces with each new computer product is what type of
Printed Circuit Board (PCB) will be required for the unit. Depending on the density of
tracks (metal interconnect traces on the PCB that act like wire to connect components
together), which is related to the density of the components, we may use a single-layer PCB,
a double-layer PCB, a four-layer PCB, or a six-layer PCB. A PCB layer is a two-
dimensional plane of interconnecting tracks. The number of layers on a PCB is the number
of parallel intercommctions in the design, the greater the number of layers required to fit the
design onto a PCB of given size. One measure of board track density is the number of nodes
required in the design. A node is created at a location in the circuit where two or more lines
(wires, tracks) meet.
Investigate the following for proper decision making logic over the described system.
1) Define the fuzzy states of nature.
2) Calculate fuzzy alternatives.
3) Illustrate orthogonal fuzzy information system.
4) Show the prior probabilities.
5) Use new data samples for computation.
6) Produce the utility values.

14. a) Demonstrate the mechanism of fuzzy rule based Aircraft Landing System with its fuzzy
conditions.
 (OR)
b) Illustrate the concept of Speed of DC Motor using Fuzzy PID Controller.

15. a)Examine the steps involved in designing simple Genetic Algorithm by explaining
population of binary strings, control parameters, fitness function, genetic operators, and
selection mechanism with encoding.
(OR)
b) Differentiate the terms gradient and non-gradient searches with accordance to the
algorithmic workflow, representation in addition to time and space complexity. Give
examples for each category.
 B.E./B.Tech.DEGREE EXAMINATION, APRIL/MAY 2018
Seventh semester
EE6006 – APPLIED SOFT COMPUTING
(Regulations 2013)
Time : Three Hours Maximum : 100 Marks
Answer ALL questions

PART – A (10 2=20)Marks)

1. Define : supervised learning . Give an example.
2. Recall the architectural components of a neural network .
3. List the applications of artificial neural network.
4. What is activation function in NN ?
5. What is FAM Table ?
6. Define defuzzification .
7. Mention the key role of knowledge based systems .
8. Recall the assumption in a Fuzzy control system Design .
9. State Darwins theory of evolution.
10. Define : Cross over .

PART-B (5 16=80 Marks)

11.a) i) Compare Bio neurons .with artificial neuron (8)
ii) Compare supervised learning with unsupervised learning . (8)
(OR)
b) Examine the working principle of feed forward neural network and compare the activation
used with other network models . Mention the error function and techniques used up to
adjust the Weights over the network. (16)

12.a) llustrate the working principle of Hebbian learning rule for Hopfield networks. Draw
and explain the architecture of discrete time Hopfield networks. (16)
. (OR)
b) Demonstrate the workflow model of neural networks for control of Inverted pendulum .
Illustrate its applicability by a mathematical model with parameters such as mass of
pendulum , mass of , cart , length of the bar , standard gravity and moment of inertia of the
bar.Draw the free body diagrams for the cart and the pendulum . (16)

13. a) In metallurgy materials are made with mixtures of various metals and other elements to
achieve certain desirable properties . In a particular preparation of steel , three elements ,
namely iron ,manganese and carbon , are mixed in two different proportions. The samples
obtained from these two different proportions are placed on a normalized scale , as shown in
the figure and the represented as fuzzy sets A1 and A2 .You are interested in finding some
 sort of “ average “ steel proportion . For the logical union of the membership functions
shown we want to find the defuzzified quantity . For each of the methods learned assess
(i)whether each is applicable and if so,ii) calculate the defuzzified value z* (16)

b) List and explain the different methods of assigning membership functions. (16)

14. a ) Discuss briefly about the fuzzy rule base for the liquid level control system with a fuzzy
rule function condition. (16)
(OR)
b) Let A = {( X1,0.2) ,( X2 ,0.7), (X3,0.4) }and B={ (Y1 , 0.8),(Y2 0.6)} be two fuzzy sets
defined on the universe of discourse X={X1,X2,X3 } and Y={Y1,Y2 } respectively . find
Cartesian product of A and B union ,intersection of A and B , algebraic sum of A and B
and algebraic product of A and B. (16)

15. a) Examine the steps involved in designing simple Genetic Algorithm by explaining
population of binary strings, fitness function , cross over and mutation operators .
(16)
(OR)
b ) Explain the step by step procedure for solving economic dispatch problems using genetic
algorithms. (16)