CHANDIGARH UNIVERSITY

Gharuan, Mohali
Institute/Department: UNIVERSITY INSTITUTE OF ENGINEERING
División: BLOCK 1
Subject Name: Artificial Neural Network
Subject Code: CSA 403/ITA-403

Assignment No.: 1
Max. Marks: 12

Date of Allotment: 24-Jul-19

Last date of Submission: 13-Aug-19

Course Outcomes:
CO Title Level
Number

CO1 To be able to understand the domain of Artificial neural network Remember

CO2 To be able to understand various problem solving techniques of Understand

artificial neural network
CO3 To use the Model algorithms and performing the ANN paradigms Understand

CO4 To implements on the basis of all the machine learning paradigms Understand
and techniques
CO5 Compute all the ANN algorithms on the basis of performances Understand

CO6 To compute real life problems like supervised and unsupervised Understand

ASSIGNMENT SET 1

Questions:
Sr. No. Question CO Number

1 Q: What is artificial neural network? Explain with example CO2

 2 Explain the limitations of perceptron learning algorithm. CO1

3 What is activation function? Describe different types of activation CO5

function with diagram.

4 . Explain the following terms: (3) CO1

a. Feed forward network
b. Activation and synaptic dynamics
c. Weights

ASSIGNMENT SET 2

Questions:
Sr. No. Question CO Number

1 Compare biological neuron with artificial neuron CO2

2 Describe some applications of ANN CO1

3 . What do you mean by supervised learning? Explain with example CO3

4 . Explain the models of neuron. . CO5

ASSIGNMENT SET 3

Sr. No. Question CO

Numbe
r
1 What is Adeline learning algorithm with diagram?
2 do you mean by McCulloch Pitts neuron? Explain with diagram CO3

3 Explain BPN algorithm. CO4

4 Illustrate the characteristics of ANN terminology. CO5

ASSIGNMENT SET 4

Sr. No. Question CO Number

1 What do you mean by role of error in adaline learning algorithm? CO1

2 What do you understand by learning rate? CO2

3 What is biological neural network? Explain with diagram and CO3

parts.
4
What do you mean by basic learning laws? Explain with example CO4

Set 5

Sr. No. Question CO

Numbe
r

1 What is the effect of Learning Rate ‘ἠ? CO1

2 Compare LMS,perceptron and delta learning laws

3 What are the variations in Back Propagation Networks ? CO3
4 Explain the influence of the momentum on weight change by taking CO4
a example
Set 6

Sr. No. Question CO Number

1 Describe some attractive features of the biological neural network CO1

that make it superior to the most sophisticated Artificial
Intelligence computer system for pattern recognition tasks.

2 What are the three differences among the three models of artficial CO2
neuron, namely, McCulloch-
Pitt, perceptron and adaline?
Explain the significance of the initial values of weights and the
learning rate parameter in the
seven basic learning laws.
4 How tuning parameters effect back propagation neural network CO5

Set 7

Sr. No. Question CO Number

1 What is meant by topology of artificial neural networks? Give a

few basic topological structure of artificial neural networks?

2 How can you compare Performanceof Computer and Biological CO2

Neural Networks

3 What are the key developments in ANN CO3

4 What are the research directions in the area of Back CO4

Propagation Network

Set 8

Sr. No. Question CO Number

 1 The symmetric sigmoid is defined ast(x) = 2s(x)−1, wheres(·) is CO2
theusual sigmoid function. Find the new expressions for the
weight correctionsin a layered network in which the nodes usetas
a primitive function

2 Implement the backpropagation algorithm and train a network CO6

that computes the parity of 5 input bits.

3 McCulloch Pitts neuron? Explain with diagram CO3

4 Different parts of biological neuron with diagram. CO4

Set 9

Sr. No. Question CO Number

1 Compute the output of M-P neuron with input 0.33,0.45 and CO1
0.63.The respective weights are 3,2 and 4. Use binary sigmoid
activation function with steepness parameter to be 0.5.

2 Distinction between learning equation and learning law? CO2

3 Different types of activation function with diagram? CO3

4 . Topology of artificial neural networks? Give a few basic CO4

topological structure of artificial neural network?
Set 10

Sr. No. Question CO Number

1 Research directions in the area of Back Propagation Network?. CO1

2 The three differences among the three models of artficial neuron, CO2
namely , McCulloch- Pitt, perceptron and adaline?

3 The significance of the initial values of weights and the CO3

learning rate parameter in the seven basic learning laws.

4 How tuning parameters effect back propagation neural

network?

Set 11

Sr. No. Question CO Number

1 The influence of the momentum on weight change by taking a CO1

example

2 Activation function? Describe different types of activation CO2

function with diagram

3 Describe some attractive features of the biological neural network CO3

that make it superior to the most sophisticated Artificial
Intelligence computer system for pattern recognition tasks.
 4 Describe some attractive features of the biological neural CO4
network that make it superior to the most sophisticated
Artificial Intelligence computer system for pattern recognition
tasks

Set 12

Sr. No. Question CO Number

1 The effect of Learning Rate ‘ἠ? CO1

2 Give two examples of pattern recognition tasks to illustrate the CO2

superiority of the biological neural network over a conventional
computer system

3 Compare the performance of a computer and that of a biological CO3

neural network in terms of speed of processing,size and
complexity,storage,fault tolerance and control mechanism

4 Additive and shunting models of activation models CO4

CHANDIGARH UNIVERSITY
Gharuan, Mohali
Institute/Department: UIET
Division: BLOCK 1
Subject Name: Artificial Neural Network
Subject Code: CSA403/ITA403

Assignment No.: 2
Max. Marks: 12

Date of Allotment: 4/10/19

Last date of Submission: 11/11/19

3 Marks Each

Course Outcomes:
CO Title Level
Number

CO1 To be able to understand the domain of Machine Learning Remember

CO2 To be able to understand various problem solving techniques of Understand

Machine Learning
CO3 To use the Model algorithms and performing the machine learning Understand
paradigms
CO4 To implements on the basis of all the machine learning paradigms Understand
and techniques
CO5 Compute all the ML algorithms on the basis of performances Understand

CO6 To compute real life problems like supervised and unsupervised Understand

Set 1

Sr. No. Question CO Number

 1 Explain stochastic networks. CO1

2 What do you mean by Synaptic dynamics model? CO2

3 Differentiate between Activation and Synaptic Dynamics Models CO3

4 Explain Hebbian learning law CO4

Set 2

Sr. No. Question CO Number

1 What are the different categories of learning? CO1

2 What is the basis for Boltzmann learning law? CO2

3 CO3
What do you mean by Feedforward neural network?

4 CO4

Explain simulated annealing?

Set 3

Sr. Question CO Number

No.

1 Explain what is meant by feature mapping? CO1

2 What are the limitations of the Boltzmann learning? CO2

3 Explain Activation dynamics model. CO3

4 What do you mean by Feedback neural network? Explain with example. CO4

Set 4

Sr. No. Question CO Number

1 What do you mean by perceptron learning? CO1

2 What do you mean by pattern association? CO2

3 Explain pattern classification? CO3

4 CO4
What do you mean by artificial neural network?

Set 5

Sr. No. Question CO Number

1 What is meant by each of the following: CO4

a) Equilibrium state
b) Stable states

2 What is meant by global pattern formation in neural CO2

networks?
3 Discuss the behaviour of stationary probability distributions of the CO3
states atdifferent temperatures in relation to the energy landscape

4 how to derive the state transition diagram for a stochastic neural CO4
network

Set 6

Sr. No. Question CO Number

1 Explain the following: CO4

a) Autonomous and nonautonomous dynamical systems
b) Additive and shunting models of activation models

2 What are the different categories of learning? example. CO3

3 CO2
What does the rank of an input matrix indicate?

4 What is the significance of the Widrow's learning for linear CO4

associative networks?

Set 7

Sr. No. Question CO Number

1 What is meant by each of the following: CO4

(a) Transient state
(b) Steady state

2 What are the requirements of learning laws for effective CO2

implementation?
3 Explain the difference between short-term memory and long-terms CO3
memory with reference to dynamics models.
4 Show the probability function for update of a neuron for different CO3
temperatures. Explain the significance of the temperature
parameter..

Set 8

Sr. No. Question CO Number

1 How to derive the state transition diagram for a stochastic neural CO5
network.

2 Bouncing ball analogy for the dynamics of a stochastic neural CO3

network. Elaborate

3 Elaborate feedforword network with example CO4

4 Global pattern formation in neural networks CO1

Set 9

Sr. No. Question CO Number

1 The different categories of learning? CO1

2 The rank of an input matrix indicate? CO2

3 The requirements of learning laws for effective implementation? CO3

4 Hebb’s law steps with detail CO4

Set 10

Sr. No. Question CO Number

1 Generalized Delta Rule: Backpropagation learning CO1

2 Different categories of learning? CO2

3 Differentiate between Activation and Synaptic Dynamics Models CO3

4 CO4
4. Basic Artificial Neural Network Models for Pattern Recognition
Problems in detail.

Set 11

Sr. No. Question CO Number

1 Hopfield model of neural network in detail CO1

2 Boltzmann Machine? Explain Boltzmann learning law of neural CO2

network

3 Supervised, reinforcement and unsupervised learning? Explain CO3

in detail with example

4 The requirements of learning laws for effective CO4

implementation?

Set 12

Sr. No. Question CO Number

1 CO4
Consider the representation of each of the ten digits (0, 1, ..., 9) by
a matrix of 10 x 10 elements, where each element is either 1 or -1.
Design a Hopfield network of 100 units to store these digits. Study
the performance of the network for recall of digits if 10% of the
elements are randomly switched.

2 Limitations of the Boltzmann learning/? CO2

3 Requirements of learning laws for effective implementation? CO3

4 Show the probability function for update of a neuron for different CO1
temperatures. Explain the significance of the temperature
parameter
CHANDIGARH UNIVERSITY
Gharuan, Mohali
Institute/Department: UIET
Division: BLOCK 1
Subject Name: Artificial Neural Network
Subject Code: CSA403/ITA403

Assignment No.: 3
Max. Marks: 12

Date of Allotment: 22/10 /19

Last date of Submission: 11/11/19

Course Outcomes:
CO Title Level
Number

CO1 To understand the nature of directed and undirected graphs Remember

CO2 To construct binary search trees and calculate the nodes in it. Understand

CO3 To implement BFS AND DFS graphs and to know their Understand
implementation for storing data
CO4 To learn pre order and post order traversals in binary trees Understand

CO5 To understand the data structure using heaps and insertions of Understand
elements in heap
CO6 To write recursive procedures and algorithms for given Understand
implementation

Set 1

Sr. No. Question CO Number

ONE MARKS QUESTIONS

1 What are the components of a competitive learning network? CO4

2 What is an instar network? CO2

FIVE MARKS QUESTIONS

3 Distinguish between content-addressable and address-address- able CO3

memories
4 What is an associative memory? CO1

Set 2

Sr. No. Question CO Number

ONE MARKS QUESTIONS

1 What is meant by on-centre off-smound feedback network? CO4

2 What is a pattern clustering network? CO1

5 MARKS QUESTIONS

3 Distinguish between heteroassociative and autoassociative CO2

memories

reduction

4 What is a linear association? What are its limitations as an associative CO4

memory?

Set 3
 Sr. No. Question CO Number

ONE MARKS QUESTIONS

1 Explain the difference between pattern clustering and feature mapping

2 What is adaptive vector quantization? What is learning vector CO2

quantization?

FIVE MARKS QUESTIONS

2 Illustrate the concept of feature mapping with the help of an example of CO6
mapping 2-D input onto a 2-D feature space.

3 What are some direct applications of the' principles of neural CO1

networks? Why are they called 'direct' applications?

Set 4

Sr. No. Question CO Number

ONE MARKS QUESTIONS

1 Explain the feature mapping of 2D input onto 1-D feature space. CO1

5 MARKS QUESTIONS

2 Explain the noise suppression property of the quadratic output function in CO2
the feedback layer of a competitive learning network

3 Discuss the performance of a competitive learning network for CO3

semilinear output function in the feedback layer

Set 5

Sr. No. Question CO Number

 FOUR MARKS QUESTIONS

1 What are the salient features of the Kohonen's self-organizing CO1

learning algorithm?

2 Explain the build up of the 1-D feature map of 1-D input values selected at CO2
random from an interval 0 to 1.

3 Explain the distinction between CO3

(a) pattern association and pattern classification tasks.
(b) pattern classification and function approximation tasks.

Set 6

Sr. No. Question CO Number

FOUR MARKS QUESTIONS

1 Find the eigenvedor corresponding to the first principal CO2

component of the following correlation matrix of a stochastic
process

stochastic process

2 Determine the self-organizing map generated by points selected at CO5

random from an annular ring formed by
two concentric circles. Consider the following two cases:
(a) The units in the SOM are arranged in a 2-dimensional plane.
(b) The units in the SOM are arranged in a 1-D layer.

3 Prove that BAM for binary or bipolar units is stable for CO6
asynchronous update of units.
 Set 7

Sr. No. Question CO Number

FOUR MARKS QUESTIONS

1 The following is the cross-correlation matrix of two stochastic CO5

processes

Find the left and the right eigen vectors corresponding to the
asymmetric principal component
2 CO2
Show that the Linsker's rule can be defined by computing average of
descent learning ∆wi = - n∂E/∂wi
on the cost function
3 CO4
Show that the Linsker's rule can be defined by computing average of
descent learning ∆wi = - n∂E/∂wi
on the cost function

Set 8
Sr. No. Question CO Number

One marks questions

1 Neocognitron – Recognition of handwritten characters CO1

5 marks questions

2 The components of a competitive learning network? CO2

3 Principal components of an autocorrelation matrix? CO3

Set 9

Sr. Question CO Number

No.

TWO MARKS QUESTION(1+1=2)

1 The difference between pattern clustering and feature mapping CO4

5 MARKS QUESTIONS

2 Cluster all the 5-bit binary vectors except the all zero vector, .10 CO2
0 . . . 0IT, using ART1 algorithm. Study the effect of vigilance
parameter on the resulting clusters
3 Illustrate the concept of feature mapping with the help of an CO1
example of mapping 2-D input onto a 2-D feature space.
 Set 10

Sr. Question CO
No. Numbe
r

FOUR MARKS QUESTIONS

1 Bidirectional Associative Memory? What is meant by 'BAM is unconditionally CO1

stable'?
2 The noise suppression property of the quadratic output function in the feedback CO3
layer of a competitive learning network

3 Discuss the performance of a competitive learning network for semilinear output CO5
function in the feedback layer

Set 11

Sr. No. Question CO Number

Four Marks Questions

1 Why MAM will have superior performance over BAM for pattern CO1
retrieval

2 Pattern association and pattern classification tasks CO2

3 Discuss a neural network model for energy minimization in a texture CO3

classification problem

Set 12

Sr. Question CO
No. Number
 Four Marks Questions

1 Pattern classification and function approximation task CO1

Q:

2 Discuss the application of neural network principles for decision making CO3

3 Construcl-dlll example to show that pattern recall is superior in a CO2

tridirectional associative memory compared to a bidirectional associative
memory