2018-2019

ARTIFICIAL NEURAL
NETWORKS
ISPE20

Dr. RUDRAMURTHY M S
Associate Professor,
Department of Information Science & Engineering

[ QUESTION BANK ]
Siddaganga Institute of Technology Tumkur 03
Department of Information Science and Engineering

1
 SIDDAGANGA INSTITUTE OF TECHNOLOGY-TUMAKURU-572103
DEPARTMENT OF INFORMATION SCIENCE AND ENGINEERING

QUESTION BANK – ISE20-ARTIFICIAL NEURAL NETWORKS

Total Hrs: 39 SEE: 100 Marks

Teaching Hrs: 3/Week CIE: 50 Marks
Course Coordinator: Dr. RUDRAMURTHY M S

UNIT-I

EACH OF THE FOLLOWING QUESTIONS CARRIES TWO MARKS

Q.Define Neural Networks?

[L1, CO1:PO1] 2M
Q. Define Neuromorphic Computing.
[L1, CO1:PO1] 2M
Q. What are dentrites?
[L1, CO1:PO1] 2M
Q. What is the function of dentrite?
[L1, CO1:PO1] 2M
Q. What is axon? Give its function.
[L1, CO1:PO1] 2M
Q. What is short-term memory in neural networks?
[L1, CO1:PO1] 2M
Q. What is Long-term memory in neural networks?
[L1, CO1:PO1] 2M
Q. What are learning laws in neural networks?
[L1, CO1:PO1] 2M
Q. List the different learning laws in neural networks.
[L1, CO1:PO1] 2M
Q. What are the requirements of learning laws?
[L1, CO1:PO1] 2M
Q. What is learning rate parameter?
[L1, CO1:PO1] 2M
Q. State the Hebbian learning law.
[L1, CO1:PO1] 2M
Q.What are the features of neural networks?
[L1, CO1:PO1] 2M
Q.What is stability-plasticity dilemma?
[L1, CO1:PO1] 2M
Q. What are the characteristics of logistic functions?
[L1, CO1:PO1] 2M
Q. Give the limitation of McCulloch Pitt’s neuron model.
[L1, CO1:PO1] 2M
Q. Give the limitation of Rosen Blatt’s perceptron neuron model.
[L1, CO1:PO1] 2M
Q. What is short-term memory in ANN?
[L2, CO1:PO1] 2M
Q. What is Long-term memory in ANN?
[L2, CO1:PO1] 2M
Q. Draw the structure of a biological neuron
[L1, CO1:PO1] 2M
Q. How is the net input calculated using matrix multiplication method?
[L2, CO1:PO1] 2M
2
 EACH OF THE FOLLOWING QUESTIONS CARRIES FOUR MARKS

Q. How Adaline neuron model is different from Mc Culloch Pitt’s and Rosen Blatt’s
perceptron model?
[L2, CO1:PO1] 4M
Q. Exemplify the motivation for ANN.
[L2, CO1:PO1] 4M
Q. Exemplify the significance of sigmoid function.
[L3, CO1:PO3] 4M
Q. For the network shown in figure, Calculate the net input to the output neuron Y.
[L3, CO1:PO2] 4M

Q. What is learning in ANN? List the important learning strategies in ANN.

[L1, CO1:PO1] 4 M
Q. Why sigmoid function is also called as squashing function?
[L2, CO1:PO1] 4M
Q. Give the terminology of Artificial Neural Networks.
[L2, CO1:PO1] 4M
Q. Draw the structure of a biological neuron and also, label its parts.
[L2, CO1:PO1] 4M
Q. State few activation functions which are used in single and multilayer networks to
calculate the output.
[L2, CO1:PO1] 4M
Q. What is the significance of weights used ANN.
[L2, CO1:PO1] 4M
Q. In what areas the artificial neural networks are used?
[L2, CO1:PO1] 4M

EACH OF THE FOLLOWING QUESTIONS CARRIES EIGHT MARKS

Q. Distinguish between Biological neural network and artificial neural networks.

[L2, CO1:PO1] 8M
Q. Exemplify linearly separable problem.
[L3, CO1:PO3] 8M
Q. Show the mapping and structural view point of ANN.
[L2, CO1:PO1] 8 M
Q. Discuss briefly Mc Culloch Pitt’s artificial neuron model. Give its limitations.
[L2, CO1:PO1] 8M
Q. Give the mathematical representation of Rosen Blatt’s perceptron neuron model.
[L2, CO1:PO1] 8M
Q. Give the mathematical representation of Widrow’s Adaline neuron model.
[L2, CO1:PO1] 8M
Q. Identify and illustrate different activation functions used in neural networks.
[L3, CO1:PO3] 8M
Q. Show the linearization of sigmoid function using Taylor’s series.
[L3, CO1:PO3] 8M
3
Q. List the important learning laws in ANN. Discuss briefly Hebbian learning.
[L2, CO1:PO1] 8M
Q. Graphically, sketch the different activation functions used in NN.
[L3, CO1:PO2] 8M
Q. Give the taxonomy of developments in Artificial Neural Networks.
[L2, CO1:PO1] 8M

EACH OF THE FOLLOWING QUESTIONS CARRIES TEN MARKS

Q. Show the graphical representation of sigmoid function and differentiate the sigmoid
function and comment on the result.
[L3, CO1:PO3] 10M
Q. Show that the sigmoid function is related to other mathematical functions.
[L3, CO1:PO3] 10M
Q. Obtain the output of the neuron Y for the network shown in figure using activation
function as binary sigmoidal and Bipolar sigmoidal.
[L3, CO1:PO3] 10M

EACH OF THE FOLLOWING QUESTIONS CARRIES TWLEVE MARKS

Q. Distinguish between Supervisory Learning and Un supervisory Learning in ANN.

[L3, CO1:PO1] 12 M
Q. Highlight briefly the Historical development of Artificial Neural Networks.
[L2, CO1:PO1] 12M
Q. Discuss briefly the structure and function of a biological neuron.
[L2, CO1:PO1] 12M
Q. List and explain different learning laws.
[L2, CO1:PO1] 12M
Q. Realize NOT function using McCulloch-Pitts neuron model.
[L3, CO1:PO3] 12 M
Q. Generate the output of logic AND function by McCulloch-Pitts neuron model.
[L3, CO1:PO3] 12 M
Q. Generate the out of logic OR function by McCulloch-Pitts neuron model.
[L3, CO1:PO3] 12 M
Q. Realize the Ex-OR function by using the McCulloch-Pitts neuron model.
[L3, CO1:PO3] 12 M

4
 UNIT 2

EACH OF THE FOLLOWING QUESTIONS CARRIES TWO MARKS

Q. What is generalization?
[L1, CO2:PO1] 2M
Q. What is Network Pruning?
[L1, CO2:PO1] 2M
Q. What is momentum?
[L1, CO2:PO1] 2M
Q. How to determine the number hidden neurons in single hidden layer feed-forward
neural network?
[L1, CO2:PO1] 2M
Q. List any two applications of Back Propagation network.
[L1, CO2:PO1] 2M
Q. Mention the demerits of Back propagation Network.
[L2, CO2:PO1] 2M
Q. Mention the merits of Back propagation Network.
[L2, CO2:PO1] 2M

EACH OF THE FOLLOWING QUESTIONS CARRIES FOUR MARKS

Q. Sketch the architecture of 10/16/5 multiplayer feed forward neural network.

[L1, CO2:PO1] 4M
Q. Narrate the different approaches and methods of improving generalization in feed
forward Neural Networks.
[L1, CO2:PO1] 4 M
Q. Feed-ward neural network for a given application uses 10 neurons in its input layer
and 5 neurons in its output layer. Determine the possible number of hidden neurons
required in the hidden layer. Assume that the network require only one hidden layer for
the given application.
[L1, CO2:PO3] 4 M
Q. For the given application, the feed-forward neural network of 10/12/6 is used.
Determine its size, width and depth of the network.
[L1, CO2:PO3] 4 M
Q. List the applications of Back propagation Network.
[L1, CO2:PO3] 4 M
Q. Show the linearization of Sigmoid Function.
[L3, CO2:PO3] 4 M

EACH OF THE FOLLOWING QUESTIONS CARRIES SIX MARKS

Q. What is network pruning? Mention different pruning techniques.
[L1, CO2:PO1] 6 M
Q. How to determine the number hidden neurons in single hidden layer feed-forward
neural network? Explain with an example.
[L2, CO2:PO3] 6 M
Q. Mention the merits and demerits of Back propagation network.
[L2, CO2:PO3] 6 M

EACH OF THE FOLLOWING QUESTIONS CARRIES EIGHT MARKS

Q. Distinguish between patterns based learning and batch mode learning.

[L1, CO2:PO1] 8 M
Q. Give the significance of momentum in training feed-forward neural network.
[L1, CO2:PO1] 8 M

5
Q. Give the flowchart for the Generalized Delta Rule (GDR) procedure.
[L2, CO2:PO1] 8 M
Q. Distinguish between online learning and off-line learning.
[L1, CO2:PO1] 8 M

EACH OF THE FOLLOWING QUESTIONS CARRIES TEN MARKS

Q. Give the step-by–step procedure of Back-Propagation Algorithm that uses GDR

procedure for three layer feed-forward neural network.
[L2, CO2:PO3] 10 M
Q. Write the flowchart of error back-propagation training algorithm.
[L1, CO2:PO1] 10 M
Q. Explain briefly the operation of 10/16/5 multiplayer feed forward neural N/W
[L1, CO2:PO1] 10M
Q. What is generalization? How to measure and evaluate generalization?
[L1, CO2:PO2] 10 M
Q. What is cross-validation? Give its significance in feed forward n/w design.
[L1, CO2:PO1] 10 M
Q. Discuss briefly the practical considerations in the design and implementation of Multi
Layer Perceptron Feed Forward Neural Network.
[L1, CO2:PO3] 10 M

Q. What do you understand by Cascade Correlation Architecture? Explain briefly.

[L1, CO2:PO2] 10 M

EACH OF THE FOLLOWING QUESTIONS CARRIES TWLEVE MARKS

Q. Develop a Back propagation algorithm for Multilayer Feed forward neural network
consisting of one input layer, one hidden layer and output layer from first principles.
[L3, CO2:PO3] 12 M
Q. Discuss briefly the practical considerations in implementation of feed-forward neural
networks.
[L2, CO2:PO2] 12 M
Q. List the factors that affect the performance of multilayer feed-forward neural network.
[L1, CO2:PO1] 12 M
UNIT 3

EACH OF THE FOLLOWING QUESTIONS CARRIES TWO MARKS

Q. What is recurrent neural network?

[L1, CO3:PO1] 2 M
Q. What is Hamming Distance?
[L1, CO3:PO1] 2 M
Q. What is Associative Memory?
[L1, CO3:PO1] 2 M
Q. What is Content Addressable Memory?
[L1, CO3:PO1] 2 M
Q. Name the two types of Associative memory.
[L1, CO3:PO1] 2 M
Q. Name the two types of Bidirectional Associative Memory.
[L1, CO3:PO1] 2 M
Q. What is Pattern Association Problem?
[L1, CO3:PO1] 2 M
Q. What is simulated-annealing?
[L1, CO3:PO1] 2 M
6
Q. What is Boltzmann Machine?
[L1, CO3:PO1] 2 M
Q. List the merits of Boltzmann’s Machine.
[L1, CO3:PO1] 2 M
Q. Mention the two different architectures of Boltzmann Machine.
[L1, CO3:PO1] 2 M
Q. List the applications of Boltzmann’s Machine
[L1, CO3:PO1] 2 M
Q. What is Hopfield Memory?
[L1, CO3:PO1] 2 M
Q. Name the two algorithms developed for Pattern Association networks.
[L1, CO3:PO1] 2 M
Q. Determine the Hamming distance between the two vectors given below.
A =[ 0, 1, 0, 0, 1, 1, 1] and B = [ 0, 1, 1, 0, 1, 0, 1].
[L2, CO3:PO3] 2 M
Q. A typical partitioning problem consisting of 1000 vectors and 20 non-empty subsets.
Determine the number of possible partitions.
[L2, CO3:PO3] 2 M
Q. Write the expression to determine the approximate possible number of Partitions in
the context of clustering problem.
[L2, CO3:PO3] 2 M

EACH OF THE FOLLOWING QUESTIONS CARRIES FOUR MARKS

Q. Give the general network synthesis procedure.
[L2, CO3:PO1] 4M
Q. Explain briefly Clustering complexity
[L2, CO3:PO1] 4M
Q. Give the weight storage prescription for Discrete BAM for binary input vector and
bipolar input vector.
[L2, CO3:PO1] 4M
Q. What are the basic parameters of recurrent network design?
[L1, CO3:PO1] 4 M
Q. How to determine the storage capacity of BAM?
[L2, CO3:PO1] 4 M
Q. List the merits, demerits and applications of Boltzmann’s Machine.
[L2, CO3:PO1] 4 M
Q. Consider the two vectors x = [1.-1,1 -1] and y = [ 1, -1, 1,-1]. Compute the Hamming
distance between them.
[L3, CO3:PO3] 4 M
Q. Determine the Hamming distance and average Hamming Distance between the
given below the two vectors A =[ 0, 1, 0, 0, 1, 1, 1] and B = [ 0, 1, 1, 0, 1, 0, 1].
[L2, CO3:PO3] 4 M
Q. Show the typical architecture of Discrete Hopfield Memory.
[L2, CO3:PO3] 4 M

EACH OF THE FOLLOWING QUESTIONS CARRIES EIGHT MARKS

Q. What is recurrent neural network? Give the classification of recurrent n/w.

[L1, CO3:PO1] 8 M
Q. Distinguish between Continuous BAM and Discrete BAM.
[L2, CO3:PO1] 8M
Q. Distinguish between Auto associative Memory and Hetero Associative Memory.
[L1, CO3:PO1] 8 M
Q. Discuss briefly the working concepts of Boltzmann Machine.
[L1, CO3:PO1] 8 M
7
Q. What is Hopfield Memory? Explain briefly.
[L1, CO3:PO1] 8 M
Q. What is content addressable memory (CAM)? Explain briefly.
[L1, CO3:PO1] 8 M
Q. State the Outer-Product Rule.
[L2, CO3:PO2] 8 M
Q. What is bi-directional memory (BAM)? Explain briefly with its architecture.
[L1, CO3:PO1] 8 M
Q. Provide the training algorithm for Hopfield Memory.
[L2, CO3:PO1] 8 M
Q. Show the step-by-step training algorithm for basic pattern association problem using
Hebb rule.
[L2, CO3:PO1] 8 M
Q. A hetero associative network is trained by Hebb outer product rule for input vector S=
[x1, x2, x3, x4] to output row vectors t = [t1,t2]. Find the weight matrix:
S1 = (1,1,0,0) , S2 = (1,1,1,0), S3 = (0,0,1,1), S4 = (0,1,0,0)
t1 = (1,0), t2 = (0.1), t3 = (1,0), t4 = (1,0)
[L3, CO3:PO3] 12 M
Q. A hetero associative network is given. Find the weight matrix and test the network
with the training input vectors.
S1 = (1,1,0,0) , S2 = (0,1,0,0), S3 = (0,0,1,1), S4 = (0,0,1,0)
t1 = (1,0), t2 = (1,0), t3 = (0,1), t4 = (0,1)
[L3, CO3:PO3] 12 M
Q. Give the step-by-step training and testing procedure of Discrete BAM
[L2, CO3:PO2] 12 M

UNIT 4
EACH OF THE FOLLOWING QUESTIONS CARRIES TWO MARKS

Q. What are competitive neural networks?

[L1, CO4:PO1] 2 M
Q. What is Self-Organization?
[L1, CO4:PO1] 2 M
Q. Show the Structure of MAXNET neural network.
[L1, CO4:PO1] 2 M
Q. Show the structure of Kohonen’s SOFM.
[L1, CO4:PO1] 2 M

EACH OF THE FOLLOWING QUESTIONS CARRIES FIVE MARKS

Q. What are competitive neural networks? Show its typical architecture.

[L1, CO4:PO1] 5 M
Q. Provide the summary of SOFM algorithm.
[L1, CO4:PO1] 5 M
Q. Write the flowchart of Kohenen SOFM.
[L1, CO4:PO1] 5 M
Q. Provide the procedure for C-means algorithm.
[L2, CO4:PO2] 5 M

EACH OF THE FOLLOWING QUESTIONS CARRIES TEN MARKS

8
Q. What are Self-organizing neural networks? Explain briefly.
[L1, CO4:PO1] 10 M
Q. Discuss briefly the formal characterization clustering and general clustering
procedures.
[L1, CO4:PO1] 10 M
Q. Explain briefly with a neat diagram typical architecture of SOFM.
[L1, CO4:PO1] 10 M
Q. What is Max-Net/ MAXNET? Explain briefly.
[L1, CO4:PO1] 10 M
Q. Provide the training algorithm for MAXNET.
[L2, CO4:PO2] 10 M
Q. Sketch the architecture of d=4 MAXNET and explain its operation briefly.
[L2, CO4:PO2] 10 M
Q. Write the structure of Kohenen’s SOFM network and give its algorithmic steps.
[L1, CO4:PO1] 10 M

EACH OF THE FOLLOWING QUESTIONS CARRIES TWELVE MARKS

Q. A Kohonen self organizing map is shown with weights in Fig given below:

C1 C2 C3 C4 C5

U U
1 2

(a) Using the square of the Euclidean distance find the cluster unit C j that is closest
to the input vector( 0.3, 0.4)
(b) Using a learning rate of 0.3, find the new weights for unit Cj
(c) Find new weights for Cj-1 and Cj+1, if they are allowed to learn.
[L3, CO4:PO3] 12 M

Q. Consider a Kohonen network with two cluster units and five input units. The weigh
vectors for the cluster units are
w1 = [0.1, 0.3, 0.5, 0.7, 0.9]
and
w2 = [0.9, 0.7, 0.5, 0.3, 0.1]
Use the square of the Euclidean distance to find the winning cluster unit for the input
pattern.
[L3, CO4:PO3] 12 M

9
 UNIT 5

EACH OF THE FOLLOWING QUESTIONS CARRIES TWO MARKS

Q. What is RBF Neural Network?

[L1, CO5:PO1] 2M
Q. What is Time-Delay Neural Networks?
[L1, CO5:PO1] 2M
Q. Mention the two layers in ART1.
[L1, CO5:PO1] 2M
Q. Give the advantage of Adaptive Resonance Theory based neural networks.
[L1, CO5:PO1] 2M
Q. Mention the applications of Time-Delay Neural Networks.
[L1, CO5:PO1] 2M
Q. Mention the applications of Adaptive Resonance Theory based neural networks.
[L1, CO5:PO1] 2M
Q. Give any one application of RBF Neural Network and Time-Delay Neural Network.
[L1, CO5:PO1] 2M
Q. List the main features of RBF NN.
[L1, CO5:PO1] 2M

EACH OF THE FOLLOWING QUESTIONS CARRIES FOUR MARKS

Q. Sketch the architecture of a typical RBF Network.

[L1, CO5:PO1] 4M
Q. Sketch the architecture of a typical Time-Delay –Neural Network (TDNN).
[L1, CO5:PO1] 4M
Q. State Cover’s theorem in the context of RBF Neural Network.
[L1, CO5:PO1] 4M

EACH OF THE FOLLOWING QUESTIONS CARRIES SIX MARKS

Q. Show the structure of a typical RBF neural network.

[L2, CO5:PO1] 6M
Q. What is stability-plasticity Dilemma? Explain briefly.
[L1, CO5:PO1] 6M
Q. Show the Architecture of ART1.
[L1, CO5:PO1] 6M
Q. What is adaptive resonance theory? Explain briefly.
[L1, CO5:PO1] 6M

EACH OF THE FOLLOWING QUESTIONS CARRIES EIGHT MARKS

Q. Discuss briefly the architecture of Time-Delay Neural Network (TDNN).

[L2, CO5:PO1] 8M
Q. List the techniques used to update the weights and centres of RBFN.
[L2, CO5:PO1] 8M
Q. Give the training procedure for RBF NN.
[L2, CO5:PO1] 8M
Q. Draw the typical architecture of ART-1 and explain its operation.
[L2, CO5:PO1] 8M

10
 EACH OF THE FOLLOWING QUESTIONS CARRIES TEN MARKS

Q. Distinguish between Radian Basis Function Neural Network and Multi layer
Perceptron Feed-Forward Neural Network.
[L2, CO5:PO1] 10M
Q. Provide the training algorithm for the RBF NN with its flowchart.
[L2, CO5:PO1] 10M

11