Jawaharlal Nehru Technological University Hyderabad 2006-2nd Year B.E Computer Science JNTU B.TECH 4th nd sem Neural Networks sets 1,2,3,4 for Engg - Question Paper

(b) discuss the limitations of perceptron? [8]
3. Implement a backpropagation algorithm to solve EX-OR issue and try the ar-
chitecture in which there is a hidden layer with 3 hidden units and the network
is fully connected. [8+8]
4. Show that the energy function of a Hopfield network may be expressed as
E = -
N
2
M
P
v=1
m2
v
where mv denotes overlaps described by
mv = 1
N
N
P
j=1
xj?v,j,v = 1, 2, .......M
where xj is the j'th element of the state vector x, ?v,j is the jth element of the
fundamental memory ?v , and M is the number of fundamental memories. Prove
that the above energy function is a Lypunov function. [16]
5. (a) What is the Kohonen layer architure and discuss its features. [4+4]
(b) discuss the Kohonen’s learning algorithm. [4+4]
6. Derive expressions for the weight updation involved in counter propagation. [16]
7. (a) elaborate the advantages of ART network. explain about gain control in ART
network. [3+5]
(b) explain in detail about orienting subsystem in an ART network. [8]
8. define how a neural network may be trained for a trend recognition task.
Illustrate with an example [16]
? ? ? ? ?
1 of 1
Code No: RR420507 Set No. 3
IV B.Tech II Semester Regular Examinations, Apr/May 2006
NEURAL NETWORKS
( Common to Computer Science & Engineering and Electronics &
Computer Engineering)
Time: three hours Max Marks: 80
ans any 5 ques.
All ques. carry equal marks
? ? ? ? ?
1. elaborate multilayer ANNs? Draw the structure of a multilayer ANN and identify
the salient characteristics and advantages. [3+5+4+4]
2. Briefly explain about linear separability and the solution for EX-OR issue.Also
suggest a network that can solve EX-OR issue. [4+6+6]
3. discuss about the generalized delta- rule and derive the weight updatation for a
multi layer feed forward neural network. [8+8]
4. (a) elaborate the limitations of Hopfield network? Suggest methods that may
overcome these limitations. [4+4]
(b) A Hopfield network made up of 5 neurons, which is needed to store the
subsequent 3 fundamental memories: [8]
?1 = [+1,+1,+1,+1,+1]T
?2 = [+1,-1,-1,+1,-1]T
?3 = [-1,+1,-1,+1,+1]T
Evaluate the 5-by-5 synaptic weight matrix of the network.
5. discuss the Kohonen’s method of unsupervised learning. explain any example as
its application. [8+8]
6. Using suitable diagrams and equations discuss the basic Bidirectional Associative
Memory configuration. Also define its energy function. [4+4+6]
7. (a) ART network exploits in full 1 of the inherent advantages of neural com-
puting technique, namely parallel processing discuss. [8]
(b) define the architecture and operation of ART2 network. [3+5]
8. discuss the concept of trend recognition and how artificial neural network is
helping in the trend recognition issues. [6+10]
? ? ? ? ?
1 of 1
Code No: RR420507 Set No. 4
IV B.Tech II Semester Regular Examinations, Apr/May 2006
NEURAL NETWORKS
( Common to Computer Science & Engineering and Electronics &
Computer Engineering)
Time: three hours Max Marks: 80
ans any 5 ques.
All ques. carry equal marks
? ? ? ? ?
1. (a) discuss about biological neuron with neat diagram ? [3+3]
(b) discuss in detail the properties of biological neuron. [4]
(c) Compare: biological neuron and Artificial neuron ? [6]
2. State and prove the perceptron convergence theorem. [2+14]
3. discuss the backpropagation algorithm and derive the expressions for weight up-
date relations? [8+8]
4. define the Hopfield model. In this model why is the energy of the all zero state
always ‘0’ in any net of any size? Use this fact to argue that at lowest 1 threshold
must be negative for the all-zero state not to be stabilize well. [4+4+8]
5. discuss the Kohonen’s method of unsupervised learning. explain any example as
its application. [8+8]
6. Derive expressions for the weight updation involved in counter propagation. [16]
7. (a) ART network exploits in full 1 of the inherent advantages of neural com-
puting technique, namely parallel processing discuss. [8]
(b) define the architecture and operation of ART2 network. [3+5]
8. discuss the concept of trend recognition and how artificial neural network is
helping in the trend recognition issues. [6+10]
? ? ? ? ?
1 of 1

Earning:   Approval pending.