H.

T No
Regulations:
Sreenidhi Institute of Science and Technology A22
(An Autonomous Institution)
Code No:9C517 Date: 23-January-2025 (AN)
B.Tech III-Year I- Semester External Examination, January-2025 (Regular)
ARTIFICIAL NEURAL NETWORKS (ECE)
Time: 3 Hours Max.Marks:60
Note: a) No additional answer sheets will be provided.
b) All sub-parts of a question must be answered at one place only, otherwise it will not be valued.
c) Missing data can be assumed suitably.
Bloom's Cognitive Levels of Learning (BCLL)
Remember L1 Apply L3 Evaluate L5
Understand L2 Analyze L4 Create L6
Part - A Max.Marks: 6x2=12
ANSWER ALL QUESTIONS, EACH QUESTION CARRIES 2 MARKS.
BCLL CO(s) Mark
s
1 Sketch and explain briefly about ‘axon’, with reference to a biological neural L2 CO1 [2
network? M]
It is the link attached to the soma.
Axon is electrically active and serves as an output channel.
Axons which mainly appear on output cells.
Axons are non-linear threshold devices.
2 Write the differences between ADALINE and MADALINE. L3 CO2 [2
Adaline is a Adaptive linear element computing model. In this model the M]
analog activation value ‘x’ is compared with the targeted output ‘b’. Thata is,
the output is a linear function of the activation value ‘x’.
Madaline: The Madaline model consists of many Adaline in parallel with a single
output unit. The Adaline layer is present between the input layer and the Madaline
layer hence Adaline layer is a hidden layer. The weights between the input layer
and the hidden layer are adjusted, and the weight between the hidden layer and the
output layer is fixed.
3 What is Auto Associative Memory in neural networks? L2 CO3 [2
If the two layers Fl and F2 coincide and the weights are symmetric, i.e., wji = M]
wij, i ≠ j, then we obtain an auto-associative memory in which each unit is
connected to every other unit and to itself as shown below.

4 What is Perceptron learning law in neural networks? L2 CO4 [2

Here the change in the weight vector is given by M]
wi = [bi - sgn(wiT a)]a
where sgn(x) is sign of x.
Therefore, the jth component of wi is given by
wij = [bi - sgn(wiT a)]aj
=  (bi – si) aj , for j = 1, 2, 3,……, M
This law is applicable only for bipolar output functions f(.).
This is also called discrete perceptron learning law.
The expression for wij shows that the weights are adjusted only if the actual output
si is incorrect, since the term in the square brackets is zero for the correct output.
5 What is Pattern clustering? Illustrate with a diagram. L3 CO5 [2
Given a set of patterns, if they are grouped according to similarity of the patterns, M]
then the resulting problem is called pattern clustering.
6 Define Radial basis function. L1 CO6 [2
Page 1 of 1
 A Radial Basis Function is a (RBF) neural network has an input layer, a hidden M]
layer and an output layer. The neurons in the hidden layer contain Gaussian transfer
functions whose outputs are inversely proportional to the distance from the center of
the neuron.

Part – B Max.Marks: 6x8=48

ANSWER ALL QUESTIONS. EACH QUESTION CARRIES 8 MARKS.
BCLL CO(s) Marks

7. Compare and Contrast the performance of a computer with biological L3 CO1 [8M]
neural networks.
ANS:Since ANNs are implemented on computers, it is worth comparing the
processing capabilities of a computer with those of the brain.The performance of
computer and biological neural networks is compared in terms of:1. Speed 2.
Processing 3. Size and Complexity 4. Storage
5. Fault Tolerance 6. Control Mechanism
Speed: Neural networks are slow in processing information. Most of the
advanced computers the cycle time corresponding to execution of one step of a
program in the central processing unit is in the range of few nanoseconds. The
cycle time corresponding to a neural event prompted by an external stimulus
occurs in milliseconds range. Thus the computer processes information nearly a
million times faster.
Processing: Neural networks can perform massively parallel operations. Most
programs have large number of instructions, and they operate in a sequential
mode one instruction after another on a conventional computer. On the other
hand, the brain operates with massively parallel operations, each of them having
comparatively fewer steps. This explains the superior performance of human
information processing for certain tasks, despite being several orders of
magnitude slower compared to computer processing of information.
Size and complexity: Neural networks have large number of computing
elements, and the computing is not restricted to within neurons.The number of
neurons in a brain is estimated to be about 10 11 and the total number of
interconnections to be around 10 15. It is this size and complexity of connections
that may be giving the brain the power of performing complex pattern recognition
tasks which we are unable to realize on a computer.
Storage: Neural networks store information in the strengths of the
interconnections. In a computer, information is stored in the memory which is
addressed by its location. Any new information in the same location destroys the
old information.In contrast, in a neural network new information is added by
adjusting the interconnection strengths, without destroying the old
information.Thus information in the brain is adaptable, whereas in the computer it
is strictly replaceable.
Fault tolerance: Neural networks exhibit fault tolerance since the information is
distributed in the connections throughout the network.Even if a few connections
are snapped or a few neurons are not functioning, the information is still
preserved due to the distributed nature of the encoded information. In contrast,
computers are inherently not fault tolerant, in the sense that information corrupted
in the memory cannot be retrieved.
Control mechanism: There is no central control for processing information in the
brain. In a computer there is a control unit which monitors all the activities of
computing. In a neural network each C neuron ads based on the information
locally available, and transmits its output to the neurons connected to it. Thus
there is no specific control mechanism external to the computing task.

OR
8 Explain briefly about the operations of a biological neural network? L1 CO1 [8M]
In the state of inactivity the interior of the neuron, the protoplasm, is negatively charged

Page 2 of 1
 against the surrounding neural liquid containing positive Sodium (Na+) ions.. The
resulting resting potential of about - 70 mV is supported by the action of the cell
membrane, which is impenetrable for the positive Sodium ions. This causes a deficiency
of positive ions in the protoplasm. Signals arriving from the synaptic connections may
result in a temporary depolarization of the resting potential. When the potential is
increased to a level above - 60 mV, the membrane suddenly loses its impermeability
against Na+ ions, which enter into the protoplasm and reduce the potential difference.
This sudden change in the membrane potential causes the neuron to discharge. Then the
neuron is said to have fired. The membrane then gradually recovers its original properties
and regenerates the resting potential over a period of several milliseconds. During this
recovery period, the neuron remains incapable of further excitation. The discharge, which
initially occurs in the cell body, propagates as a signal along the axon to the synapses. The
intensity of the signal is encoded in the frequency of the sequence of pulses of activity,
which can range fiom about 1 to 100 per second. The speed of propagation of the
discharge signal in the cells of the human brain is about 0.5-2 mls. The discharge signal
travelling along the axon stops at the synapses, because there exists no conducting link to
the next neuron. Transmission of the signal across the synaptic gap is mostly effected by
chemical activity. When the signal arrives at the presynaptic nerve terminal, special
substances called neurotransmitters are produced in tiny amounts. The neurotransmitter
molecules travel across the synaptic junction reaching the postsynaptic neuron within
about 0.5 ms. These substances modify the conductance of the postsynaptic membrane for
certain ions, causing a polarization or depolarization of the postsynaptic potential. If the
induced polarization potential is positive, the synapse is termed excitatory, because the
influence of the synapse tends to activate the postsynaptic neuron. If the polarization
potential is negative, the synapse is called inhibitory, since it counteracts excitation of the
neuron. All the synaptic endings of an axon are either of an excitatory or an inhibitory
nature.
9. Explain the perceptron model of neuron. L2 CO2 [8M]
ANS:

Page 3 of 1
 OR
10 Draw the schematic diagram of McCulloch-Pitts Model of neuron and write the L4 CO2 [8M]
equations that describes the operation of MP model of neuron.

11 What are Recurrent Neural Networks? Explain briefly with a neat diagram. In L1 CO3 [8M]
what way they are different from.
Recurrent neural networks are designed to learn sequential or time varying patterns. A
recurrent net is a neural network with feedback (closed loop) connections.
The architectures range from fully interconnected (Figure 1) to partially connected nets
(Figure 2), including multilayer feed-forward networks with distinct input and output
layers. Fully connected networks do not have distinct input layers of nodes, and each node
has input from all other nodes. Feedback to the node itself is possible.

Simple partially recurrent neural networks (Figure 2) have been used to learn strings of
characters. Athough some nodes are part of a feedforward structure, other nodes provide
the sequential context and receive feedback from other nodes. Weights from the context
units (C1 and C2) are processed like those for the input units, for example, using
Page 4 of 1
 backpropagation. The context units receive time-delayed feedback from, in the case of
Figure 2, the second layer units. Training data consists of inputs and their desired
successor outputs. The net can be trained to predict the next letter in a string of characters
and to validate a string of characters.
OR
12 Explain the following Neural Network Architectures in detail: L3 CO3 [8M]
i) Counter propagation neural network (CNN)
The counter propagation network is basically constructed from an instar-outstar
model. This model is three layer neural network that performs input-output data
mapping, producing an output vector y in response to input vector x, on the basis
of competitive learning.This is a multilayer network based on the various
combining structures of input, clustering and output layers. CPN reduces the time
by one hundred times. CPN provides solution for those applications which cannot
have larger iterations. CPN can be used for data compression, approximation of
functions, pattern association, pattern completion and signal enhancement
applications.

ii) Recurrent neural network (RNN)

Recurrent neural networks allow previous outputs to be used as inputs while
having hidden states.
13 Explain in detail about any three Basic Learning Laws. L1 CO4 [8M]
Learning laws:
Hebb's Law
Perceptron Learning Law
Delta Learning Law
Widrow and Hoff LMS Learning Law
Correlation Learning Law
lnstar (Winner-take-all) Learning Law
Outstar Learning Law
Learning law is a law of adjusting the weights or weight adjustment.That is the weight
vector for the ith processing unit at time instant (t + 1) in terms of the weight vector at
time instant (t) as follows:
wi(t + 1) = wi(t) + wi(t)
where wi(t) is the change in the weight vector. There are different methods for
implementing the learning feature of a neural network, leading to several learning laws.
Hebb's Law: Here the change in the weight vector is given
by wi =f (wiT a)a Therefore, the jth component of wi is
given by
wij = f (wiT a) aj
=  si aj , for j = 1, 2, 3,……, M
where si is the output signal of the ith unit. The law states that the weight increment is
proportional to the product of the input data and the resulting output signal of the unit.
This law requires weight initialization to small random values around wij = 0 prior to
learning. This law represents an unsupervised learning.

Perceptron Learning Law:Here the change in the weight vector is given by

wi
= [bi - sgn(wi a)]a where
T

sgn(x) is sign of x.
Therefore, the jth component of wi is given by
wij = [bi - sgn(wiT a)]aj
=  (bi – si) aj , for j = 1, 2, 3,……, M
This law is applicable only for bipolar output functions f(.). This is also called discrete
perceptron learning law.The expression for wij shows that the weights are adjusted
only if the actual output si is incorrect, since the term in the square brackets is zero for
Page 5 of 1
the correct output.
This is a supervised learning law, as the law requires a desired output for each input. In
implementation, the weights can be initialized to any random initial values, as they are
not critical. The weights converge to the final values eventually by repeated use of the
input-output pattern pairs, provided the pattern pairs are representable by the system.

Delta Learning Law:

Here the change in the weight vector is given by

wi =[bi-
T T
f(wi a)] f(wi a)a where is
the derivative with respect to x.
Hence,
wij = [bi - f(wiT a)] f(wiT a)aj
=  (bi – si) aj , for j = 1, 2, 3,……, M
This law is valid only for a differentiable output function, as it depends on the
derivative of the output function . It is a supervised learning law since the change in
the weight is based on the error between the desired and the actual output values for a
given input. Delta learning law can also be viewed as a continuous perceptron learning
law. In implementation, the weights can be initialized to any random values as the
values are not very critical. The weights converge to the final values eventually by
repeated use of the input-output pattern pairs. The convergence can be more or less
guaranteed by using more layers of processing units in between the input and output
layers. The delta learning law can be generalized to the case of multiple layers of a
feedforward network.
Widrow and Hoff
LMS Learning Law:
Here the change in the weight vector is given
by
wi = [bi - wiT a]a

wij = [bi - wiT a] f(wiT a)aj , for j = 1, 2, 3,……, M

This is a supervised learning law and is a special case of the delta learning law, where the
output function is assumed linear, i.e., f(xi)  xi. In this case the change in the weight
is made proportional to the negative gradient of the error between the desired output
and the continuous activation value, which is also the continuous output signal due to
linearity of the output function. Hence, this is also called the Least Mean Squared (LMS)
error learning law. This is same as the learning law used in the Adaline model of neuron.
In implementation, the weights may be initialized to any values. The input-output pattern
pairs data is applied several times to achieve convergence of the weights for a given set of
training data. The convergence is not guaranteed for any arbitrary training data set.

Correlation Learning Law:

Here the change in the weight vector is given by
wi = bia
Hence, wij = biaj , for j = 1, 2, 3,……, M
This is a special case of the Hebbian learning with the output signal (si) being replaced
by the desired signal (bi). But the Hebbian learning is an unsupervised learning, whereas
the correlation learning is a supervised learning, since it uses the desired output value to
adjust the weights. In the implementation of the learning law, the weights are initialized
to small random values close to zero, i.e., wij = 0.

lnstar (Winner-take-all) Learning Law:This is relevant for a collection of neurons,

organized in a layer as shown in Figure.

Page 6 of 1
 All the inputs are connected to each of the units in the output layer in a feedforward
manner. For a given input vector a, the output from each unit ‘i’ is computed using the
weighted sum wiTa. The unit ‘k’ that gives maximum output is identified.
That is, wkT a = (wiT a). Then the weight vector leading to the kth unit is adjusted as
follows:

wk = (a-wk)
Therefore wkj = [aj - wkj] , for j = 1, 2, 3,……, M
The final weight vector tends to represent a group of input vectors within a small
neighborhood. This is a case of unsupervised learning. In implementation, the values of
the weight vectors are initialized to random values prior to learning, and the vector
lengths are normalized during learning.

Outstar Learning Law:The outstar learning law is also related to a group of units
arranged in a layer as shown in Figure.

In this law the weights are adjusted so as to capture the desired output
pattern characteristics. The adjustment of the weights is given by
wjk = [bj - wjk] , for j = 1, 2, 3,……, M
where the kth unit is the only active unit in the input layer. The vector b = (bl, b2, ...,
bMIT is the desired response from the layer of M units. The outstar learning is a
supervised learning law, and it is used with a network of instars to capture the
characteristics of the input and output patterns for data compression. In
,implementation, the weight vectors are initialized to zero prior to blaming.
OR
14 Explain the significance of the initial values of weights and the learning rate L3 CO4 [8M]
parameter in the seven basic learning laws?
Initial Weight Values:
Impact on Activation Functions: The initial weight values significantly impact the
output of activation functions within a neuron, which in turn affects the signal
propagation through the network.
Avoiding Vanishing/Exploding Gradients: Choosing appropriate initial weights helps
prevent the "vanishing gradient" problem (where gradients become very small during
back propagation) or the "exploding gradient" problem (where gradients become very
large), both of which can hinder learning.
Initialization Techniques: Different weight initialization methods, like Xavier
initialization or He initialization, are designed to mitigate these issues depending on the
activation function used.


Page 7 of 1
 Learning Rate Parameter:
Control of Weight Updates: The learning rate directly controls the magnitude of weight
adjustments made during each training iteration, essentially dictating how fast the model
learns.
Convergence Speed: A high learning rate can lead to faster convergence but may
overshoot the optimal solution, while a low learning rate might result in slow
convergence or getting stuck in a local minimum.
Adapting Learning Rate: Techniques like learning rate decay can dynamically adjust the
learning rate during training to address issues like early convergence or slow learning in
later stages.

15 Explain the following and Distinguish between them: L2 CO5 [8M]

i)Pattern Association
The pattern association problem is illustrated in the Figure. The input patterns are
shown as a1, a2, a3 and the corresponding output patterns as b 1, b2, b3.

The objective of designing a neural network is to capture the association between

input-output pattern pairs in the given set of training data, so that when any of the
inputs al is given, the corresponding output bl is retrieved. If the amplitude of the
noise added to ‘al’ is so large that the noisy input pattern is closer to some pattern
(say ‘ak’) other than the correct one (‘al’), then the network produces an incorrect
output pattern ‘bk’, k ≠ l. Thus an incorrect output pattern would be retrieved for
the given noisy input.

ii)Pattern Classification
If a group of input patterns correspond to the same output pattern, then typically
there will be far fewer output patterns compared to the number of input patterns.

In other words, if some of the output patterns in the pattern association problem
are identical, then the number of distinct output patterns can be viewed as class
labels and the input patterns corresponding to each class can be viewed as
samples of that class.The network displays an accretive behaviour in this case.

iii)Pattern Mapping
Given a set of input-output pattern pairs as in the pattern association problem, if
the objective is to capture the implied mapping, instead of association, then the
problem becomes a pattern mapping problem.

In a pattern mapping problem both the input and the output patterns are only
samples from the mapping system. Once the system behaviour is captured by the
network, the network would produce a possible output pattern for a new input
pattern, not used in the training set.The possible output pattern would be
approximately an interpolated version of the output patterns corresponding to the
input training patterns close to the given test input pattern. Thus the network
Page 8 of 1
 displays an interpolative behavior. Typically the input and output pattern spaces
are continuous in this case, and the mapping function must be smooth for the
interpolation to work satisfactorily.An example of the data for a pattern mapping
problem could be the input data given to a complex physical system and the
corresponding output data from the system for a number of trials. The objective is
to capture the unknown system behavior from the samples of the input-output pair
data.

OR
16 Compare and Contrast between Auto association problem and Hetero association L3 CO5 [8M]
problem?
Autoassoclatlon problem:
If each of the output patterns bl in a pattern association problem is identical to the
corresponding input patterns al , then the output pattern space is identical to the
input pattern space (as shown in the Figure).

In such a case the problem becomes an autoassociation problem. This is a trivial

case where the network merely stores the given set of input patterns. When a
noisy input pattern is given, the network retrieves the same noisy pattern. Thus
there is an absence of accretive behaviour. If each of the output patterns bl in a
pattern association problem is identical to the corresponding input patterns al ,
then the output pattern space is identical to the input pattern space (as shown in
the Figure).

Hetero association problem

In a hetero association problem, the recovered pattern is different from the input pattern in
type, format, and content.
The hetero association problem is also known as the hetero-associative correlator.
The hetero association problem is a type of bidirectional associative memory (BAM).
The hetero association problem can be solved using a hetero-associative neural network
model.
The hetero association problem can be solved using a model that stores objects as an
indeterminate content on a medium.
17 Explain pattern recognition tasks performed by competitive learning neural L4 CO6 [8M]
network with neat sketches.
Temporary pattern storage:
If a given input pattern is stored in a network, even in a transformed form, in such
a way that the pattern remains only until a new pattern input is given, then the
problem becomes that of a short term memory or temporary storage problem.
Pattern clustering problem:
Given a set of patterns, if they are grouped according to similarity of the patterns,
then the resulting problem is called pattern clustering. It is illustrated in Figure.

There are two types of problems here. In one case the network displays an
accretive behaviour (Figure. a). That is, if an input pattern not belonging to any
Page 9 of 1
 group is presented, then the network will force it into one of the groups. The input
pattern space is typically a continuous space. The test input patterns could be the
same as the ones used in the training or could be different. The output pattern
space consists of a set of cluster centres or labels.The second type of problem
displays interpolative behaviour as shown in Figure. b.In this case, a test input
pattern not belonging to any group produces an output which is some form of
interpolation of the output patterns or cluster centers, depending on the proximity
of the test input pattern to the input pattern groups formed during training.
Feature mapping problem:
In the pattern clustering problem a group of approximately similar input patterns
are identified with a fixed output pattern or a group label.
On the other hand, if similarities of the features of the input patterns have to be
retained in the output, the problem becomes one of feature mapping.
In this, a given set of input patterns are mapped onto output patterns in such a
way that the proximity of the output patterns reflect the similarity of the features of
the corresponding input patterns.
When a test input pattern is given, it will generate an output which is in the
neighbourhood of the outputs for similar patterns.

OR
18 Sketch and Explain briefly about the Architecture of Simple Forward CPN. L5 CO6 [8M]
The counter propagation network is basically constructed from an instar-outstar
model. This model is three layer neural network that performs input-output data
mapping, producing an output vector y in response to input vector x, on the basis
of competitive learning.

This is a multilayer network based on the various combining structures of input,

clustering and output layers. CPN reduces the time by one hundred times. CPN
provides solution for those applications which cannot have larger iterations. CPN
can be used for data compression, approximation of functions, pattern
association, pattern completion and signal enhancement applications.

-- 00 -- 00 –

Page 10 of 1