International Journal of Emerging Engineering Research and Technology

Volume 2, Issue 2, May 2014, PP 143-148

Artificial Neural Networks-A Study

Er.Parveen Kumar1, Er.Pooja Sharma2,
1
Department of Electronics and Communication Engineering,
Galaxy Global Group of Institutions, Ambala, India (M.Tech Student)
[email protected]
2
Department of Electronics and Communication Engineering,
Galaxy Global Group of Institutions, Ambala, India (Assistant Professor)
[email protected]

Abstract: Artificial neural networks commonly referred as the neural networks are the information or signal
processing mathematical model that is based on the biological neuron. A neural network is a complex structure
which consist a group of interconnected neurons which provides a very exciting alternatives for complex
problem solving and other application which can play important role in today’s computer science field so
researchers from the different discipline are designing the artificial neural networks to solve the problems of
pattern recognition, prediction, optimization, associative memory and control. In this paper we have presented
the basic study of the artificial neural network, its characteristics and its applications.

Keywords: Artificial Neural Network (ANN), Feedback Network, Feed-Forward Network, Artificial Neuron,
Biological Paradigm, Pattern Recognition.

1. INTRODUCTION particular the brain) which is capable of machine

The study of brain is an interesting area since a learning as well as pattern recognition. Whereas
long time. With advancement in the field of animal’s nervous system is more complex than
electronics and computer science, it was the the human so the system designed like this will
assumed that we can use this natural way of this be able to solve more complex problems.
thinking process of brain to design some Artificial neural networks are generally
artificial intelligence system. presented as systems of highly interconnected
"neurons" which can compute values from
The first step toward artificial intelligence came
inputs.
into existence in 1943 when Warren McCulloch,
a neurophysiologist, and a mathematician,
Walter Pitts, wrote a paper on how neurons
work. Mathematical analysis has solved some of
the mysteries posed by the new models but has
left many questions for future investigations.
There is no need to say, the study of neurons,
their interconnections, and their role as the
brain’s elementary building blocks is one of the
most dynamic and important research fields in
modern world of electronics and computer
science.
2. ARTIFICIAL NEURAL NETWORKS Fig.1 A Simple Neural Network

In electronics engineering and related fields, Neural Network is just like a website network of
artificial neural networks (ANNs) are interconnected neurons which can be millions in
mathematical or computational models that are number. With the help of these interconnected
inspired by a human’s central nervous system (in neurons all the parallel processing is being done

©IJEERT www.ijeert.org 143

Er.Parveen Kumar & Er.Pooja Sharma

in body and the best example of Parallel Neural networks are similar to biological neural
Processing is human or animal’s body. networks in performing functions collectively
and in parallel by the units, rather than there
Currently, artificial neural networks are the
being a clear delineation of subtasks to which
clustering of the primitive artificial neurons.
various units are assigned. The term "neural
This clustering occurs by creating layers which
network" usually refers to models employed in
are then connected to one another. How these
statistics, cognitive psychology and artificial
layers connect is the other part of the "art" of
intelligence. Neural network models which
engineering networks to resolve the complex
emulate the central nervous system are part of
problems of the real world.
theoretical neuroscience and computational
So neural networks, with their stronger ability to neuroscience.
derive meaning from complicated or imprecise
2.2 Working of Neural Networks
data, can be used to extract patterns and detect
trends that are too complex to be noticed by The working of neural networks revolves around
the myriad of ways these individual neurons can
either humans or other computer techniques. be clustered together. This clustering occurs in
2.1 Background the human mind in such a way that information
can be processed in a dynamic, interactive, and
The examination of the central nervous system self-organizing way. Biologically, neural
of human brain was the inspiration of neural networks are constructed in a three-dimensional
networks. In an Artificial Neural Network, world from microscopic components. These
simple artificial nodes, known as "neurons", neurons seem capable of nearly unrestricted
"processing elements" or units", are connected interconnections. That is not true of in the case
together to form a network which is called a of any proposed, or existing, man-made
biological neural network. network. Integrated circuits, using current
technology, are two-dimensional devices with a
limited number of layers for interconnection.
This physical reality restrains the types, and
scope, of artificial neural networks that can be
implemented in silicon. Currently, neural
networks are the simple clustering of the
primitive artificial neurons. This clustering
occurs by creating layers which are then
connected to one another. How these layers
connect is the other part of the "art" of
Fig.2 Genetic Neuron engineering networks to resolve real world
problems.
There is no single formal definition of an
artificial neural network. However, a class of 3. ANN MODELS
statistical or mathematical or computational Neural network models in artificial intelligence
models may commonly be called "Neural
are essentially simple mathematical models
Networks" if they possess the following
characteristics: defining a function f: X->Y or a distribution
over X or both X and Y, but sometimes models
1. Consist of sets of adaptive weights, i.e.
numerical parameters that are tuned by a are also intimately associated with a particular
learning algorithms, and learning algorithm or learning rule.
2. Capable of approximating non-linear A common use of the ANN model really means
functions of their inputs. the definition of a class of such functions (where
The adaptive weights are conceptually members of the class are obtained by varying
connection strengths between neurons, which parameters, connection weights, or specifics of
are activated during training and prediction. the architecture such as the number of neurons
or their connectivity).

International Journal of Emerging Engineering Research and Technology 144

Artificial Neural Networks-A Study

3.1 Network function 4. CHARACTERISTICS OF NEURAL

NETWORK
The word network in the term 'artificial neural
network' refers to the interconnections between Basically Computers are good in calculations
the neurons in the different layers of each that takes inputs process then and gives the
system. An example system has three layers. result as per the calculations which is done by
The first layer has input neurons which send using the particular Algorithm which are
data via synapses to the second layer of neurons, programmed in the software’s but ANN uses its
and then via more synapses to the third layer of own rules, the more decisions they make, the
output neurons. More complex systems will better decisions may become. [6] The
have more layers of neurons with some having Characteristics are basically those which should
increased layers of input neurons and output be present in intelligent System like robots and
neurons. The synapses store parameters called other Artificial Intelligence Applications. There
"weights" that manipulate the data in the are six characteristics of Artificial Neural
calculations. An ANN is typically defined by Network which are basic and important for this
three types of parameters: technology which are showed with the help of
diagram:
1. The interconnection pattern between the
different layers of neurons
2. The learning process for updating the weights
of the interconnections
3. The activation function that converts a
neuron's weighted input to its output activation.

Fig.4 Characteristics of ANN

4.1 The Network Structure

The Network Structure of ANN should be
simple and easy. There are basically two types
Fig.3 Nonlinear Model of Neuron
of structures recurrent and non recurrent
structure. The Recurrent Structure is also known
as Auto associative or Feedback Network and
the Non Recurrent Structure is also known as
Associative or feed-forward Network. [3, 4, 6, 7]
In Feed forward Network, the signal travel in
one way only but in Feedback Network, the
signal travel in both the directions by
introducing loops in the network. As shown in
the figures below:
Fig.4 Multilayer Artificial Neural Network

International Journal of Emerging Engineering Research and Technology 145

Er.Parveen Kumar & Er.Pooja Sharma

the result is summation of all the outputs which

comes after processing various inputs. The
Partial answer is worthless for any user in the
ANN System. [7]
4.6 Learning Ability
In ANN most of the learning rules are used to
develop models of processes, while adopting the
network to the changing environment and
discovering useful knowledge. These Learning
methods are Supervised, Unsupervised and
Reinforcement Learning. [7]

Fig. 5(a). Feed Forward Network 5. ADVANTAGES OF NEURAL NETWORKS

The neural networks have a lot of applications
here we have discussed some of the most
important applications of the neural networks
[2]:
1. Adaptive learning: A neural networks
have the ability to learn how to do
things.
2. Self-Organisation: A neural network or
ANN can create its own representation
of the information it receives during
learning.
3. Real Time Operation: In neural network
Fig. 5(b) Feedback Network or ANN computations can be carried out
in parallel.
4.2 Ability of Parallel Processing
4. Pattern recognition is a powerful
ANN is only the concept of parallel processing technique for the data security. Neural
in the computer field. Parallel Processing is done networks learn to recognize the patterns
by the human body in human neurons that is which exist in the data set.
very complex but by applying basic and simple
parallel processing techniques we implement it 5. The system is developed by learning
in ANN like Matrix and some matrix rather than programming. Neural
calculations. [7] networks teach themselves the patterns
in the data freeing the analyst for more
4.3 Distributed Memory interesting work.
ANN is very vast system so single unit memory 6. Neural networks are flexible in a
or centralized memory cannot fulfill the need of changing environment. Although neural
ANN system so in this condition we need to networks may take some time to learn a
store information in weight matrix which form a sudden drastic change but they are
long term memory because information is stored excellent in adapting the constantly
as patterns throughout the network structure. [7] change in information.
4.4 Fault Tolerance Ability 7. Neural networks can build informative
models whenever conventional
ANN is a very complex system so it is necessary approaches fail. Because neural
that it should be a fault tolerant. Because if any networks can handle very complex
part becomes fails it will not affect the system as interactions they can easily model data
much but if the all parts fails at the same time which is too difficult to model with
the system will fails completely. [7] traditional approaches such as
4.5 Collective Solution inferential statistics or programming
logic.
ANN is a interconnected system the output of a
system is a collective output of various input so

International Journal of Emerging Engineering Research and Technology 146

Artificial Neural Networks-A Study

8. Performance of neural networks is very (vehicle control, process control), game-

good and better on most of the playing and decision making (chess,
problems. The neural networks can racing), pattern recognition (radar
build models that are more complex in systems, face identification, object
the structure of the data in significantly recognition, etc.), sequence recognition
less time. (gesture, speech, handwritten text
recognition), medical diagnosis,
6. LIMITATIONS OF NEURAL NETWORK financial applications, data mining (or
In this world everything has some merits and knowledge discovery in databases,
demerits, so the neural network system also has "KDD").
some merits and demerits. The limitations of 8. Another interesting use is when using
ANN [6] are: the Smartphone as a media hub; a user
can dock the device to the TV and watch
1. ANN or Neural Networks is not a daily content from the device- while
life problem solver. controlling the content in a touch-free
manner from a far.
2. There is no structured methodology
available. 9. If your hands are wet or dirty or a
3. There is no single standardized person hates smudges, touch-free
paradigm for Neural Networks controls are a benefit.
development. 8. CONCLUSION
4. The Output Quality of an ANN can be In this paper we discussed about the artificial
unpredictable. neural network, working of neural networks,
5. Many ANN Systems does not describe characteristics of ANN, its advantages,
how they solve the problems. limitations and applications of ANN. There are
various advantages of ANN over conventional
6. Nature of ANN is like a Black box. approaches. Depending on the nature of the
7. APPLICATION application and strength of the internal data
patterns you can generally expect a network to
The real time applications of Artificial Neural train quite well. This applies to problems where
Networks are: the relationships may be quite dynamic or non-
linear. By studying Artificial Neural Network
1. Functional approximation, including we had concluded that as the technology is
time series prediction and modelling. increasing the need of Artificial Intelligence is
2. Call control- answer an incoming call also increasing because of parallel processing,
(speaker-ON) with a swipe of the hand because by using parallel processing we can do
while driving. more than one task at a time. So Parallel
Processing is needed in this present time because
3. Classification, including pattern and with the help of parallel processing we can save
sequence recognition, pattern detection more and more time and money in any task
and sequential decision making. related to electronics, computers and robotics. If
4. Skip tracks or control volume on your we talk about the Future work we can say that
media player using simple hand we have to develop more algorithms and
motions. programs so that we can remove the limitations
of the Artificial Neural Network and can make it
5. Data processing, including filtering, more and more useful for the various kinds of
clustering, blind signal separation and applications. If the Artificial Neural Network
compression. concept is combined with the Computational
6. Scroll Web Pages, or in an eBook with Automata, FPGA and Fuzzy Logic we will
simple left and right hand gestures, this definitely solve some of the limitations of neural
is ideal when touching the device is a network technology.
barrier such as wet hands are wet, with
REFERENCES
gloves, dirty etc.
[1] Haykin S., “Neural Networks A
7. Application areas of ANNs include
Comprehensive Foundation”, 2nd edition,
system identification and control
Pearson Education, 1999.

International Journal of Emerging Engineering Research and Technology 147

Er.Parveen Kumar & Er.Pooja Sharma

[2] Ms. Sonali. B. Maind et. al., Research Parveen Kumar Student
Paper on Basic of Artificial Neural M.Tech, I passed my B.Tech
Network, International Journal on Recent
and Innovation Trends in Computing and ECE from Modern Institute of
Communication Volume: 2 Issue: 1 | Engineering & Technology
January 2014 Mohri, Kurukshetra, India in
[3] Vidushi et al., International Journal of 2012 and presently a student
Advanced Research in Computer Science of M.Tech ECE final Year at Galaxy Global
and Software Engineering 2 (10), October-
Group of Institutions Dinarpur, Ambala, India
2012, pp.278-284
[4] About Feed Back Network from website
http://www.idsia.ch/ ~juergen/rnn.html .
[5] Sucharita Gopal, “Artificial Neural
Networks for Spatial Data Analysis”,
Boston, 1988.
[6] Eldon Y. Li, “Artificial Neural Networks
and their Business Applications”, Taiwan,
1994. FLEXChip Signal Processor
(MC68175/D), Motorola, 1996.
[7] Christos Stergiou and Dimitrios Siganos,
“Neural Networks”.
[8] About Neural Network from website
http://en.wikipedia.org /
wiki/Neural_network .
[9] Girish Kumar Jha, “Artificial Neural
Network and its Applications”, IARI New
Delhi.
[10] Image of a Neuron form website
http://transductions.net/
2010/02/04/313/neurons/
[11] Ugur HALICI, “ Artificial Neural
Networks”, Chapter 1, ANKARA.
[12] Ajith Abraham, “Artificial Neural
Networks”, Stillwater,OK, USA, 2005.
[13] Lippmann, R.P., 1987. An introduction to
computing with neural nets. IEEE Accost.
Speech Signal Process. Mag., April: 4-22.
[14] Carlos Gershenson, “Artificial Neural
Networks for Beginners”, United kingdom.
AUTHORS’ BIOGRAPHY
Er.Pooja Sharma passed
B.Tech ECE from Kurukshetrs
University Kurukshetra in
2007 and then passed M.Tech
ECE from MM University
Mullana, Ambala,India and
She is presently working as an Assistant
professor at Galaxy Global Group of Institutions
Dinarpur, Ambala, India.

International Journal of Emerging Engineering Research and Technology 148