E3S Web of Conferences 287, 03001 (2021) https://doi.org/10.

1051/e3sconf/202128703001
ICPEAM2020

A View of Artificial Neural Network Models in Different

Application Areas
Kumaravel ArulRaj 1,*, Muthu Karthikeyan2, and Deenadayalan Narmatha3
1Department of Mechanical Engineering, Einstein College of Engineering, Tirunelvelli, Tamilnadu, India.
2Department of Mechanical Engineering, Raja Rajeshwari College of Engineering, Bengaluru, Karnataka, India.
3Department of Electronics and Communication Engineering, Einstein College of Engineering, Tirunelvelli, Tamilnadu,
India.

Abstract. Neural network is a web of million numbers of inter-connected neurons which

executes parallel processing. An Artificial neural network is a nonlinear mapping structure;
an information processing pattern is stimulated by the approach as biological nervous system
(brain) process the information. It is used as a powerful tool for modeling the data in the
application domains where incomplete understanding of the data relationship to be solved
with the readily available trained data. The basic element for this processing pattern is the
structure of the data which is the collection of densely interconnected neurons to elucidate
the problems. A prominent part of these network is their adaptive nature to “learn by
example” just like human substitutes “programming” in resolving the problems. Through
learning process, neural net is designed for data classification and prediction where statistical
techniques and regression model have been employed. This report is an overview of artificial
neural networks in different application areas and it also illustrate the architecture structure
formed for the applications. It also provides information about the training algorithm used for
certain application.

Keywords: Artificial Neural Network and applications, data classification and prediction

1.Introduction furnished with other benefits as fault tolerance, self-

organization, real time process and adaptive learning.
Artificial neural network (ANN) as a branch
of artificial intelligence have established in a variety
of problem-solving areas. The rapid growth of
research activity in neural network applications has
been accompanied by advancement in the commercial
use of ANNs. The concept of artificial neural network
is familiarized from the biological neural network
which plays a vital role in human body. The human
brain comprises of million and million number of
neurons that performs a task when existed with the
stimulus, it produces an output into other neurons Fig.1 Artificial Neural Network Structure
through synaptic connection. Similarly the artificial
neurons perform a task by adding the inputs with a The main differences between Artificial Neural
transfer function in hidden layer and computing the Networks and Biological Neural Networks (BNN) are
total value as an output. Based on the structure of the described with some parameters as follows;
net, an ANN can be single layer or multilayer that is 1. ANN has few kinds of high-speed
input, feedback or hidden and output as shown in processing unit whereas BNN has simple
Fig.1. In ANN architecture, training or learning is and more than 100 kinds of low speed
termed as the programming for adjusting the weights. biological neurons is shown in Fig2.
The learning process can be done by given weights 2. The human brain has distributed memory
computed from trained data set or by fine- tuning the then integrated into processor while ANN
weights automatically. Trained neural network is having localized memory separated from
expertise in analyzing the information has been processor.
3. The computational method in biological

© The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution
License 4.0 (http://creativecommons.org/licenses/by/4.0/).
E3S Web of Conferences 287, 03001 (2021) https://doi.org/10.1051/e3sconf/202128703001
ICPEAM2020

neurons is occur in neurons, dendrites and Dase et al., presented a review of application
synapses with self-learning process in a of ANN for the prediction of stock market index
distributed method while in neural net which is a challenging task. ANN is capable of
computing the data in sequential and calculating precise stock index and it also extract the
centralized method. information from the big data thereby it predicts to
buy or sell or hold the stock market shares. [4]
4. The reliability of the biological network is Vishnu et al., learned that neural network
robust and in ANN is vulnerable. need not be reprogrammed. When the processing
5. The operating environment is element of the neural net fails, the network can
poorly defined and unconstrained in continue its function by parallel processing. ANN is
biological network and for neural net is well more prevalent in the prediction of outcome from the
defined environment. process parameters by appropriately training the data.
6. Error removal capability of human brain is [5]
stronger than an ANN by recognizing the Mitali et al., briefly explained the prediction
error easily. technique in neural network. This paper reports the
7. ANN knowledge can be replaceable different artificial neural network trainings which
whereas BNN knowledge is adaptable. minimize the mean square error and the steps used in
Matlab. ANN is used to increase the prediction
accuracy with minimal dependent on experimental
data and helps in optimal selection of machining
parameter for its optimization. [6]
Rani Pagariya et al., reported the artificial
neural network introduction, explained the types of
neural network and describe the benefits of neural
network with historical background. The
mathematical models for the back-propagation
algorithm are presented. This paper also investigates
the architecture of the neural network. [7]
Foram S. Panchal et al. emphasized that
that artificial neural network is more suitable for
Fig.2 Biological Neuron
pattern recognition, signal processing, forecasting,
To cover wide area of applications, artificial classification and data analysis. To get proper
intelligence will enable by integrating the artificial results, neural network is used. It needs trained data,
neural network with the expert systems. An ANN has selected architecture and preprocessed data but still
the self-learning capability, the requirement is to the performance is based on the network size. In the
provide data and train it; the outcome depends on the neural network, hidden node selection causes
structure of the network. The expert system overfitting or underfitting problem. This work finds
knowledge is restricted with the human knowledge the solution for the selection of hidden node issue by
domain; the outcome of this system is subject to how choosing the optimal number of hidden nodes based
the knowledge has been signified. Both systems have on similarities of input data after training the
their own characteristics but can cooperate among network with real data. [8]
them and for some issues they overlap in use. [1] Hong Hui Tan et al., explained about the
second order optimization technique. This technique
2. Literature Review
offers additional curvature information which
Rouse viewed that neural network is a web reduces the iteration in training the data and attains
of programs and structure of data which is similar to quick convergence by optimizing the step length in
the human brain function. A neural network training phase of neural network. It also reviews the
comprises of densely interconnected processing unit techniques involved in second order optimization as
in parallel each with its own domain of knowledge follows, Hessian calculation method is appropriate
and data access in its memory. [2] for network training and it is efficient by configuring
Kayri observed that in neural network, with conjugate gradient algorithm, Gauss-Newton
training algorithm plays an important role in and Levenberg- Marquardt method uses less
adjusting the weight and minimizing the mean square
memory, with the support of parallel processing
error between the experimental data and the predicted
computation power per iterations is reduced. The
data output. In the trained data, the error can be
feasibility and optimization technique performance
reduced to exact threshold level. In this paper,
Gradient descent algorithm is used to lessen the mean is summarized. [9]
square error. [3] Subhadra et al., proposed a system of

2
E3S Web of Conferences 287, 03001 (2021) https://doi.org/10.1051/e3sconf/202128703001
ICPEAM2020

predicting the heart disease with suitable diagnosis Alcohol abuse, Both right and left Arm and Leg
by using multilayer perceptron neural network. The weakness, Speech Slurring, Giddiness, Headache,
data is trained by the use of back propagation Vomiting, Memory Deficits, Swallowing Difficulties,
algorithm. The inputs of the network are age, sex, Vision loss, isolated vertigo, Hand & leg Numbness
chest, resting blood pressure, serum cholesterol, and Dizziness. The network architecture has 20 input
fasting blood sugar, resting electrocardiograph, nodes, 10 hidden nodes and 10 output nodes.
heart rate, exercise induced angina, sloping rate and
major vessels count. Genetic algorithm, K-mean
algorithm are commonly used data mining method
for predicting the heart disease. The performance
metrics such as sensitivity, specificity, precision
and accuracy are calculated for various approaches
and proved that the proposed system is proficient in
predicting the heart disease. [10]

2. Aspects and Applications of

Neural Networking
An artificial neural network is an approach
which provides the nearer results to human
perception and recognition than conventional
methods. The distinctive features make neural
networks valuable in a variety of applications. One of
the characteristics of ANN is that, for an incomplete
input, it can bring forth sensible outcomes.
1. Self-Organizing Traffic Control System

A four-layer ANN is used to approximate

optimal splits of signal phases in traffic control
system. The traffic control system is built on two
conditions as, constant cycle length over the road
network and no offsets between nearby joints. The
inputs are control variables such as signal phases split
lengths and on inflow links traffic volumes. The Fig.3 ANN Architecture for Stroke disease
output from the neural network are queue length and Prediction
performance measures. In the training process, by
altering the synaptic weights, the targeted outcomes The types of stroke diseases such as TIA,
were achieved. [11] Left Hemiplegia, Right Hemiplegia, Dysphasia,
Monoplegia, Left Hemianopia, Aphasia, Right
2. Thrombo Embolic stroke Prediction Hemianthesia, Dysphagia, Quadruplegia are the
output parameters (D1 to D10 in Fig3). It helps the
ANN exhibits better performance in the doctors to provide earlier diagnosis and better
stroke disease prediction by using back propagation medication to the patient. Neural network provides
algorithm to train the stroke data structure and it has the improved classification or predication of disease
been tested for the types of stroke disease viz. outcome with certain probability. [12]
cerebrovascular accident, ischemic stroke, embolic
stroke, thrombotic stroke and brain attack. Here 3. Shear Wave Velocity Prediction
neural network helps the doctor in decision making,
Shear wave velocity approximation is the
screening and to double check their diagnosis.
important parameter in the seismic exploration and
Backward stepwise feature selection method is used
representation of a carbonate reservoir. To optimize
for input feature selection.
the petro physical parameters multiple regression
The network structure for the prediction of
method is used. The fast training neural network can
stroke disease has the twenty input parameters (as
predict shear wave velocity from carbonate rocks
shown in Fig.3 X1 to X20) as Hypertension, Diabetes
porosity, density and compressional wave velocity.
mellitus, Myocardial infarction, cardiac Failure,
[13]
Atrial Fibrillation, Smoking, High Blood Cholesterol,

3
E3S Web of Conferences 287, 03001 (2021) https://doi.org/10.1051/e3sconf/202128703001
ICPEAM2020

4. Lung Cancer Diagnosing rainfall, snowmelt, evaporation and temperature. The

artificial neural network is the effective tool to
In carcinogenesis, neural network has its role handle the complex problem and has the ability to
in both pre-clinical and post-clinical diagnosis with a model rainfall runoff in the arid and semiarid areas
high precision rate which supports the clinicians to where the uneven downpours occur. The impacts of
detect the cancer at its early stage. The back training process and input dimension on rainfall
propagation algorithm is used for faster execution and runoff prediction competence of the neural network
for greater accuracy rate with the trained and test data were applied in many hydrological fields. [16]
set obtained from the Lung cancer patients. [14]
7. Data Mining
5. Cardiac Arrhythmias Detection
Data Mining is the process of extracting
Cardiac Arrhythmias is the heart disease
data from the larger database. Data mining can
shows a condition of atypical electrical activity in
estimate the upcoming trends and supports for
the heart. It can be detected by using
decision making. For a data mining task, some of the
electrocardiogram signal which is a graphical
features like simplicity of the architecture deigning
representation of the electrical activity of the heart.
and low computational complexity are achieved by
The neural network is trained by using Levenberg
using the artificial neural network. To improve the
Marquardt Back propagation algorithm to identify
artificial neural network performance, genetic
the disease with the trained dataset comprising the
algorithm is used in neural network design and
QRS complex features as minimum and maximum
implements the Neuro- fuzzy system. [17]
QRS width, moderate QRS width and the Heart rate
[15]. 8. Image Processing
The network architecture for the disease Artificial neural network is used to resolve
prediction consists of input layer consists of five the real-world problems in the image processing
neurons with tangent sigmoid transfer function and field. In an image processing, the most important
output layer comprises of six neurons which step is preprocessing which can be categorized into
represent the presence of diseases shown in Fig.4 image construction, restoration and enhancement.
with linear transfer function. Here artificial neural Neural network is directly applied to the image
network is used to identify and predict features from features and pixel data. Regression feed forward
highly abnormal ECG and it provides better network is used in preprocessing to reconstruct the
images, to suppress noise in image restoration and
proficiency for the heartbeats classification
act as an edge detectors in an image enhancement. In
according to different arrhythmias.
image compression application neural net perform
the encoding and in feature extraction applications
supervised and unsupervised networks are used to
extract the features. The optimization can be done by
Hopfield artificial neural network. [18]
9. Weather Prediction
Weather forecasting is to determine the state of
temperature in future at a given location. For the
temperature prediction, three layered back propagation
neural network technique is used. It is observed that
neural network model produces less error with high
efficiency compared with the numerical differentiation
method. [19]
Agricultural and industrial field requires
accurate weather forecasting which also alert about
natural disaster. The input data for the weather
prediction in the neural network consists of
temperature, air pressure, wind direction, cloudiness,
humidity and wind speed collected from
Fig. 4 Network Architecture
meteorological experts. The data are trained using
6. Rainfall-run off Prediction various techniques such as back propagation, optical
neural network, radial basis function, regression, and
Rainfall-runoff plays a major role in the fuzzy ARTMAP [20]
water resource management. The inputs for an ANN
model are the meteorological variables include 10. Civil Engineering

4
E3S Web of Conferences 287, 03001 (2021) https://doi.org/10.1051/e3sconf/202128703001
ICPEAM2020

Artificial neural networks have provided an emboli classification based on trans-cranial

effective tool for various engineering applications ultrasound signals, tissue temperature modeling
including wave-induced seabed instability based on imaging transducer’s raw data and
identification, earthquake- induced liquefaction and identification of ischemic cerebral vascular accident
tide forecasting. In Tidal level forecasting, to predict areas based on computer tomography images. In all
the tidal level Ann method is used. Similarly to predict the cases, ANN shows its high performance and
the wave induced and earthquake liquefaction potential makes it useful for easy diagnosis. [24]
ANN produces effective response. [21]
In near future ANN approach in medical field
11. Abrasive Wear Behavior Prediction plays a vital role during emergency situations. Due to
the environmental change, there will be possibility of
Artificial neural networks approach for the more diseases. But nowadays technology has its wide
prediction of abrasive wear behavior of carbon fabric growth, through that self-diagnosis may applicable.
reinforced epoxy composite shows that the data Implementing neural networks for medical reasoning
predictions are effective when compared with the provides solutions for common problems by
experimental data. To predict the wear properties, back identifying with the symptoms, recognizes the
propagation neural network architecture with different disease and give suggestions for first aid. [25]
training algorithm is used and compared the results in
which Levenberg Marquardt algorithm provides better ANN has its applications in diagnosis of urinary
results. system disease. Feed forward back propagation act as
a classifier to distinguish infected or non-infected
urinary system disease such as urinary bladder
inflammation and nephritis of renal pelvis. Based on
the symptoms, supervised neural networks classify
the infected and non-infected person. ANN shows
significant results in diagnosis of the disease. [26]

4. Conclusion
The fundamental concepts of neural
network, the similarities and differences of artificial
neural networks with the biological neural network
and expert systems were presented. This work reveals
the appropriateness and applicability of artificial
neural network techniques in various fields. Artificial
neural networks are intended to become the powerful
Fig.5 Network Structure for Wear Prediction tool in enhancing the state of artificial intelligence
applications. It also explains the training algorithm
In figure.5, Filter content, Applied Load and sliding and the network structure for various applications of
Distance are the input parameters and wear rate is the artificial neural network.
output parameter for the Ann architecture. The
performance measures such as Sum Squared error, References
Mean relative error, Mean square error and regression 1. Rajan Dharwal, Loveneet Kaur, IJST, ISSN
are observed for various algorithms. Trained Ann 0974-6846, Vol.9, (2016).
System helps in estimating the weight loss in the 2. M. Rouse, Neural Network Definition, available
abrasive wear of polymer composites. [22] at what’s.com Cson, accessed (2015).
3. Kayri Murat, JAMC, Vol. 21, Issue 20,
12. Power System
pp. 2- 11, (2016).
In multi-area power system, to calculate the transfer 4. R.K Dase, D.DPawar . IJMI, ISSN: 0975– 2927,
capabilities artificial neural networks model was 2, 2(2010).
explored. The obtained power transfer capability, 5. Vishnu D. Asal and R. I. Patel, IJSR, Vol.2 (3),
voltage magnitudes and voltage angles from repeated (2013).
power flow method is used for training the input and 6. Mitali S Mhatre , Fauzia Siddiqui ,
output patterns of the neural networks. Ann method MugdhaDongre , Paramjit Thakur, IJSER, ISSN
produces the effective result for the power transfer 2229-5518, 6, (2015).
capability [23] 7. Rani Pagariya, MahipBartere, IJAR, 4, 6,
(2013).
13.Medical Diagnosis 8. Foram S. Panchal, Mahesh Panchal, IJCSMC,
311, 455-464, (2014).
ANN can be used as a classifier and diagnosis tool in 9. Hong Hui Tan and King Hann Lim, IOP
medical application because of it intelligent Conference Series: Materials Science and
computational techniques. Some of the case studies Engineering495, 012003, (2019).
of ANN in medical applications includes blood flow 10. K. Subhadra., B.Vikas , (IJITEE) ISSN: 2278-

5
E3S Web of Conferences 287, 03001 (2021) https://doi.org/10.1051/e3sconf/202128703001
ICPEAM2020

3075, 8-5, (2019).

11. T. Nakatsuji and T. Kaku. In Transportation
Research Record 1324, TRB, National Research
Council, Washington, D.C., (1991).
12. D. Shanthi, G.Sahoo, N.Saravanan,IJBB, 3,1.

13. H. Eskandari., M.R.Rezaee, and.

Mohammadnia, CSEG RECORDER, (2004).
14. N.Ganesan, K. Venkatesh, M. A. Rama,
A. Malathi Palani, IJCA, 1,6, (2010).
15. Himanshu Gothwal, Silky Kedawat ,JBSE,4,
289-296, (2011).
16. Karim Solaimani, JAES, 5 (6): 856-865, (2009).
17. Nashaat El-KhamisyMohamed, Ahmed Shawky
Morsi El-Bhrawy, IOSR-JCE, e-ISSN: 2278-
0661,p-ISSN: 2278-8727, 18, , Ver. III (2016).
18. Rajesh G, A.Muthukumaravel, IJIRCCE,4, 8,
(2016)
19. Jyosthna Devi, Syam Prasad Reddy.B, Vagdhan
Kumar. K, Musala Reddy.B, Raja.N, IJETT,
3,1,(2012).
20. Subhajini A.C, IJWCCCR,4,1,8-18,(2018)
21. Jeng D.S, Cha D.H, Blumenstein.M, ICAIPSIR,
(2003).
22. Sudarshan Rao.K, Varadarajan Y.S, Rajendra N
(2014), IJEMS, vol 21,pp.16-22.
23. Palukuru Nagendra, sunita Halder nee Dey,
Tanaya Dutta (2010), LEJPT, ISSN 1583-1078,
Issue 16, p.119-128.
24. Maria Graca Ruano, Antonio E. Ruano (2012),
AISC 195, pp 433-451.
25. Sarfaraz Ahmed (2014) IJEDR, ISSN 2321-
9939.
26. Qeerthara Kadhim Al-Shayea, Itedal S.H.Bahia
(2010), IJCSNS, vol.10, No.7.
27. ArdeshirFaghri, Jiuyi Hua, Transportation
Research Record 1358, Department of Civil
Engineering, University of Delaware, Newark.
28. Furqan Yaqui, IJIRSET, 7, 1, (2018).
29. Mishra B.B, Dehuri .S, JCS3 (12); 948-955,
(2007).