Copyright © 2018 Norma Alias et. al. This is an open access article distributed under the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
International Journal of Engineering & Technology, 7 (4) (2018) 3255-3262
International Journal of Engineering & Technology
Website: www.sciencepubco.com/index.php/IJET
doi: 10.14419/ijet.v7i4.15154
Research paper
A modeling of animal diseases through using artificial
neural network
Norma Alias
1
*, Fatin Naemah Mohd Farid
1
, Waleed Mugahed Al-Rahmi
2
, Noraffandy Yahaya
2
,
Qusay Al-Maatouk
3
1
Ibnu Sina Institute for Scientific &Industrial Research (ISI-SIR), Faculty of Science, Universiti Teknologi Malaysia, 81310, UTM
Skudai, Johor, Malaysia
2
Faculty of Education, Universiti Teknologi Malaysia, 81310, UTM Skudai, Johor, Malaysia
3
Faculty of Engineering, Computing and Technology, Asia Pacific University of Technology & Innovation (APU), Technology Park
Malay-sia, Bukit Jalil-57000 Kuala Lumpur, Malaysia
*Corresponding author E-mail: norm@ibnusina.utm.my
Abstract
This paper studied the implementation of Artificial Neural Network (ANN) where it well-known recently in veterinary disease research
field in Malaysia. The parameter identification under consideration is types of animal disease, types of species and locations of disease
based on the Geographical Information System (GIS) data set. There are many types of animal diseases that affect farm animals in Ma-
laysia. In this research, the method of multilayer perceptron neural network is used as main model since it is an effective solving method
in predicting the future of veterinary disease. ANN has ability to visual animal diseases involving the computational model. The model is
to present the rela-tionship between causes of the species and location and consequence of animal disease without emphasizing the pro-
cess, considering the initial and boundary condition and considering the nature of the relations. The data collection of animal disease is
considered as a large sparse data set. Therefore method of ANN is well suited for optimizing of the data, to train the data operational and
to predict the parameter identification of animal disease. The output layers of ANN are plotted in SPSS software for statistical solution
and MATLAB programming for sequential ANN implemented. The ANN will be compare to genetic algorithm for the performance and
effectiveness of the method. The numerical simulation of ANN helps in future prediction of animal disease based on the species and
location parameters.
Keywords: Mathematical Modeling; Artificial Neural Network (ANN); Geographical Information System (GIS).
1. Introduction
Neural network consist a number of simple neuron like processing
elements and a number of weighted connections between the ele-
ments. Those weights are on connections encode the network
(Volna, 2012). In this research, ANN is used to give an output to
predict animal disease, hence, it will help to predict the disease
based on location and species happen for future study. The basic
definition or concept can be seen through paper as written by
Krenker et al. in 2011 says that Artificial neural network is a com-
putational model that tries to account for the parallel nature of
human brain. Besides ANN is one of the most accurate and widely
used forecasting model applied in forecasting social, and econom-
ic (Al Shamisi et al., 2011, Alias et al., 2012). The characteristics
of ANN are fault tolerance, distributed memory, learning ability,
collective solution, weighted parameter and network structures
(Pintér, 2012). In machine learning and related fields, artificial
neural network (ANN) is a computational model inspired by a
central nervous system (neuron(s) in brain), and are used to esti-
mate functions that can depend on a large number of inputs and
are generally unknown (Martínez-Morales et al., 2014). The de-
velopment of neurobiology study enables researchers to create
mathematical models of neurons to stimulate neural behavior. Safi,
(2011) stated that, to train the network, we used the well-back
propagation algorithm (BP) of MLP, which consists of an optimi-
zation procedure aimed at minimizing the RMSE error observed at
the output layer. This algorithm uses a supervised learning mode,
meaning that the output corresponding to each input is a priori
known, which makes it possible to compute signal errors and try
to reduce them through epochs.
2. Multilayer perceptron (MLP) model
For multilayer perceptron, it contains several single layer percep-
tron. Those single layer perceptron were arranged according some
hierarchy. The hierarchy should follow the characteristics. Some
of the characteristics were inputs of first layer is taking with the
number of perceptron same the number of vectors X of the prob-
lem. Then, the output layer generate outputs with the number of
neuron equal to the desired number of quantities computed from
the inputs as well as there is one layer between single perceptron
with another (Morens and Fauci, 2013). Detailed description of
MLP architecture of ANN is published previously by Seyam and
Mogheir, (2011) where he was used identical approach to ANN
construction as well as its training or testing ratio. Neural Net-
works are able to learn because they can change the connection
weights between units. After learning, the knowledge is stored in
the weights. For the purpose of training, 70% of the staring dataset
is usually used. Training of the neural network is terminated when
the network has learned to generalize the underlying trends of