Agriculture Reference
In-Depth Information
neural network with two hidden layers to
predict hen and pullet egg production. The
model successfully learned the relationship
between input (hen age) and output (egg
production). The results suggested that the
ANN model could provide an effective means
of recognizing data patterns and accurately
predicting the egg production of laying hens
based on their age.
The group method of data handling
(GMDH) algorithm considers the responses to
polynomial regressions of all pairs obtained
from original data as inputs to multilayer
neural networks (Ivakhnenko, 1971). This ap-
proach has been successfully applied in sev-
eral fields, but it is rarely used in poultry
science. The results of Ahmadi
et al
. (2007)
indicated that a GMDH neural network is an
efficient means of recognizing patterns in data-
sets and in accurately predicting performa-
nce based on input investigation. It may also
be used to optimize broiler performance as a
function of nutritional factors.
Chicken mechanically separated meat
(MSM) is a raw material from chicken meat
processing derived from low-commercial
value chicken parts including back and neck
and it is produced using specific equipment,
such as deboning machines. Back-propagation
ANNs with five input layers (Ca, Fe, P, Mg, Zn),
a five-neuron hidden layer and one output
layer were trained to determine MSM content
in meat products. However, the application
of the networks to commercial samples (val-
idation) was inadequate because of the diffe-
rence between the ingredient composition
of the sausages used during training and the
commercial samples. The neural network
built to determine MSM content was effi-
cient during training and network testing
(Sousa
et al
., 2003).
Conclusions
The objective of this chapter was to intro-
duce the subject of neural networks, particularly
multi-layer ANNs. All fields of knowledge
now apply ANNs as powerful analysis tools,
and there are already many applications in
animal science, as shown in this chapter.
However, this number is still small. As an
emerging field in data analysis, the study of
ANNs has experienced exponential growth
in the past few decades. In the near future,
the human brain will be better understood
and new computer technologies will emerge,
allowing for the development of more so-
phisticated hybrid models of ANNs.
References
Ahmadi, H., Mottaghitalab, M. and Nariman-Zadeh, N. (2007) Group method of data handling-type
neural network prediction of broiler performance based on dietary metabolizable energy, methionine,
and lysine.
Journal of Applied Poultry Research
16,
494-501.
Akkurt, S, Ozdemir, S., Tayfur, G. and Akyol, B. (2003) The use of GA-ANNs in the modelling of com-
pressive strength of cement mortar.
Cement and Concrete Research
33,
973-979.
Anderson, J.A. (1983) Cognitive and psychological computation with neural models.
IEEE Transactions
on Systems Management and Cybernetics
13,
799-814.
Basheer, I.A. and Hajmeer, M. (2000) Artificial neural networks: fundamentals, computing, design, and
application.
Journal of Microbiological Methods
43,
3-31.
Chen, M.S. (1991) Analysis and design of the multilayer perceptron using polynomial basis functions.
PhD Dissertation, The University of Texas at Arlington, Texas, USA.
Fausett, L. (1994)
Fundamentals of Neural Networks: Architecture
,
Algorithms and Applications
.
Prentice-Hall, Englewood Cliffs, New Jersey, USA.
Galan-Marin, G. and Perez, J. (2001) Design and analysis of maximum Hopfield networks.
IEEE Trans-
actions on Neural Networks
12,
329-339.
Ghazanfari, S., Nobari, K. and Tahmoorespur, M. (2011) Prediction of egg production using artificial
neural network.
Iranian Journal of Applied Animal Science
1,
10-16.
Haykin, S. (2001)
Redes Neurais, Princípios e Prática, Second Edition
. Bookman, Porto Alegre, Rio
Grande do Sul, Brazil.
