MITS5002 Report: Artificial Neural Networks and EAs

This report provides a comprehensive review of a journal paper focused on artificial neural networks (ANNs) and evolutionary algorithms (EAs) within the context of software engineering methodologies. The report begins with an introduction to ANNs, defining their function and the report's objective: to critically analyze a research paper titled "Evolutionary artificial neural networks: a review." The paper, written in 2013, investigates the use of evolutionary algorithms to optimize ANNs. The report examines the paper's intention, content, research methods (including qualitative research design and secondary data collection), and data analysis techniques. Key findings highlight the potential of ANNs and EAs, along with the issues and limitations. The report concludes that evolutionary algorithms are an effective approach for enhancing ANN performance and addresses the challenges associated with implementation. Finally, it emphasizes the importance of understanding ANN and EA, as well as their limitations, to improve skills and knowledge in the field of artificial intelligence and methodologies.

Software Engineering Methodology

SOFTWARE ENGINEERING METHODOLOGY 1
Introduction
The term artificial neural network is defined as the computing system which constitutes
animal brains and performs various kinds of tasks. Mainly, artificial neural network processes
data or information by their dynamic state response to external inputs [1]. The objective of
this report is to examine the fundamental concept of an artificial neural network and evaluate
a journal paper based on the artificial neural networks. The title of the journal paper is
“Evolutionary artificial neural networks: a review which will be discussed in this report”. The
identification of this paper is to critically review the concept of artificial neural networks and
describe their working principle. This report is divided into several sections for example
intention and content of the paper, research methods, and findings, issues highlighted by the
researchers and results and discussion.
Intention and content of the paper
This journal paper was written by Shifei Ding and other authors in the year 2013 and it is
completely based on the artificial neural networks. The key purpose of this investigation is to
examine the utilization of evolutionary algorithms to optimize artificial neural networks [4].
As per the author's identification artificial neural networks are adaptive nonlinear information
system which combines the various kinds of a computing system with numbers of
characteristics, for example, self-organizing, self-adapting and so on. It is one of the common
technologies which are used in the field of medical but selection of the structure and
parameter is a very difficult task for the ANN system.
The authors argued that the benefits of the artificial neural networks are represented by the
network architecture and the source code. This journal paper mainly focuses on evolutionary
algorithms in order to optimize ANN networks [2]. According to the authors the artificial
neural network includes a set of processing unit which is also called as neurons and the
architecture of the neural network is connected with the neurons. It is observed that the
learning process in the artificial neural network is mainly implemented by training
programmes because the learning process is achieved by iteratively controlling the
connection weights [3]. From this paper, it has found that the major limitation of the artificial
neural network takes more time while adjusting the architecture with the neurons. It is
identified that the performance of the EAs may not be similar to the advanced source code for
several issues. But such kind of process can be used for processing numerous kinds of issues,

SOFTWARE ENGINEERING METHODOLOGY 2
for example, designing the network architecture, learning rules, and training of the
connection process and so on.
The researchers suggested that the evolution of the connection weights draws into a self-
adapting and global technique for training [4]. There are several problems included in the
field of ANN networks such as encoding system, find the fitness function, evolution
approach, train the network and so on. After reading the research paper it has found that the
EAs processes are utilized to select the effective input variables for neural networks from raw
data which is included in the input features [5].
Research methods
Research design
In this research paper, the authors used a qualitative research design while conduction the
research. Mainly, the qualitative research focuses on the theoretical information about the
artificial neural network and helped the authors for reducing research issues and challenges
occurred while conducting the investigation [6]. Therefore, by using such kind of research
design the author improved the quality of the research.
Research strategy
The authors used various kinds of research methods in this paper such as research design,
research approach, and data collection methods and data analysis techniques. In order to
collect the relevant data and information the authors developed and implemented several
strategies including philosophy of the research, previous studies, observation, experiments
and secondary research method. By using all these research strategies the researchers
produced a hypothesis of the research [7].
Data collection method
A secondary research method adopted by the authors in order to obtain the facts and
information about artificial neural networks. The secondary data about ANN networks
collected from various sources, for example, research papers, online websites, books and
other offline resources. With the help of secondary research method, the authors collected
facts and data about artificial neural networks and achieved aims and objectives of the study.

SOFTWARE ENGINEERING METHODOLOGY 3
Data analysis
The researchers used a descriptive content analysis process in order to analyse and evaluate
secondary data and information. Data analysis method has the ability to control and manage
the quality of research by providing effective information about the research topic [8]. By
using this research method the authors critically analysed the viewpoints of other authors and
gained an effective conclusion at the end of the investigation.
Findings
From this paper, it has found that the artificial neural network has the potential to combine
the various processing units with the real-time learning system. The evolutionary algorithm is
one of the best approaches for artificial neural networks that optimize the overall
performance of the system [9]. This research paper provided an in-depth analysis of ANN
networks and identified that EAs can be utilized to numerous kinds of issues such as
designing the architecture and extracting data from artificial neural networks.
Issues highlighted by the researchers
The authors highlighted that overfitting is one of the common problem associated with the
artificial neural networks which also impact on the performance of the system. It is very
complex to completely perform the evolutionary process for ANN networks and solving
practical issues is another problem highlighted by the authors. It is identified that the main
problem of EA is that it is less effective and performance may not be equal with the special
algorithms [10].
Results
After analysing this paper it has identified that ANNs and EAs both are the AI technologies
which borrow thoughts from behaviour characteristics and structure of the biological world.
EA is one of the effective techniques for ANNs in order to simulate the evolution
characterises of the biological aspects and helped the neural networks for managing the
system in an effective manner. Such kind of process has the potential to solve several
challenges such as the realization of ANN and monitoring the performance of neural
networks.

SOFTWARE ENGINEERING METHODOLOGY 4
Conclusion of the paper
This journal paper shows that fundamental principle of artificial neural networks and
provided complete information about evolutionary algorithms. It is concluded that the
evolutionary algorithm is an effective technique for improving the overall performance of
ANNs and also solve the designing and implementing related issues in the system. This paper
identified that evolutionary strategy emphasizes the change in the nature of individual which
is mainly included in the field of the evolutionary algorithm. In the future investigation the
researchers will improve the effectiveness of the paper by conducting a literature review and
provide information about applications of artificial neural networks along with their
limitations. This paper is related to the unit of study because it described the software
engineering methodology which is the major part of this unit.
Conclusion
From this study, it is concluded that the artificial neural network is a kind of process that has
the capability to connect numbers of neurons with each other. This report described and
reviewed the journal paper based on the artificial neural networks along with results and
findings of the article. It is identified that the authors provided in-depth analysis about ANNs
and EAs process by which the students can gain their skills and knowledge in the field of
artificial intelligence and methodologies.

SOFTWARE ENGINEERING METHODOLOGY 5
References
[1]. M.A., Ahmadi, M., Ebadi, A. Shokrollahi and, S.M.J., Majidi, “Evolving artificial
neural network and imperialist competitive algorithm for prediction oil flow rate of
the reservoir,” Applied Soft Compauting, vol. 13, no. 2, pp.1085-1098, 2013.
[2]. D.Z., Antanasijević, V.V., Pocajt, D.S., Povrenović, M.Đ. Ristić and, A.A., Perić-
Grujić, “PM10 emission forecasting using artificial neural networks and genetic
algorithm input variable optimization,” Science of the Total Environment, vol. 443,
no. 5, pp.511-519, 2013.
[3]. E., Asadi, M.G., da Silva, C.H., Antunes, L. Dias and, L., Glicksman, “Multi-
objective optimization for building retrofit: A model using genetic algorithm and
artificial neural network and an application,” Energy and Buildings, vol. 81, no. 6,
pp.444-456, 2014.
[4]. S., Ding, H., Li, C., Su, J. Yu and, F., Jin, “Evolutionary artificial neural networks: a
review,” Artificial Intelligence Review, vol. 39, no. 3, pp.251-260, 2013.
[5]. J. Hagenauer and, M., Helbich, “Mining urban land-use patterns from volunteered
geographic information by means of genetic algorithms and artificial neural
networks,” International Journal of Geographical Information Science, vol. 26, no. 6,
pp.963-982, 2012.
[6]. A., Nguyen, J. Yosinski and, J., Clune, “Deep neural networks are easily fooled:
High confidence predictions for unrecognizable images,” In Proceedings of the IEEE
conference on computer vision and pattern recognition, vol. 12, no. 6, pp. 427-436,
2015.
[7]. S., Oreski, D. Oreski and, G., Oreski, “Hybrid system with genetic algorithm and
artificial neural networks and its application to retail credit risk assessment,” Expert
systems with applications, vol. 39, no. 16, pp.12605-12617, 2012.
[8]. P. Tahmasebi and, A., Hezarkhani, “A hybrid neural networks-fuzzy logic-genetic
algorithm for grade estimation,” Computers & geosciences, vol. 42, no. 6, pp.18-27,
2012.
[9]. L., Wang, Y. Zeng and, T., Chen, “Back propagation neural network with adaptive
differential evolution algorithm for time series forecasting,” Expert Systems with
Applications, vol. 42, no. 2, pp.855-863, 2015.

SOFTWARE ENGINEERING METHODOLOGY 6
[10]. F. Yu and, X., Xu, “A short-term load forecasting model of natural gas based
on optimized genetic algorithm and improved BP neural network,” Applied
Energy, vol. 134, no. 6, pp.102-113, 2014.