JUITA 8 (1) - Compressed

09/03/22 06.
34 Editorial Team
Home (http://jurnalnasional.ump.ac.id/index.php/JUITA/index)
/ About the Journal (http://jurnalnasional.ump.ac.id/index.php/JUITA/about)
/ Editorial Team (http://jurnalnasional.ump.ac.id/index.php/JUITA/about/editorialTeam)
EDITORIAL TEAM
Editor in Chief
Hindayati Mustafidah, Teknik Informatika - Universitas Muhammadiyah Purwokerto, Indonesia
Editorial Board
Tatik Maftukhah, Puslit LIPI Kompleks Puspiptek Tangerang Selatan - Banten, Indonesia
Sri Kusumadewi, Universitas Islam Indonesia, Indonesia
Feri Wibowo, Universitas Muhammdiyah Purwokerto, Indonesia
Ardhi Wijayanto, Informatika, Universitas Sebelas Maret, Indonesia
Erick Alfons Lisangan, Teknik Informatika, Universitas Atma Jaya Makassar, Indonesia
Section Editors
Slamet Widodo, Universitas Muhammadiyah Purwokerto, Indonesia
Muhammad Zaeni Lathif, Universitas Muhammadiyah Purwokerto, Indonesia
Dimara Kusuma Hakim, Universitas Muhammadiyah Purwokerto, Indonesia
Ermadi Satriya Wijaya, Universitas Muhammadiyah Purwokerto, Indonesia
ISSN: 2579-8901
Reference Management Tool (https://www.mendeley.com/download-desktop/)
(https://www.mendeley.com/download-desktop/)
Peer-Reviewers (/index.php/JUITA/about/displayMembership/76)
Publication Ethics (/index.php/JUITA/pages/view/publicationethics)
Register (/index.php/JUITA/user/register)
Author Guidelines (/index.php/JUITA/about/submissions#authorGuidelines)
jurnalnasional.ump.ac.id/index.php/JUITA/about/editorialTeam 1/3
09/03/22 06.34 Editorial Team
(https://drive.google.com/open?id=1f1
Zx7sOwRBuMeigZ3h_JOBv4Ki)
Notifications
View (http://jurnalnasional.ump.ac.id/index.php/JUITA/notification)
Subscribe (http://jurnalnasional.ump.ac.id/index.php/JUITA/notification/subscribeMailList)
Article Template (https://docs.google.com/document/d/1JoUHP707SsdT_4s6_0tXX6pXiQTmJW-1/edit?usp=sharing&ouid=114837924174631254413&rtpof=true&sd=t
(https://docs.google.com/document/d/1JoUHP707SsdT_4s6_0tXX6pXiQTmJW-1/edit?usp=sharing&ouid=114837924174631254413&rtpof=true&sd=true)
(https://drive.google.com/file/d/1Sodj0Rxpn3mlwqVwO6XCT-ItPTwlyrHZ/view?usp=sharing)
User
Username
Password
Remember me
Journal Content
Search
Search Scope
All
Browse
By Issue (http://jurnalnasional.ump.ac.id/index.php/JUITA/issue/archive)
By Author (http://jurnalnasional.ump.ac.id/index.php/JUITA/search/authors)
By Title (http://jurnalnasional.ump.ac.id/index.php/JUITA/search/titles)
Other Journals (http://jurnalnasional.ump.ac.id/index.php/index)
(https://www.statcounter.com/)
09/03/22 06.34 Editorial Team
(http://statcounter.com/p11268461/summary/?guest=1)
Font Size
Information
For Readers (http://jurnalnasional.ump.ac.id/index.php/JUITA/information/readers)

For Authors (http://jurnalnasional.ump.ac.id/index.php/JUITA/information/authors)
Keywords
Android (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Android)
Aplikasi (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Apl
Augmented Reality (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Augmented%20Reality)
Beasiswa-PPA (http://jurnalnasional.ump.ac.id/index.php/JU
subject=Beasiswa-PPA)
Forward Chaining (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Forward%20Chaining)
Fuzzy Quantification Theory I
(http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Fuzzy%20Quantification%20Theory%20I)
Mamdani (http://jurnalnasional.ump.ac.id/index.php/JUIT
subject=Mamdani)
MySQL (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=MySQL)
Rekayasa Perangkat Lunak
(http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Rekayasa%20Perangkat%20Lunak)
SMS Gateway (http://jurnalnasional.ump.ac.id/index.php/JUITA/s
subject=SMS%20Gateway)
Sistem informasi (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Sistem%20informasi)
Tsukamoto
(http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Tsukamoto)
Web Service (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Web%20
backpropagation (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=backpropagation)
backward chaining (http://jurnalnasional.ump.ac.id/index.php/
subject=backward%20chaining)
decision support system (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=decision%20support%20system)
expert
(http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=expert%20system)
interest in learning (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?
subject=interest%20in%20learning)
learning achievement (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=learning%20achievement)
learning mot
(http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=learning%20motivation)
motivasi belajar (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?
subject=motivasi%20belajar)
JUITA : Jurnal Informatika e-ISSN: 2579-8901 p-ISSN: 2086-9398
(https://scholar.google.co.id/citations?user=l2ufArUAAAAJ&hl=en) (https://doaj.org/toc/2579-8901)
(http://id.portalgaruda.org/?ref=browse&mod=viewjournal&journal=624) (http://sinta2.ristekdikti.go.id/journals/detail?id=1153)
(https://search.crossref.org/?q=2579-8901)
(http://creativecommons.org/licenses/by/4.0/)
JUITA : Jurnal Informatika is licensed under a Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses
09/03/22 06.35 JUITA Vol. 8 Nomor 1, Mei 2020
Home (http://jurnalnasional.ump.ac.id/index.php/JUITA/index)
/ Archives (http://jurnalnasional.ump.ac.id/index.php/JUITA/issue/archive)
/ JUITA Vol. 8 Nomor 1, Mei 2020 (http://jurnalnasional.ump.ac.id/index.php/JUITA/issue/view/436)
JUITA VOL. 8 NOMOR 1, MEI 2020
JUITA
Cover (http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/9069) PDF (http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/9

JUITA: Jurnal Informatika
Pengantar Redaksi dan Daftar Isi

(http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/9070) (http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/9
JUITA: Jurnal Informatika
Articles
Boyer-Moore String Matching Algorithm and SHA512 Implementation for Jpeg/exif File Fingerprint
Compilation in DSA (http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/4413) (http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/
10.30595/juita.v8i1.4413 (http://dx.doi.org/10.30595/juita.v8i1.4413)
Rachmad Fitriyanto, Anton Yudhana, Sunardi Sunardi
Market Basket Analysis dengan Algoritma Apriori pada Ecommerce Toko Busana Muslim PDF (Bahasa
Trendy(Market Basket Analysis with Apriori Algorithms in Ecommerce Trendy Muslim Clothing (http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/
Stores) (http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/4550)
Aji Setiawan, Rizka Mulyanti
Pembangunan Aplikasi Identifikasi Typographical Error Dokumen Berbahasa Indonesia PDF (Bahasa
Menggunakan Algoritma Jaro-Winkler Distance (http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/
(http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/6344)
Grelly Lucia Yovellia Londo, Yohanes Sigit Purnomo W.P., Martinus Maslim
Performance of Levenberg-Marquardt Algorithm in Backpropagation Network Based on the Number

of Neurons in Hidden Layers and Learning Rate (http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view
Hindayati Mustafidah, Suwarsito Suwarsito
Deteksi Penyakit Malaria Menggunakan Convolutional Neural Network Berbasis Saliency PDF (Bahasa
(http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/6671) (http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/
Yohannes Yohannes, Siska Devella, Kelvin Arianto
Rekomendasi Makanan untuk Ibu Hamil Menggunakan Algoritma Genetika(Food Recommendations PDF (Bahasa
for Pregnant Women Using Genetic Algorithms) (http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/
Nadya Satya Handayani, Sri Kusumadewi, Edi Fitriyanto
Implementasi Algoritma Kunci Matriks Untuk Keamanan Data Akademik PDF (Bahasa
Joko Soebagyo, Imay Kurniawan
jurnalnasional.ump.ac.id/index.php/JUITA/issue/view/436 1/4
Analisis Bukti Digital TRIM Enable SSD NVMe Menggunakan Metode Static Forensics PDF (Bahasa
Imam Riadi, Sunardi Sunardi, Abdul Hadi
Image Tracking Berbasis Augmented Reality Untuk Relasi Antara Sunnah Nabi dan Manfaat Secara PDF (Bahasa
Medis (http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/5541) (http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view
Febrian Wahyu Christanto, Muhammad Fauzan Affandi, Titin Winarti
Implementasi dan Analisa SARDrive Sebagai Media Penyimpanan Cloud PDF (Bahasa
Muhammad Khairul Anam, Habibi Ulayya
Perancangan dan Pembuatan Sistem Pelaporan Kinerja Berbasis Online untuk Pranata Komputer PDF (Bahasa
Imam Teguh Islamy, Hanim Maria Astuti, Radityo Prasetianto Wibowo
Monitoring Status Gizi Balita Secara Online PDF (Bahasa

Deniar Risma Putri, Endah Sudarmilah
Development of Android Based Online Nurse Test Preparation

Ambar Relawati, Guntur Maulana Zamroni
Prediksi Kelompok UKT Mahasiswa Menggunakan Algoritma K-Nearest Neighbor PDF (Bahasa
Sukamto Sukamto, Yanti Adriyani, Rizka Aulia
ISSN: 2579-8901
Reference Management Tool (https://www.mendeley.com/download-desktop/)
(https://www.mendeley.com/download-desktop/)
Peer-Reviewers (/index.php/JUITA/about/displayMembership/76)
Publication Ethics (/index.php/JUITA/pages/view/publicationethics)
Register (/index.php/JUITA/user/register)
Author Guidelines (/index.php/JUITA/about/submissions#authorGuidelines)
(https://drive.google.com/open?id=1f1
Zx7sOwRBuMeigZ3h_JOBv4Ki)
Notifications
View (http://jurnalnasional.ump.ac.id/index.php/JUITA/notification)
Subscribe (http://jurnalnasional.ump.ac.id/index.php/JUITA/notification/subscribeMailList)
Article Template (https://docs.google.com/document/d/1JoUHP707SsdT_4s6_0tXX6pXiQTmJW-1/edit?usp=sharing&ouid=114837924174631254413&rtpof=true&sd=t
(https://docs.google.com/document/d/1JoUHP707SsdT_4s6_0tXX6pXiQTmJW-1/edit?usp=sharing&ouid=114837924174631254413&rtpof=true&sd=true)
(https://drive.google.com/file/d/1Sodj0Rxpn3mlwqVwO6XCT-ItPTwlyrHZ/view?usp=sharing)
User
Username
Password
Remember me
Journal Content
Search
Search Scope
All
Browse
By Issue (http://jurnalnasional.ump.ac.id/index.php/JUITA/issue/archive)
By Author (http://jurnalnasional.ump.ac.id/index.php/JUITA/search/authors)
By Title (http://jurnalnasional.ump.ac.id/index.php/JUITA/search/titles)
Other Journals (http://jurnalnasional.ump.ac.id/index.php/index)
Font Size
Information
For Readers (http://jurnalnasional.ump.ac.id/index.php/JUITA/information/readers)

For Authors (http://jurnalnasional.ump.ac.id/index.php/JUITA/information/authors)
Keywords
Android (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Android)
Aplikasi (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Apl
Augmented Reality (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Augmented%20Reality)
Beasiswa-PPA (http://jurnalnasional.ump.ac.id/index.php/JU
subject=Beasiswa-PPA)
Forward Chaining (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Forward%20Chaining)
Fuzzy Quantification Theory I
(http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Fuzzy%20Quantification%20Theory%20I)
Mamdani (http://jurnalnasional.ump.ac.id/index.php/JUIT
subject=Mamdani)
MySQL (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=MySQL)
Rekayasa Perangkat Lunak
(http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Rekayasa%20Perangkat%20Lunak)
SMS Gateway (http://jurnalnasional.ump.ac.id/index.php/JUITA/s
subject=SMS%20Gateway)
Sistem informasi (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Sistem%20informasi)
Tsukamoto
(http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Tsukamoto)
Web Service (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=Web%20
backpropagation (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=backpropagation)
backward chaining (http://jurnalnasional.ump.ac.id/index.php/
subject=backward%20chaining)
decision support system (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=decision%20support%20system)
expert
(http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=expert%20system)
interest in learning (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?
subject=interest%20in%20learning)
learning achievement (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=learning%20achievement)
learning mot
(http://jurnalnasional.ump.ac.id/index.php/JUITA/search?subject=learning%20motivation)
motivasi belajar (http://jurnalnasional.ump.ac.id/index.php/JUITA/search?
subject=motivasi%20belajar)
JUITA : Jurnal Informatika e-ISSN: 2579-8901 p-ISSN: 2086-9398
(https://scholar.google.co.id/citations?user=l2ufArUAAAAJ&hl=en) (https://doaj.org/toc/2579-8901)
(http://id.portalgaruda.org/?ref=browse&mod=viewjournal&journal=624) (http://sinta2.ristekdikti.go.id/journals/detail?id=1153)
(https://search.crossref.org/?q=2579-8901)
(http://creativecommons.org/licenses/by/4.0/)
JUITA : Jurnal Informatika is licensed under a Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses
JUITA: Jurnal Informatika e-ISSN: 2579-8901; Volume 8, Nomor 1, Mei 2020
Performance of Levenberg-Marquardt Algorithm

in Backpropagation Network Based on the Number
of Neurons in Hidden Layers and Learning Rate
Hindayati Mustafidah1, Suwarsito2
1
Informatic Engineering, Universitas Muhammadiyah Purwokerto
2
Geography Education, Universitas Muhammadiyah Purwokerto
Purwokerto, Central Java, 53182, Indonesia
1fida.mustafidah@ump.ac.id
2suwarsito.ito@ump.ac.id
Abstract - One of the supervised learning paradigms in Artificial Neural Networks (ANN) are biologically
artificial neural networks (ANN) that are in great developed inspired computational models. ANN consists of several
is the backpropagation model. Backpropagation is a processing elements (neurons) and there is a relationship
perceptron learning algorithm with many layers to change between neurons that will transform information
weights connected to neurons in hidden layers. The
received by one neuron to another neuron. This
performance of the algorithm is influenced by several
network parameters including the number of neurons in the
relationship is called weight. Deboeck and Kohonen
input layer, the maximum epoch used, learning rate (lr) describe ANN as a collection of mathematical techniques
value, the hidden layer configuration, and the resulting error that can be used for signal processing, forecasting and
(MSE). Some of the tests conducted in previous studies grouping, and are referred to as non-linear, multi-layered
obtained information that the Levenberg-Marquardt parallel regression techniques [2]. ANN as one of the
training algorithm has better performance than other main components of forming soft computing have been
algorithms in the backpropagation network, which produces widely applied in various fields of human life both for
the smallest average error with a test level of α = 5% which the purposes of research and solving technical problems
used 10 neurons in a hidden layer. The number of neurons such as forecasting, diagnostics, and pattern recognition
in hidden layers varies depending on the number of neurons
[3], [4].
in the input layer. In this study an analysis of the
performance of the Levenberg-Marquardt training
Backpropagation is the most widely used type of
algorithm was carried out with 5 neurons in the input layer, learning paradigm with or without supervision in ANN,
a number of n neurons in hidden layers (n = 2, 4, 5, 7, 9), especially in developing systems to solve problems.
and 1 neuron in the output layer. Performance analysis is Systems known to have used backpropagation have been
based on network-generated errors. This study uses a mixed studied to detect intrusions in the banking system [5] and
method, namely development research with quantitative and to estimate the longitudinal velocity fields at open
qualitative testing using ANOVA statistical tests. Based on channel junctions [6]. In other cases, backpropagation as
the analysis, the Levenberg-Marquardt training algorithm a multilayer perceptron was used in simulating the
produces the smallest error of 0.00014 ± 0.00018 on 9 characteristics of open channel bends and subsequently
neurons in hidden layers with lr = 0.5.
used in prediction of flow parameters in 90° open
Keywords: hidden layer, backpropogation, MSE, learning
channel arches [7], [8]. The network structure in this
rate, Levenberg-Marquardt paradigm uses more than one layer (multi-layer) to
change the weight associated with neurons in the hidden
I. INTRODUCTION layer. Learning for ANN is a process in which free
parameters of ANN are adapted through a continuous
Soft computing has come as an impact of the stimulation process by the environment in which the
development of computer science technology which is an network is located [9]. ANN learns from its experience.
approach technique in solving problems [1]. Soft The usual learning process includes three tasks, namely:
computing is part of an intelligent system which is a 1) network output, 2) comparing the output with the
model approach to computation by imitating human desired target, and 3) adjusting the weight and repeating
reason and has the ability to reason and learn in an the process.
environment filled with uncertainty and inaccuracy.
Performance of Levenberg-Marquardt … | Mustafidah, H., Suwarsito, 29 – 35 29

There are 12 training algorithms in the but the network performance slows down even though
backpropagation model that can be used [10], namely the the network speed in carrying out the training process is
Fletcher-Reeves Update algorithm, Polak-Ribiere, also influenced by the learning rate (lr) value used.
Powell-Beale Restarts, Scaled Conjugate Gradient, Information about the number of neurons in the hidden
Gradient Descent with Momentum and Adaptive layer that has the most optimal performance is unknown.
Learning Rate, Resilent Backpropagation, BFGS, One Therefore, in this study an analysis and testing of the
Step Secant, Levenberg-Marquardt. Some researches performance of the Levenberg– Marquardt training
related to the application of this training algorithm are algorithm was conducted based on variations in the
[11]; [12]; [13]; [14]; [15]. Up to this stage a training number of neurons in hidden layers and learning rate (lr).
algorithm has been implemented to help solve a case and
has not yet been tested for other training algorithms. II. METHOD
Further testing is carried out by [16]; [17]; [18]; [19];
[20]. The testing was conducted on the twelve training This research is a mixed method research in the form
algorithms and generated information that the of developing computer programs with quantitative and
Levenberg-Marquardt algorithm is the most optimal qualitative testing using ANOVA statistical tests.
algorithm using 5, 10, and 15 neurons in the input layer.
In the study 10 neurons were used in hidden layers. A. Research Variables
Meanwhile, the number of neurons in the hidden layer is
very influential on network performance, especially in The research variables in the form of ANN
the error or MSE (Mean Squared Error) produced which parameters are the maximum epoch of 1000 (103), the
has an impact on the level of accuracy of network output. value of lr = 0.01, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7,
MSE is known as a method that produces errors that are 0.8, 0.9, 1, target error = 0.001 (10-3), input neurons (X)
likely to be better for small errors, but sometimes make as many as 5, and 1 output neuron (Y), as well as 2, 4, 5,
a big difference [21]. In theory, the more neurons in a 7, 9 neurons in the hidden layer (Z). Network
hidden layer the more accurate the output is produced, configuration is presented in Fig. 1.
Fig. 1 Design of artificial neural networks with 5 input neurons, n neurons in

hidden layers (n = 2, 4, 5, 7, 9) and 1 neuron in the output layer
30 Performance of Levenberg-Marquardt … | Mustafidah, H., Suwarsito, 29 – 35

B. Research Data 1) Determine the hypothesis

H0: there is no difference in error produced by the
Network input data and targets are acquired from
Levenberg-Marquardt algorithm on the
research [16].
number of n neurons in hidden layers (n = 2,
C. Development of Computer Programs 4, 5, 7, 9) for each value of lr.
H1: there are differences in errors generated by the
The design of a computer program to obtain network
Levenberg-Marquardt algorithm on the
output data is built as shown in Fig. 2.
number of n neurons in hidden layers (n = 2,
D. Data Analysis 4, 5, 7, 9) for each value of lr
2) Determine the alpha value (α) (in this study used
Output data of network generated by the Levenberg-
α = 5%)
Marquardt algorithm were analyzed using ANOVA
3) Taking conclusions. Conclusions are taken
statistical tests. Tests were carried out on many neurons
based on the significant value obtained (sig.) With the
in hidden layers (n = 2, 4, 5, 7, 9) at each learning rate
provision H0 rejected if the sig value<α and the
(Fig. 3). Furthermore, from the results of this test, it is
provisions of H0 are accepted if the value of sig≥α.
analyzed again to get the smallest MSE. The stages of
the ANOVA test were carried out [22]:
Fig. 2 Flowchart of ANN program development
Fig. 3 Testing statistics design of many neurons n in hidden layers (n = 2, 4,

5, 7, 9) at each learning rate

III. RESULTS AND DISCUSSION in Figure 2. The computer program was coded with
MATLAB as in Fig. 4.
A. Research Data
B. Data Analysis
Network input data (X) is the value of 5 neurons in
the input layer and target (Y) are random data acquired ANOVA statistical tests were performed using SPSS
from the research of [16] as in Table 1. Data input and software. The test results for errors generated by the
target of network are run on the Levenberg-Marquardt Levenberg-Marquardt algorithm on the number of n
algorithm to obtain MSE data. The Levenberg- neurons in the hidden layer for each value of lr with n =
Marquardt algorithm is run 20 times for each number of 2, 4, 5, 7, 9 are presented in Table 2, 3, 4, 5, and Table 6
n neurons in the hidden layer and every lr as the design respectively.
TABLE I
INPUT AND TARGET DATA
X1 X2 X3 X4 X5 Y
9.5013 7.6210 6.1543 4.0571 0.5789 2.0277
2.3114 4.5647 7.9194 9.3547 3.5287 1.9872
6.0684 0.1850 9.2181 9.1690 8.1317 6.0379
4.8598 8.2141 7.3821 4.1027 0.0986 2.7219
8.9130 4.4470 1.7627 8.9365 1.3889 1.9881
Fig. 4 Source code of ANN computer program
TABLE II
RESULTS OF ANOVA TEST ON 2 NEURONS IN A HIDDEN LAYER
Sum of Squares df Mean Square f Sig.
Between Groups 5.657 11 .514 .547 .870
Within Groups 214.404 228 .940
Total 220.061 239

TABLE III
Sum of Squares df Mean Square F Sig.
Between Groups 1.743 11 .158 1.052 .401
Total 36.070 239
TABLE IV
Between Groups .494 11 .045 .858 .583
Total 12.444 239
TABLE V
Between Groups .001 11 .000 1.575 .107
Within Groups .009 228 .000
Total .010 239
TABLE VI
Between Groups .000 11 .000 1.202 .286
Within Groups .000 228 .000
Total .000 239
Based on Tables 2, 3, 4, 5, and Table 6, there are 5 number of neurons in HL for each value of lr used. While
Sig. all of which are ≥ α (= 5%) so that H0 is accepted. in research of [23], testing is performed on each value of
In accordance with the proposed hypothesis, there is no lr used for each number of neurons in HL.
significant difference in MSE for each n neuron in the The MSE difference that occurs is suspected to be a
hidden layer (n = 2, 4, 5, 7, 9) based on the learning rate. correlation between the value of lr and MSE. Therefore
However, the average MSE value generated by the a correlation test is performed using the Pearson method
Levenberg-Marquardt algorithm for each of the n and produce data as in Table 8. From Table 8 can be seen
neurons in each learning rate (lr) can be known through that the correlation between learning rate (lr) and MSE is
descriptive analysis. The results of the data description -0.048. This means that the correlation between lr and
are presented in Table 7. MSE is very small and inversely correlated. The greater
Table 7 shows the difference in the smallest error rate the value of lr, the smaller the MSE. Because the value
for each value of lr and the number of neurons in the of sig.> α (= 5%), it can be said that there is no significant
hidden layer. The data in blue in the table shows the correlation between lr and MSE. This is in line with the
smallest MSE value for each number of neurons in HL at results of research by [24] which states that there is no
the corresponding lr value. Overall, the smallest error correlation between MSE and lr in backpropagation
(MSE) was achieved on 9 neurons in the hidden layer networks using 10 neurons in hidden layers.
with learning rate = 0.5. The MSE value is 0,00014 ± In the studies mentioned, the Levenberg - Marquardt
0,00018. This result is in line with research conducted by algorithm provides the smallest MSE value compared to
[23] which gives the smallest MSE value achieved by the other training algorithms. This is reasonable because the
LM algorithm of 0.00019584038 ± 0.000239300998. algorithm uses a Newtonian method that is very fast and
The MSE results were achieved using a different test accurate to get the minimum error [10].
direction. In this study, testing was carried out on the

TABLE VII
MEAN OF MSE FOR EACH LEARNING RATE (LR) FOR N NEURONS IN A HIDDEN LAYER (N = 2, 4, 5, 7, 9)
2 neurons 4 neurons 5 neurons 7 neurons 9 neurons
No. lr
Mean ± stdev. Mean ± stdev. Mean ± stdev. Mean ± stdev. Mean ± stdev.
1 0.01 0.74591±1.12645 0.00844±0.01963 0.08880±0.36589 0.00040±0.00031 0.00038±0.00028
2 0.05 0.49334±0.95028 0.08763±0.36596 0.00294±0.01202 0.00596±0.01814 0.00016±0.00021
3 0.1 0.66188±0.98916 0.08855±0.36588 0.00029±0.00035 0.00016±0.00016 0.00024±0.00032
4 0.2 0.50436±0.93995 0.12977±0.54840 0.00027±0.00032 0.00020±0.00028 0.00023±0.00030
5 0.3 0.67330±1.04447 0.12517±0.49946 0.00295±0.01201 0.00290±0.01203 0.00023±0.00023
6 0.4 0.72417±1.09239 0.01342±0.02758 0.00407±0.01744 0.00017±0.00028 0.00023±0.00023
7 0.5 0.55946±0.95146 0.12582±0.54907 0.00016±0.00022 0.00008±0.00014 0.00014±0.00018
8 0.6 0.59960±1.02243 0.00803±0.02399 0.00303±0.01200 0.00027±0.00030 0.00028±0.00034
9 0.7 0.47074±0.92942 0.00416±0.01742 0.11422±0.49708 0.00027±0.00032 0.00018±0.00021
10 0.8 0.65068±0.99104 0.09129±0.36541 0.11422±0.49708 0.00028±0.00032 0.00020±0.00026
11 0.9 0.54508±0.93392 0.00931±0.02250 0.00299±0.01200 0.00028±0.00032 0.00030±0.00032
12 1 0.14428±0.54541 0.31443±0.74240 0.00024±0.00027 0.00017±0.00022 0.00025±0.00030
TABLE VIII
CORRELATIONS TEST BETWEEN LEARNING RATE VALUES AND MSE
learning rate MSE
learning rate Pearson Correlation 1 -,048
Sig. (2-tailed) ,715
N 60 60
MSE Pearson Correlation -,048 1
Sig. (2-tailed) ,715
N 60 60
IV. CONCLUSION REFERENCES

Based on the results of the research that has been [1] S. Kusumadewi and S. Hartati, Neuro-Fuzzy : Integrasi
done, it can be concluded that the Levenberg–Marquardt Sistem Fuzzy dan Jaringan Syaraf. Yogyakarta: Graha
training algorithm has the best performance when using Ilmu, 2006.
9 neurons in the hidden layer and lr = 0.5. This [2] S. Shanmuganathan and S. Samarasinghe, Artificial
performance is indicated by the MSE value of 0.00014 ± Neural Network Modelling, vol. 628. Springer
0.00018 from the target error 0.001. With information International Publishing, 2016.
generated from this study, the Levenberg–Marquardt [3] M. T. T. Jones, Artificial Intelligence A Systems
training algorithm can be used as an alternative in the Approach. New Delhi: Infinity Science Press LLC, 2008.
development of ANN-based applications. [4] J. J. Siang, Jaringan Syaraf Tiruan dan
Pemrogramannya Menggunakan MATLAB. Yogyakarta:
ACKNOWLEDGEMENTS ANDI, 2009.
[5] A. B. Adetunji, A. Q. Ayinde, and C. O. Akanbi,
Authors thank DP2M - The Ministry of Research, “Application of Neural Network to Detect Intrusion in
Technology, and Directorate General of Higher Banking System.,” Am. J. Sci. Ind. Res., vol. 5, no. 2, pp.
Education (DIKTI) through KOPERTIS Region VI, who 53–59, 2014.
has provided funds and Universitas Muhammadiyah [6] A. H. Zaji and H. Bonakdari, “Application of artificial
Purwokerto for providing facilities in the neural network and genetic programming models for
implementation of this research. estimating the longitudinal velocity field in open channel
junctions,” Flow Meas. Instrum., vol. 41, pp. 81–89,

2015. Neural Network,” in Proceeding International

[7] A. Gholami, H. Bonakdari, A. H. Zaji, and A. A. Akhtari, Conference on Education, Technology, and Science
“Simulation of open channel bend characteristics using (NETS) 2013, “Improving The Quality Of Education To
computational fluid dynamics and artificial neural Face The Impact Of Technology”. December 28th, 2013,
networks,” Eng. Appl. Comput. Fluid Mech., vol. 9, no. 2013.
1, pp. 355–369, 2015. [16] H. Mustafidah and S. Suwarsito, “Error Rate Testing of
[8] A. Gholami, H. Bonakdari, A. H. Zaji, S. Ajeel Fenjan, Training Algorithm in Back Propagation Network,” Int.
and A. A. Akhtari, “Design of modified structure multi- J. Soft Comput. Eng., vol. 5, no. 4, pp. 46 – 50, 2015.
layer perceptron networks based on decision trees for the [17] H. Mustafidah and S. Suwarsito, “Model Parameter
prediction of flow parameters in 90° open-channel Jaringan Syaraf Tiruan untuk Pemilihan Algoritma
bends,” Eng. Appl. Comput. Fluid Mech., vol. 10, no. 1, Pelatihan Jaringan Backpropagation yang Paling
pp. 193–208, 2016. Optimal,” Purwokerto, Central Java, Indonesia, 2015.
[9] A. Desiani and M. Arhami, Konsep kecerdasan buatan. [18] H. Mustafidah and S. Suwarsito, “Uji Keoptimalan
Yogyakarta: Andi Offset, 2006. Algoritma Pelatihan pada Jaringan Syaraf Tiruan,” in
[10] S. Kusumadewi, Membangun Jaringan Syaraf Tiruan Prosiding Seminar Nasional SENATKOM 2015, 2015,
Menggunakan MATLAB & EXCEL LINK. Yogyakarta: pp. 243–248.
Graha Ilmu, 2004. [19] H. Mustafidah and S. Suwarsito, “Inferensi Tingkat
[11] H. Harjono and D. Aryanto, “Application of Artificial Kesalahan dalam Jaringan Backpropagation Berdasarkan
Neural Networks to Predict Student Achievement Laju Pemahaman,” in Prosiding Seminar Nasional
Study,” SAINTEKS, vol. 5, no. 2, 2009. Aptikom 2016 (SEMNASTIKOM), Hotel Lombok Raya
Mataram, 28 – 29 Oktober 2016, 2016, pp. 576–580.
[12] H. Mustafidah, D. K. Hakim, and S. Sugiyanto, “Tingkat
Keoptimalan Algoritma Pelatihan pada Jaringan Syaraf [20] H. Mustafidah and H. Harjono, “Korelasi Tingkat
Tiruan (Studi Kasus Prediksi Prestasi Belajar Kesalahan dan Epoh dalam Jaringan Backpropagation,”
Mahasiswa) Optimization Level of Training Algorithms in Prosiding SEMNASTIKOM 2017, 3 November 2017,
in Neural Network (Case Studies of Student Learning ISBN: 978-602-50434-0-6, 2017, pp. 55–61.
Achievement Predictions),” JUITA, vol. II, no. 3, pp. [21] E. L. Lehmann and G. Casella, Theory of Point
159–166, 2013. Estimation, Springer t. Springer, 2003.
[13] H. Mustafidah, D. Aryanto, and D. K. Hakim, “Uji [22] T. Taniredja and H. Mustafidah, Penelitian Kuantitatif
Optimalisasi Algoritma Pelatihan Conjugate Gradient (Sebuah Pengantar). Bandung: ALFABETA, 2011.
pada Jaringan Syaraf Tiruan,” in Prosiding SENATEK, [23] H. Mustafidah, Suwarsito, and S. N. C. Permatasari,
ISBN: 978-602-14355-0-2, 21 September 2013, 2013, p. “Accuracy of the neurons number in the hidden layer of
B-9-1. the levenberg-marquardt algorithm,” Int. J. Recent
[14] F. Wibowo, S. Sugiyanto, and H. Mustafidah, “Tingkat Technol. Eng., vol. 8, no. 4, pp. 2349–2353, 2019.
Ketelitian Pengenalan Pola Data pada Algoritma [24] H. Mustafidah and S. Suwarsito, “Correlation Analysis
Pelatihan Perbaikan Metode Batch Mode dalam Jaringan Between Error Rate of Output and Learning Rate in
Syaraf Tiruan,” JUITA (Jurnal Inform., vol. II, no. 4, pp. Backpropagation Network,” Adv. Sci. Lett., vol. 24, no.
259 – 264, 2013. 12, pp. 9182–9185, 2018.
[15] H. Mustafidah, S. Hartati, R. Wardoyo, and A. Harjoko,
“Prediction of Test Items Validity Using Artificial

36

JUITA 8 (1) - Compressed

Uploaded by

Copyright:

Available Formats

JUITA 8 (1) - Compressed

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

JUITA 8 (1) - Compressed

Uploaded by

Copyright:

Available Formats

09/03/22 06.

Reference Management Tool (https://www.mendeley.com/download-desktop/)

Publication Ethics (/index.php/JUITA/pages/view/publicationethics)

Author Guidelines (/index.php/JUITA/about/submissions#authorGuidelines)

Article Template (https://docs.google.com/document/d/1JoUHP707SsdT_4s6_0tXX6pXiQTmJW-1/edit?usp=sharing&ouid=114837924174631254413&rtpof=true&sd=t

For Readers (http://jurnalnasional.ump.ac.id/index.php/JUITA/information/readers)

JUITA : Jurnal Informatika e-ISSN: 2579-8901 p-ISSN: 2086-9398

JUITA VOL. 8 NOMOR 1, MEI 2020

Cover (http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/9069) PDF (http://jurnalnasional.ump.ac.id/index.php/JUITA/article/view/9

Pengantar Redaksi dan Daftar Isi

Performance of Levenberg-Marquardt Algorithm in Backpropagation Network Based on the Number

Monitoring Status Gizi Balita Secara Online PDF (Bahasa

Development of Android Based Online Nurse Test Preparation

Reference Management Tool (https://www.mendeley.com/download-desktop/)

Publication Ethics (/index.php/JUITA/pages/view/publicationethics)

Author Guidelines (/index.php/JUITA/about/submissions#authorGuidelines)

Article Template (https://docs.google.com/document/d/1JoUHP707SsdT_4s6_0tXX6pXiQTmJW-1/edit?usp=sharing&ouid=114837924174631254413&rtpof=true&sd=t

For Readers (http://jurnalnasional.ump.ac.id/index.php/JUITA/information/readers)

JUITA : Jurnal Informatika e-ISSN: 2579-8901 p-ISSN: 2086-9398

Performance of Levenberg-Marquardt Algorithm

Performance of Levenberg-Marquardt … | Mustafidah, H., Suwarsito, 29 – 35 29

Fig. 1 Design of artificial neural networks with 5 input neurons, n neurons in

30 Performance of Levenberg-Marquardt … | Mustafidah, H., Suwarsito, 29 – 35

B. Research Data 1) Determine the hypothesis

Fig. 2 Flowchart of ANN program development

Fig. 3 Testing statistics design of many neurons n in hidden layers (n = 2, 4,

Performance of Levenberg-Marquardt … | Mustafidah, H., Suwarsito, 29 – 35 31

Fig. 4 Source code of ANN computer program

32 Performance of Levenberg-Marquardt … | Mustafidah, H., Suwarsito, 29 – 35

Performance of Levenberg-Marquardt … | Mustafidah, H., Suwarsito, 29 – 35 33

IV. CONCLUSION REFERENCES

34 Performance of Levenberg-Marquardt … | Mustafidah, H., Suwarsito, 29 – 35

2015. Neural Network,” in Proceeding International

Performance of Levenberg-Marquardt … | Mustafidah, H., Suwarsito, 29 – 35 35

You might also like