cover
Contact Name
-
Contact Email
-
Phone
-
Journal Mail Official
-
Editorial Address
-
Location
Kota bandung,
Jawa barat
INDONESIA
Indonesian Journal on Computing (Indo-JC)
Published by Universitas Telkom
ISSN : 24609056     EISSN : -     DOI : -
Core Subject : Science,
Indonesian Journal on Computing (Indo-JC) is an open access scientific journal intended to bring together researchers and practitioners dealing with the general field of computing. Indo-JC is published by School of Computing, Telkom University (Indonesia).
Arjuna Subject : -
Articles 251 Documents
Pneumonia Classification from X-ray Images Using Residual Neural Network Abdan Hafidh Ahnafi; Anditya Arifianto; Kurniawan Nur Ramadhani
Indonesia Journal on Computing (Indo-JC) Vol. 5 No. 2 (2020): September, 2020
Publisher : School of Computing, Telkom University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.34818/INDOJC.2020.5.2.454

Abstract

Pneumonia is a virus, bacterium, and fungi infection disease which causes alveoli swelling and gets worse easily if it is not taken care of immediately. There are symptoms that can be recognized through x-ray images, for example the appearance of white mist in the lungs. A pneumonia classification system has already developed, but it still produced low accuracy. In this research we develop classification system by increasing the depth of CNN architecture using Residual Neural Network to improve accuracy from previous research. The dataset contains 2 classes which are pneumonia and normal, and trained to produce the best learning strategy with various scenarios. The model trained using data train that has been oversampling. The best scenario is achieved by ResNet152 architecture using dropout 0.5. This scenario achieved a result of 0.88 precision, 0.95 recall, 0.92 f1-score, and 0.89 of accuracy. The classification model on this research produces higher accuracy compared to the research of Enes Ayan, et.al. in 2019 which produced 0.87.
Classifying Skin Cancer in Digital Images Using Convolutional Neural Network with Augmentation Zeyhan Aliyah; Anditya Arifianto; Febryanti Sthevanie
Indonesia Journal on Computing (Indo-JC) Vol. 5 No. 2 (2020): September, 2020
Publisher : School of Computing, Telkom University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.34818/INDOJC.2020.5.2.455

Abstract

Skin cancer is a hazardous disease that can induces death if it is not taken care of immediately. The disease is hard to identified since the symptoms have similarities with other disease. An automatically classification system of skin cancer has been developed, but it still produced low accuracy. We use Convolutional Neural Network  to enhance the accuracy of the classification. There are 2 main scenarios conducted in this research using HAM10000 dataset which has 7 classes. We compared ResNet and VGGNet architectures and obtained ResNet50 with augmentation as the best model with the accuracy of 99% and 99% macro avg.
LBP Advantages over CNN Face Detection Method on Facial Recognition System in NOVA Robot Luqman Bramantyo Rahmadi; Kemas Muslim Lhaksmana; Donny Rhomanzah
Indonesia Journal on Computing (Indo-JC) Vol. 5 No. 2 (2020): September, 2020
Publisher : School of Computing, Telkom University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.34818/INDOJC.2020.5.2.456

Abstract

Network-optimized virtual assistant (NOVA) is a robot developed by Bandung Techno Park (BTP) that can interact with humans for various purposes, such as a receptionist robot. NOVA robot is still in development and one of the main focuses is adding face recognition features so that the robot can actively greet and interact with humans. Therefore, we propose a face recognition and tracking system based on neural networks. This system is developed using the Google FaceNet feature extraction method. Previously, face detection in NOVA robot was implemented by employing the multi-task cascaded convolutional networks (MTCNN) method, whereas face tracking on the system was realized by using the modification of the MOSSE object tracking method. However, we found that the implementation of MTCNN in NOVA robot cannot run better than 30 fps. Therefore, this paper aims to solve this issue by investigating conventional face detection methods that could outperform MTCNN in this regard. Tests conducted on the ChokePoint dataset demonstrates that the system with LBP can achieve 30.44 fps framerate with a precision of 95% and recall of 83%. The test results show that LBP is not only better than MTCNN in identifying faces but also more efficient to compute.
Analysis and Implementation of Signature Based Method and Structure File Based Method for File Carving Anjar Afrizal; Niken Dwi Wahyu Cahyani; Erwid Musthofa Jadied
Indonesia Journal on Computing (Indo-JC) Vol. 6 No. 1 (2021): April, 2021
Publisher : School of Computing, Telkom University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.34818/INDOJC.2021.6.1.457

Abstract

File Carving is a data recovery technique based on file structure and content without relying on filesystem information or metadata. The problem in carving files is its high false positive value especially when the file is fragmented (either linear fragmented or non-linear fragmented). The aim of this study is to implement and analyze the performance of two file carving method (Signature Based and File Structure Based) as a solution to the problem of the carving process. By focusing on JPEG, GIF and PNG files, two datasets are used, namely: CFReDS Project (NIST Project) and Basic Data Carving Test (Nick Mikus Project). The analysis is based on the recovery performance (carving recall, supported recall, carving precision), execution time, and memory usage. From the recovery performance parameter, the File Structure Based method gets a higher overall value than the Signature Based method. However, based on the execution time performance parameter, the Signature Based method has better execution time and use fewer resources compared to the File Structure Based method.
A Parallel Implementation of Dual-Pivot Quick Sort for Computers with Small Number of Processors Mohammad F. J. Klaib; Mutaz Rasmi Abu Sara; Masud Hasan
Indonesia Journal on Computing (Indo-JC) Vol. 5 No. 2 (2020): September, 2020
Publisher : School of Computing, Telkom University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.34818/INDOJC.2020.5.2.487

Abstract

Sorting algorithms are heavily used. Quicksort is one of the fastest comparison-based sorting algorithms. These days almost all computing devices have multiple processors. There is a strong need of finding efficient parallel versions of the most common algorithms that are widely used. The basic version of quick sort is sequential and uses only one pivot.  Recently, Yaroslavsky has proposed a modified version of the quick sort that uses two pivots and runs much faster than the single-pivot quick sort. Since then Java has incorporated this dual-pivot quick sort into its standard library for sorting. Although there are many parallel versions of the original single-pivot quick sort, there is a very few for the dual-pivot. Those few parallel versions of the dual-pivot quick sorts are compared with standard sort functions, rather than the dual-pivot quick sort itself. In this paper, we provide a parallel version of the dual-pivot quick sort algorithm of Yaroslavsky and implement it in Java. For comparison, we run both in small number of parallel processors. The experimental results show that our algorithm runs significantly faster than the Yaroslavsky’s algorithm. Moreover, our algorithm performs gradually better as the number of processors and the input size increase.
Sequence Chunking on Quran in English Translation using Bidirectional Long Short-Term Memory Try Arie; Muhammad Arif Bijaksana
Indonesia Journal on Computing (Indo-JC) Vol. 5 No. 2 (2020): September, 2020
Publisher : School of Computing, Telkom University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.34818/INDOJC.2020.5.2.492

Abstract

Every Moslem is obliged to read and understand the meanings of the Quran. The problem is the amount of information contained in the Quran so that ordinary people have difficulty understanding the Quran as a whole. Neural networks can be used to extract important information in the Quran to solve this problem. Therefore, the author proposes a model to identify and classify tags using sequence chunking. The system will use the Bi-LSTM model where the system will be given various token from the Quran as the inputs to be identified as the correct tags. The author is using the dataset obtained from website quran.com. The evaluation of the proposed model produces an f-measure value of 0.903.
Grid-based Image Encryption using Code-based Cryptography Dian Anggoro Putro Bhagaskoro; Ari Moesriami Barmawi
Indonesia Journal on Computing (Indo-JC) Vol. 5 No. 3 (2020): December, 2020
Publisher : School of Computing, Telkom University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.34818/INDOJC.2020.5.3.493

Abstract

Recently, image is frequently sent or exchanged electronically, such that image security is important. A method for securing images is using RSA. However, RSA needs more time for securing an image. For overcoming the RSA problem, McEliece Cryptosystem is introduced to grid-based image encryption. The image is divided into blocks and each block is divided into grids, then finally McEliece Cryptosystem is applied to the pixels in the grids. Based on the experiment’s result, it was proven that the execution time of the proposed method is less than the previous one, while maintaining the security. Keywords: McEliece Cryptosystem, RSA, Image Encryption, Image Decryption, Grid 
Anaphora Resolution on Al-Quran with Indonesian Translation Arlinda Dwi Ardiyani; Moch Arif Bijaksana; Arief Fatchul Huda
Indonesia Journal on Computing (Indo-JC) Vol. 5 No. 2 (2020): September, 2020
Publisher : School of Computing, Telkom University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.34818/INDOJC.2020.5.2.496

Abstract

Al-Quran is the holy book of Islam, in Al-Quran we often find many cases of anaphora. Anaphora is a pronoun, for example “it” which refers to an object (antecedent) in the previous sentence. Antecedent of a pronoun is very important to understand the Al-Quran. Coreference Resolution with the classification model using the support vector machine method are needed to find out the antecedent. In this research, we use i feature and j feature for the extraction process. Based on the evaluation results, the system is able to find the antecedent of an anaphor with the best accuracy value of 86.36%.
The Effect of Information Gain Feature Selection for Hoax Identification in Twitter Using Classification Method Support Vector Machine Isep Mumu Mubaroq; Erwin Budi Setiawan
Indonesia Journal on Computing (Indo-JC) Vol. 5 No. 2 (2020): September, 2020
Publisher : School of Computing, Telkom University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.34818/INDOJC.2020.5.2.499

Abstract

Nowadays social media twitter is popular media for news dissemination. News has elements that can be distinguished types of news, such as hoax that has elements of panic, worry, and anxiety that can have a significant impact in various fields of social, economic, educational, and political. Hoax prevention efforts need as possible before news viral, by to be developed method with functions to identify and hoax analyze. in this research we have proposed an approach Machine Learning with method Support Vector Machine (SVM) supported by feature selection Information Gain (IG) added Term Frequency–Inverse Document Frequency (TF-IDF) for word weighting system performance is very optimal in increasing accuracy by 37,51%, with accuracy reaching 96.55%.
Implementation of Dependency Parser Using Artificial Neural Network Methods Nurul Izzah; Moch Arif Bijaksana; Arief Fatchul Huda
Indonesia Journal on Computing (Indo-JC) Vol. 5 No. 3 (2020): December, 2020
Publisher : School of Computing, Telkom University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.34818/INDOJC.2020.5.3.504

Abstract

In recent years, parsing has become very popular within the scope of NLP (Natural Language Processing) with the presence of Dependency Parser. However, almost all existing Dependency Parser do classifications based on millions of sparse indicator features. This feature is not only bad in drawing conclusions, but also significantly limits the speed of parsing so that the resulting parsing is not optimal. To overcome these problems, changing the use of sparse features becomes dense features to reduce sparsity between words. The Artificial Neural Network classification method is used to produce fast and concise parsing in the Transition-Based Dependency Parser by using 2 hyperparameters. The dataset used in this study is Arabic, Chinese, English, and Indonesian. Based on the evaluation that has been done, it shows a higher result using the second hyperparameter. In testing with English test data, the accuracy value of LAS (Labeled Attachment Score) is 80.4% and UAS (Unlabelled Attachment Score) is 83%, Then with dev data obtained an accuracy value of LAS 81.1% and UAS 83.7%, and parsing speed of 98 sentences per second (sent/s).Keywords: Parsing, dependency parser, transition-based dependency parsing.