Thursday, May 26, 2022

Journal of Information and Communication Technology (JICT) Vol.21, No.2, April 2022

Noor Ain Syazwani Mohd Ghani & Abdul Kadir Jumaat
Abdurrakhman Prasetyadi, Budi Nugroho & Adrin Tohari
Nazmus Sakib Akash, Sigma Jahan, Shakir Rouf, Amlan Chowdhury, Amitabha Chakrabarty & Jia Uddin
Rifo Ahmad Genadi, Masayu Leylia Khodra

1Noor Ain Syazwani Mohd Ghani & 2Abdul Kadir Jumaat
1&2Faculty of Computer and Mathematical Sciences, Universiti Teknologi MARA, Malaysia;
One of the most important steps in image processing and computer vision for image analysis is segmentation, which can be classified into global and selective segmentation. Global segmentation models can segment whole objects in an image. Unfortunately, these models are unable to segment a specific object that is required for extraction. To overcome this limitation, the selective segmentation model, which is capable of extracting a particular object or region in an image, must be prioritised. Recent selective segmentation models have shown to be effective in segmenting greyscale images. However, if the input is vector-valued or identified as a colour image, the models simply ignore the colour information by converting that image into a greyscale format. Colour plays an important role in the interpretation of object boundaries within an image as it helps to provide a more detailed explanation of the scene's objects. Therefore, in this research, a model for selective segmentation of vector-valued images is proposed by combining concepts from existing models. The finite differences method was used to solve the resulting Euler Lagrange (EL) partial differential equation of the proposed model. The accuracy of the proposed model's segmentation output was then assessed using visual observation as well as by using two similarity indices, namely the Jaccard (JSC) and Dice (DSC) similarity coefficient. Experimental results demonstrated that the proposed model is capable of successfully segmenting a specific object in vector-valued images. In the future, this research can be extended further in 3-dimensional modelling.
Keywords: Active contour, colour image segmentation, selective segmentation, variational model, vector-valued image.
btn fulltext

1*Abdurrakhman Prasetyadi, 1Budi Nugroho & 2Adrin Tohari
1Research Center for Informatics, National Research and Innovation Agency, Indonesia
2Research Center for Geotechnology, National Research and Innovation Agency, Indonesia
Cluster methods such as k-means have been widely used to group areas with a relatively equal number of disasters to determine areas prone to natural disasters. However, it is difficult to obtain a homogeneous clustering result of the k-means method because this method is sensitive to a random selection of the centers of the cluster. This paper presents the result of a study that aimed to apply a proposed hybrid approach of the combined k-means algorithm and hierarchy to the clustering process of anticipation level datasets of natural disaster mitigation in Indonesia.  This study also added keyword and disaster-type fields to provide additional information for a better clustering process. The clustering process produced three clusters for the anticipation level of natural disaster mitigation. Based on the validation from the expert, 67 districts/cities (82.7%) fall into cluster 1 (low anticipation), 9 districts/cities (11.1%) is classified into cluster 2 (medium), and the remaining 5 districts/cities (6.2%) is categorized in cluster 3 (high anticipation). From the analysis of the calculation of the silhouette coefficient, the hybrid algorithm provides relatively homogeneous clustering results. Furthermore, applying the hybrid algorithm to the keyword segment and the type of disaster also produces a homogeneous clustering as indicated by the calculated purity coefficient and the total purity values. Thus, the proposed hybrid algorithm can provide relatively homogeneous clustering results in natural disaster mitigation.
Keywords: Clustering, Hybrid, K-means, Mitigation, Natural disaster.
btn fulltext

1Nazmus Sakib Akash, 2Sigma Jahan, 3Shakir Rouf, 3Amlan Chowdhury,3Amitabha Chakrabarty & 4Jia Uddin
1Department of Computing & Information System, Daffodil International University, Bangladesh
2Faculty of Computer Science, Dalhousie University, Canada
3Department of Computer Science & Engineering, BRAC University, Bangladesh
4AI and Big Data Department, Endicott College, Woosong University, South Korea;;;;;
With rapid technological progress in the Internet of Things (IoT), it has become imperative to concentrate on its security aspect. This paper represents a model that accounts for the detection of botnets through the use of machine learning algorithms. The model examines anomalies, commonly referred to as botnets, in a cluster of IoT devices attempting to connect to a network. Essentially, this paper exhibits the use of transport layer data (User Datagram Protocol - UDP) generated through IoT devices.  We propose an intelligent novel model comprising Random Forest Classifier with Independent Component Analysis (ICA) for botnet detection in IoT devices. Various machine learning algorithms are also implemented upon the processed data for comparative analysis. The experimental results of the proposed model generated state-of-the-art results for three different datasets, achieving up to 99.99% accuracy effectively with the lowest prediction time of 0.12 seconds without overfitting. The significance of our study lies in detecting botnets in IoT devices effectively and efficiently under all circumstances by utilizing ICA with Random Forest Classifier, which is a simple machine learning algorithm.
Keywords: Botnets, Distributed Denial of Service, Independent Component Analysis, Internet of Things, Random Forest Classifier.
btn fulltext 

1Ahmed Elaraby & 2Zohair Al-Ameen
1Department of Computer Science, Faculty of Computers and Information, South Valley University, Egypt
2Department of Computer Science, College of Computer Science and Mathematics, University of Mosul, Iraq;
Edge detection is the diverse way used to detect boundaries in digital images. Many methods exist to achieve this purpose, yet not all of them can produce results with high detection ratios, some may have high complexity, and others may require numerous inputs. Hence, a new multi-phase algorithm that depends on information theory is introduced in this article to detect the edges of aerial images adequately in a fully automatic manner. The proposed algorithm works by utilizing Shannon and Hill entropies with specific rules along with a non-complex edge detector to record the vital edge information. The proposed algorithm is examined with different aerial images, its performances appraised against six existing approaches and the outcomes are appraised using three image evaluation methods. From the results, promising performances were recorded as the proposed algorithm performed the best in many aspects and provided satisfactory results. The results of the proposed algorithm have high edge detection ratios as it was able to capture most of the significant edges of the given images. Such findings make the proposed algorithm desirable to be used as a key image detection method with other image-related applications. 
Keywords: aerial images, edge detection, image processing, information theory.
btn fulltext

Rifo Ahmad Genadi & Masayu Leylia Khodra
School of Electrical Engineering and Informatics, Institut Teknologi Bandung, Indonesia;
In aspect-based sentiment analysis (ABSA), tasks are diverse and consist of aspect term extraction, aspect categorization, opinion term extraction, sentiment polarity classification, and relation extractions of aspect and opinion terms. These tasks are generally carried out sequentially using more than one model. However, this approach is inefficient and likely to reduce the model’s performance due to cumulative errors in previous processes. The co-extraction approach with Dual crOss-sharEd RNN (DOER) and span-based multitask (SpanMLT) acquired better performance than the pipelined approaches in English review data. Therefore, this research focuses on adapting the co-extraction approach where the extraction of aspect terms, opinion terms, and sentiment polarity are conducted simultaneously from review texts. The co-extraction approach was adapted by modifying the original frameworks to perform unhandled subtask to get the opinion triplet. Furthermore, the output layer on these frameworks was modified and trained using a collection of Indonesian-language hotel reviews. The adaptation was conducted by testing the output layer topology for aspect and opinion term extraction as well as variations in the type of recurrent neural network cells and model hyperparameters used, and then analysing the results to obtain a conclusion. The two proposed frameworks were able to carry out opinion triplet extraction and achieve decent performance. The DOER framework achieves better performance than the baselines on aspect and opinion term extraction tasks.
Keywords: aspect-based sentiment analysis, opinion triplet, co-extraction, Dual crOss-sharEd RNN, span-based multitask
btn fulltext

Universiti Utara Malaysia Press
 Universiti Utara Malaysia, 06010 UUM Sintok
Kedah Darul Aman, MALAYSIA
Phone: +604-928 4816, Fax : +604-928 4792

Creative Commons License

All articles published in Journal of Information and Communication Technology (JICT) are licensed under a Creative Commons Attribution 4.0 International License.