Article Per Year (5 Year)
p-Index From 2021 - 2026
0.61
P-Index
This Author published in this journals
All Journal Knowledge Engineering and Data Science ILKOM Jurnal Ilmiah
Manga', Abdul Rachman
Unknown Affiliation
Computer Science & IT Engineering

Published : 3 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 3 Documents
Search

 1 
Utilization of Deep Learning YOLO V9 for Identification and Classification of Toraja Buffalo Breeds Manga', Abdul Rachman; Herawati, Herawati; Purnawansyah, Purnawansyah
ILKOM Jurnal Ilmiah Vol 17, No 1 (2025)
Publisher : Prodi Teknik Informatika FIK Universitas Muslim Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33096/ilkom.v17i1.2349.12-19

Abstract

This study aims to develop and evaluate a buffalo breed detection system that supports the cultural practices of the Toraja community, particularly in the context of the Rambu Solo’ ceremony. The ceremony places significant importance on the types of buffaloes used, as each breed symbolizes different social statuses and cultural meanings. In response to the need for an accurate and efficient identification method, this research utilizes the YOLOv9 (You Only Look Once version 9) deep learning model to detect and classify Toraja buffalo breeds. A dataset comprising 2,656 annotated images was used, representing five distinct buffalo categories: bongga sori, bonga ulu, moon, saleko, and todi. The images were collected from both field documentation and online sources. The YOLOv9 model was trained across 90 epochs, aiming to achieve high accuracy in breed detection and classification. The evaluation results demonstrate the model's strong performance, achieving a precision of approximately 0.9 and a recall of 0.8. These metrics indicate the model's ability to correctly identify the buffalo breeds with a high degree of reliability. However, during the training process, certain patterns of overfitting and underfitting were observed, suggesting that the model's performance could still be improved. These issues can potentially be addressed by increasing the volume and diversity of training data, applying data augmentation techniques, and fine-tuning hyperparameters to achieve a more balanced generalization. Overall, the findings show that YOLOv9 is a promising tool for supporting cultural preservation through technology by automating the identification of buffalo types used in traditional ceremonies. This system can assist in maintaining the accuracy and consistency of buffalo classification according to local customs. Future research is recommended to explore broader datasets, compare alternative object detection algorithms, and develop an integrated application for practical field use.
Analysis of the Ensemble Method Classifier's Performance on Handwritten Arabic Characters Dataset Manga', Abdul Rachman; Handayani, Anik Nur; Herwanto, Heru Wahyu; Asmara, Rosa Andrie; Sulistya, Yudha Islami; Kasmira, Kasmira
ILKOM Jurnal Ilmiah Vol 15, No 1 (2023)
Publisher : Prodi Teknik Informatika FIK Universitas Muslim Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33096/ilkom.v15i1.1357.186-192

Abstract

Arabic character handwriting is one of the patterns and characteristics of each person's writing. This characteristic makes Arabic writing more challenging if the letter recognition process is based on a dataset of Arabic scripts. This Arabic script has been presented in a dataset totaling 16800, each representing a class of hijaiyah letters starting from alif to yes, consisting of 600 data for each class. The accuracy of the data used can be increased using the ensemble method. By using multiple algorithms at simultaneously, the ensemble technique can raise the level or result of a score in machine learning. This study's primary goal is to evaluate the ensemble method classifier's performance on datasets of handwritten Arabic characters. The classifier uses the ensemble method by applying the proposed soft voting to provide a multiclass classification of three machine learning algorithms, namely, SVM, Random Forest, and Decision Tree for classification. This research process produces an accuracy value for the voting classifier of 0.988 and several other SVM algorithms with an accuracy of 0.103, a random forest with an accuracy of 1.0, and a decision tree with an accuracy of 0.134. The test results used the confusion matrix evaluation model, including accuracy, precision, recall, and f1-score of 0.99.
Comparative Performance of VGG16 and EfficientNetB0-Based Transfer Learning for Brain Tumor Classification Azis, Huzain; Jalil, Rizqi Ananda; Manga', Abdul Rachman
Knowledge Engineering and Data Science
Publisher : citeus

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

The classification of brain tumors using Magnetic Resonance Imaging (MRI) images is essential for early diagnosis but remains challenging due to tumor diversity. This study evaluates the effectiveness of two distinct architectural approaches for feature extraction: VGG16, representing a classic sequential design, and EfficientNetB0, a modern architecture optimized for parameter efficiency through compound scaling. Using a dataset of 2,870 MRI images categorized into four classes, we implemented a static transfer learning strategy by freezing all pre-trained ImageNet weights to act as fixed feature extractors. Features were extracted from specific layers, the final pooling layer for VGG16 and the Global Average Pooling (GAP) layer for EfficientNetB0. To optimize the discriminative power of these static features, this study implements an ensemble-fusion framework, specifically Voting and Stacking classifiers, to integrate the strengths of diverse base learners (SVM, Random Forest, and XGBoost) in identifying complex brain tumor patterns. Results demonstrate that VGG16-based models consistently and significantly outperform EfficientNetB0-based models across all evaluation metrics. The VGG16 + Random Forest combination achieved a peak accuracy of 73.9%, whereas EfficientNetB0-based models struggled, with peak accuracies reaching only 36.1%. Statistical analysis confirms that VGG16’s uniform  convolutional layers provide more stable and discriminative feature representations for medical textures compared to the mobile-optimized blocks of EfficientNetB0 when weights are not fine-tuned. This research highlights that, for static feature extraction in resource-constrained environments, classic architectural stability offers superior reliability compared to modern parameter efficiency.
Co-Authors Anik Nur Handayani Herawati Herawati Heru Wahyu Herwanto Jalil, Rizqi Ananda Kasmira, Kasmira Purnawansyah Purnawansyah Rosa Andrie Asmara Yudha Islami Sulistya