cover
Article Per Year (5 Year)
All
Home Page
OAI Link
Editorial Team
Contact
Reviewer
Google Scholar
Contact Name
-
Contact Email
-
Phone
-
Journal Mail Official
jurnal@if.uinsgd.ac.id
Editorial Address
Gedung Fakultas Sains dan Teknologi Lt. 4 Jurusan Teknik Informatika Jl. A.H. Nasution No. 105 Cibiru Bandung 40614 Telp. (022) 7800525 / Fax (022) 7803936 Email : jurnal@if.uinsgd.ac.id
Location
Kota bandung,
Jawa barat
INDONESIA
JOIN (Jurnal Online Informatika)
Published by Universitas Islam Negeri Sunan Gunung Djati Bandung
ISSN : 25281682     EISSN : 25279165     DOI : 10.15575/join
Core Subject : Science,
Computer Science & IT
JOIN (Jurnal Online Informatika) is a scientific journal published by the Department of Informatics UIN Sunan Gunung Djati Bandung. This journal contains scientific papers from Academics, Researchers, and Practitioners about research on informatics. JOIN (Jurnal Online Informatika) is published twice a year in June and December. The paper is an original script and has a research base on Informatics.
Arjuna Subject : -
Articles 508 Documents
 47   48   49   50   51 
CPSO-LSTM: Chaotic Particle Swarm Optimization improved LSTM Hyperparameters for Air Pollution Prediction Andi, Tri; Pranolo, Andri; Ismail, Amelia Ritahani; Kusuma, Candra Juni Cahyo
JOIN (Jurnal Online Informatika) Vol 11 No 1 (2026)
Publisher : Department of Informatics, UIN Sunan Gunung Djati Bandung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15575/join.v11i1.1689

Abstract

Accurate air pollution predictions are crucial for public health and environmental management, but achieving high prediction accuracy remains a challenge due to the complexity of temporal patterns in pollution data. This study aims to improve performance of Long Short-Term Memory (LSTM) by optimizing hyperparameters tuning based Chaotic Particle Swarm Optimization (CPSO) for air pollution predictions. Hyperparameter optimization included the number of hidden layers, neurons, activation functions, loss functions, optimizers, batch sizes, and epochs. The proposed model LSTM-CPSO compared to other models, baseline LSTM and PSO-LSTM, to predict the concentrations of PM2.5, PM10, NO2, SO2, CO, and O3 in Jakarta based on Mean Squared Error (MSE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE) metrics. The experimental results show that CPSO-LSTM achieves superior performance with MSE 0.012105, MAE 0.086356, RMSE 0.110022, and MAPE 32.31%, outperforming the baseline LSTM by 38.1% on the MSE metric and 11.9% on MAPE. Interestingly, LSTM-CPSO produces better architecture with 2 hidden layers and 91 neurons than LSTM-PSO that requires 7 hidden layers with 51 neurons. Similarly, LSTM-CPSO has shortest 5 training epochs better than LSTM-PSO with 16 epochs. This research demonstrates that chaos-based metaheuristic optimization can select the best hyperparameters to improve the performance of LSTM for air quality forecasting.
Comparative Performance of Fine-Tuned IndoBERT BASE and LARGE Variants for Emotion Detection in Indonesian Tweets Winarno, Sri; Novita Dewi, Ika; Nugraha, Adhitya; Firdausillah, Fahri; Fitri, Maulatus Shaffira; Ramadhani, Talitha Olga; Widhiyanti, Erna Amalia; Rizqi, Ainur Rahma Miftakhul
JOIN (Jurnal Online Informatika) Vol 11 No 1 (2026)
Publisher : Department of Informatics, UIN Sunan Gunung Djati Bandung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15575/join.v11i1.1704

Abstract

In the digital era, where emotions play a crucial role in shaping human behavior, communication, and decision-making, their expressions are often conveyed through short and informal texts on platforms such as Twitter. This research aims to improve the accuracy of emotion detection in Indonesian text using the IndoBERT-BASE-P2 and IndoBERT-LARGE-P2 transformer models. The dataset consists of 7,080 tweets annotated with six basic emotion categories (anger, fear, joy, love, neutral, and sad). The research methodology included text preprocessing, class balancing using SMOTE, and fine-tuning with optimized training parameters. Evaluation results show that IndoBERT-BASE-P2 achieved an accuracy of 84.43% and a macro F1-score of 84.33%, surpassing previous studies, while the larger IndoBERT-LARGE-P2 model tended to overfit and offered no meaningful improvement. Error analysis showed the neutral class was the most difficult to classify. These findings demonstrate that with effective preprocessing and parameter optimization, a smaller model can be a highly efficient solution for emotion classification in Indonesian text, especially in resource-constrained conditions.
Optimizing Hypertension Prediction from Electronic Health Records via Hybrid ANN–LSTM with Ant Colony and Bayesian Optimization Junadhi, Junadhi; Zulham, Zulham
JOIN (Jurnal Online Informatika) Vol 11 No 1 (2026)
Publisher : Department of Informatics, UIN Sunan Gunung Djati Bandung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15575/join.v11i1.1723

Abstract

Hypertension is a chronic medical condition that, if undetected early, can result in life-threatening complications such as cardiovascular disease and stroke. Despite numerous studies on predictive modeling for hypertension, existing approaches often suffer from limited accuracy due to suboptimal feature selection, inadequate hyperparameter tuning, and imbalanced datasets. This study aims to address these limitations by proposing a hybrid deep learning framework that integrates Artificial Neural Networks (ANN) and Long Short-Term Memory (LSTM) models with Ant Colony Optimization (ACO) for feature selection and Bayesian Optimization (BO) for hyperparameter tuning. The proposed method is trained on Electronic Health Records (EHR) and employs the Synthetic Minority Over-sampling Technique (SMOTE) to mitigate class imbalance. Experimental results show that the optimized ANN achieves an accuracy of 94.3% and the optimized LSTM reaches 95.1%, outperforming baseline models without optimization. Improvements in precision, recall, and F1-score further demonstrate the model’s robustness in identifying hypertension cases. The main contribution of this research lies in the integration of ACO-based feature optimization and BO-based hyperparameter tuning within a hybrid ANN–LSTM framework, resulting in a clinically applicable model for early hypertension prediction. These findings suggest that the proposed approach has strong potential for deployment in electronic medical record systems to support timely and accurate clinical decision-making.
Early Fusion of Visual and Ingredient Representations for Multimodal Food Classification Rahma Salsabila, Navira; Regita Azzahra, Adela; Utaminingrum, Fitri; Henryranu Prasetio, Barlian
JOIN (Jurnal Online Informatika) Vol 11 No 1 (2026)
Publisher : Department of Informatics, UIN Sunan Gunung Djati Bandung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15575/join.v11i1.1725

Abstract

Identifying the most appropriate food dish based on available kitchen ingredients remains a practical yet challenging task in everyday life. To address this, this study specifically aims to develop an intelligent food classification system using a multimodal approach. We propose a multimodal food classification method that performs early fusion by combining visual and textual features extracted using the Contrastive Language–Image Pretraining (CLIP) model. Features from food images and ingredient lists are fused and classified through a two-layer multilayer perceptron. The model is evaluated on the Recipes5k dataset with 4,826 samples across 101 food categories. Results show that the proposed multimodal model achieves 91.32% accuracy, outperforming text-only (85.65%) and image-only (57.26%) baselines. The main contribution of this work lies in demonstrating the effectiveness of early fusion for combining cross-modal representations in food classification. Unlike prior methods, our model supports flexible inference with either text or image input, enabling practical real-world applications. These findings highlight the potential of multimodal learning for food recommendation systems, offering both accuracy and contextual relevance beyond unimodal approaches.
Comparative Analysis of Pre-Trained Deep Learning Models for Classifying Tropical Fungal Skin Infections Aras, Suhardi; Darwis, Muhammad RIzwan; Arkan, Muhammad Zhaky
JOIN (Jurnal Online Informatika) Vol 11 No 1 (2026)
Publisher : Department of Informatics, UIN Sunan Gunung Djati Bandung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15575/join.v11i1.1758

Abstract

Tropical fungal skin infections, including Tinea corporis, Tinea versicolor, Tinea pedis, and Tinea nigra, are common health problems in tropical countries such as Indonesia. Although not life-threatening, these diseases can cause discomfort, reduce self-confidence, and interfere with daily activities. Conventional diagnostic methods still rely on subjective visual observation, which is often inaccurate—especially in regions with limited infrastructure and scarce access to specialists. Moreover, existing studies rarely provide a comparative evaluation of deep learning architectures for tropical fungal infections using small and diverse datasets. Therefore, this study aims to address these challenges by conducting a systematic comparative evaluation of three pre-trained models—MobileNetV3, EfficientNet-B2, and SE-ResNet101—to determine the most accurate and computationally efficient architecture for multi-class classification of tropical fungal skin diseases. In this study, a dataset of 660 clinical skin images sourced from the Kaggle repository was used, covering four tropical fungal infection classes. The dataset consisted of 165 images per class, which were divided into training, validation, and testing subsets. Experimental results demonstrated that MobileNetV3 achieved the best performance, with a validation accuracy of 95.08%, a test accuracy of 97.34%, and the shortest training time of 15 minutes, compared to EfficientNet-B2 (16 minutes) and SE-ResNet101 (22 minutes). The main contribution of this study is to provide a systematic comparative evaluation of deep learning models for the classification of tropical fungal skin infections, while recommending MobileNetV3 as the most suitable model for practical implementation of automated image-based diagnosis in primary healthcare services with limited resources.
Maleo-Short: An "In-the-Wild" Indonesian Dataset for Speaker Diarization Mardiana, Ardi; Muslimah, Dinda Desmonda; Bastian, Ade; Irawan, Eka Tresna
JOIN (Jurnal Online Informatika) Vol 11 No 1 (2026)
Publisher : Department of Informatics, UIN Sunan Gunung Djati Bandung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15575/join.v11i1.1781

Abstract

Speaker diarization (SD), the task of partitioning an audio stream into speaker-homogenous segments, is fundamental for analyzing multi-speaker recordings. Its application to “in-the-wild” data, such as content from the YouTube platform, poses significant challenges, including overlapped speech, ambient noise, and rapid speaker turns, thereby constituting an active research area. While numerous SD datasets are available, they predominantly focus on English and other high-resource languages. A notable scarcity of publicly accessible datasets exists for the Indonesian language, as extant corpora are primarily engineered for Automatic Speech Recognition (ASR). To address this resource deficit, this research introduces Maleo-Short, a new Indonesian multi-speaker dataset derived from YouTube. The dataset comprises 110 short conversational clips, with a total duration of 1 hours 32 minutes. A reliable ground truth was established through a meticulous manual annotation process using ELAN to generate precise speaker segmentation and transcription files. To validate its utility and assess its complexity, the dataset was evaluated using pre-trained baseline models. The empirical results confirm its status as a challenging benchmark, with the most effective models achieving a Diarization Error Rate (DER) of 32.64% and a Word Error Rate (WER) of 33.78%. Maleo-Short is presented as a valuable, publicly accessible resource intended to catalyze advancements in Indonesian speaker diarization research by facilitating the development and rigorous evaluation of SD systems on acoustically complex and realistic conversational data. Maleo-Short is available at https://doi.org/10.57967/hf/7944.  
Optimizing Stacking Ensemble Models for Customer Churn Prediction in the Telecommunications Industry Rofik, Rofik; Unjung, Jumanto; Pertiwi, Dwika Ananda Agustina; Muslim, Much Aziz
JOIN (Jurnal Online Informatika) Vol 11 No 1 (2026)
Publisher : Department of Informatics, UIN Sunan Gunung Djati Bandung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15575/join.v11i1.1783

Abstract

One of the biggest challenges in the telecommunications industry is predicting churn, which is the condition when a customer unsubscribes and switches to another service provider. In an era of competitive market conditions, retaining customers is much more efficient than acquiring new customers. Conventional prediction models are often unable to capture the complexity of customer behavior patterns, resulting in a lower accuracy than optimal. This study aims to optimize customer churn prediction performance by developing a stacking ensemble model that combines several classification algorithms to improve model performance. Fourteen algorithms were tested, and the six algorithms with the best accuracy were selected as base learners, while Logistic Regression was selected as the meta-learner. The stacking model testing was carried out sequentially through a combination of 6 algorithms with the same meta-learner algorithm. Testing was also carried out with and without using the SMOTE data balancing method to evaluate the effect of data balancing on the prediction results. The results of this study show that the combination of the Adaboost, Ridge Classifier, and Logistic Regression algorithms can produce the highest accuracy of 82.97%, which exceeds the prediction performance of a single algorithm. This research contributes to demonstrating an effective stacking ensemble configuration for predicting customer churn in the telecommunications industry and emphasizes that the selection of the right algorithm combination has a greater impact on model performance than the number of algorithms used.
Denial of Service (DOS) Attack Detection on MQTT Protocol Using the Random Forest Method Monika Dian Pertiwi, Kharisma; Azizi Hasibuan, Nurul; Putri Rahmawati, Dyah
JOIN (Jurnal Online Informatika) Vol 11 No 1 (2026)
Publisher : Department of Informatics, UIN Sunan Gunung Djati Bandung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15575/join.v11i1.1784

Abstract

The Message Queuing Telemetry Transport (MQTT) protocol serves as a critical lightweight communication infrastructure for Internet of Things (IoT) systems. Still, it remains highly vulnerable to Denial of Service (DoS) attacks that compromise network availability and security. Despite extensive IoT security research, existing MQTT-based intrusion detection systems predominantly employ binary classification approaches and lack comprehensive multi-class attack differentiation capabilities, limiting their practical deployment in real-world scenarios. This study addresses this critical gap by developing a multi-class DoS attack detection system utilizing the Random Forest algorithm to simultaneously classify normal traffic, MQTT flooding attacks, and SYN flood attacks. The methodology encompasses four systematic stages: collecting an MQTT network traffic dataset containing 1,634,286 records across three attack categories through controlled simulations; performing rigorous data preprocessing for cleaning and normalization; strategically extracting 60 MQTT-specific attributes to identify attack signatures; and implementing Random Forest with optimized hyperparameters for multi-class classification. Experimental results demonstrate optimal performance using an 80:20 train-test split with 5-fold cross-validation, achieving 95.27% precision, 95.09% recall, 95.08% F1-score, and 95.09% accuracy. A comprehensive evaluation using macro and micro-averaged metrics confirms the model's ability to autonomously classify MQTT network traffic types with high accuracy and balanced performance across all attack categories, offering a practical security solution for MQTT-enabled IoT infrastructure.

Page 51 of 51 | Total Record : 508

 47   48   49   50   51 

Filter by Year

2016 2026


Reset
Filter By Issues
All Issue Vol 11 No 1 (2026) Vol 10 No 2 (2025) Vol 10 No 1 (2025) Vol 9 No 2 (2024) Vol 9 No 1 (2024) Vol 8 No 2 (2023) Vol 8 No 1 (2023) Vol 7 No 2 (2022) Vol 7 No 1 (2022) Vol 6 No 2 (2021) Vol 6 No 1 (2021) Vol. 5 No 2 (2020) Vol 5, No 1 (2020) Vol 5 No 1 (2020) Vol 4 No 2 (2019) Vol 4 No 1 (2019) Vol 3 No 2 (2018) Vol 3 No 1 (2018) Vol 2 No 2 (2017) Vol 2 No 1 (2017) Vol 1 No 2 (2016) Vol 1 No 1 (2016) More Issue