Article Per Year (5 Year)
p-Index From 2021 - 2026
13.688
P-Index
This Author published in this journals
All Journal Jurnal Buana Informatika Sinkron : Jurnal dan Penelitian Teknik Informatika U Karst JURNAL MEDIA INFORMATIKA BUDIDARMA Indonesian Journal of Artificial Intelligence and Data Mining Jurnal Informatika Universitas Pamulang Jiko (Jurnal Informatika dan komputer) J-Dinamika: Jurnal Pengabdian Kepada Masyarakat Building of Informatics, Technology and Science Buletin Ilmiah Sarjana Teknik Elektro Jurnal Informatika Ekonomi Bisnis Abdi: Jurnal Pengabdian dan Pemberdayaan Masyarakat Jurnal Teknik Informatika C.I.T. Medicom JOURNAL OF INFORMATION SYSTEM RESEARCH (JOSH) Journal of Computer System and Informatics (JoSYC) Jurnal Sistem Komputer dan Informatika (JSON) Reswara: Jurnal Pengabdian Kepada Masyarakat Jurnal Teknologi Informatika dan Komputer Journal of Computer Networks, Architecture and High Performance Computing Jurasik (Jurnal Riset Sistem Informasi dan Teknik Informatika) Jurnal Teknik Informatika (JUTIF) Yumary: Jurnal Pengabdian kepada Masyarakat International Journal of Robotics and Control Systems KLIK: Kajian Ilmiah Informatika dan Komputer J-SAKTI (Jurnal Sains Komputer dan Informatika) Brilliance: Research of Artificial Intelligence Malcom: Indonesian Journal of Machine Learning and Computer Science Jurnal Informatika Ekonomi Bisnis Jurnal Pengabdian Masyarakat Charitas Kesatria : Jurnal Penerapan Sistem Informasi (Komputer dan Manajemen)
Airlangga, Gregorius
Atma Jaya Catholic University Of Indonesia
Religion Agriculture, Biological Sciences & Forestry Humanities Civil Engineering, Building, Construction & Architecture Computer Science & IT Control & Systems Engineering Decision Sciences, Operations Research & Management Economics, Econometrics & Finance Education Electrical & Electronics Engineering Energy Engineering Environmental Science Health Professions Industrial & Manufacturing Engineering Languange, Linguistic, Communication & Media Law, Crime, Criminology & Criminal Justice Mathematics Medicine & Pharmacology Public Health Social Sciences Other

Published : 116 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 116 Documents
Search

 8   9   10   11   12 
Enhancing Facial Emotion Recognition on FER2013 Using Attention-based CNN and Sparsemax-Driven Class-Balanced Architectures Suwartono, Christiany; Bata, Julius Victor Manuel; Airlangga, Gregorius
Buletin Ilmiah Sarjana Teknik Elektro Vol. 7 No. 4 (2025): December
Publisher : Universitas Ahmad Dahlan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12928/biste.v7i4.14510

Abstract

Facial emotion recognition plays a critical role in various human–computer interaction applications, yet remains challenging due to class imbalance, label noise, and subtle inter-class visual similarities. The FER2013 dataset, containing seven emotion classes, is particularly difficult because of its low resolution and heavily skewed label distribution. This study presents a comparative investigation of advanced deep learning architectures against traditional machine-learning baselines on FER2013 to address these challenges and improve recognition performance. Two novel architectures are proposed. The first is an attention-based convolutional neural network (CNN) that integrates Mish activations and squeeze-and-excitation (SE) channel recalibration to enhance the discriminative capacity of intermediate features. The second, FastCNN-SE, is a refined extension designed for computational efficiency and minority-class robustness, incorporating Sparsemax activation, Poly-Focal loss, class-balanced reweighting, and MixUp augmentation. The research contribution is demonstrating how combining attention, sparse activations, and imbalance-aware learning improves FER performance under challenging real-world conditions. Both models were extensively evaluated: the attention-CNN under 10-fold cross-validation, achieving 0.6170 accuracy and 0.555 macro-F1, and FastCNN-SE on the held-out test set, achieving 0.5960 accuracy and 0.5138 macro-F1. These deep models significantly outperform PCA-based Logistic Regression, Linear SVC, and Random Forest baselines (≤0.37 accuracy and ≤0.29 macro-F1). We additionally justify the differing evaluation protocols by emphasizing cross-validation for architectural stability and held-out testing for generalization and note that FastCNN-SE contains ~3M parameters, enabling efficient inference. These findings demonstrate that architecture-level fusion of SE attention, Sparsemax, and Poly-Focal loss improves balanced emotion recognition, offering a strong foundation for future studies on efficient and robust affective-computing systems.
Improved Text Classification for Indonesian Hate Speech Detection: FastText-LSTM Model with Easy Data Augmentation Wicaksana, Hilman Singgih; Huda, Khairul; Airlangga, Gregorius
Jurnal Sistem Komputer dan Informatika (JSON) Vol. 7 No. 3 (2026): Maret 2026
Publisher : Universitas Budi Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30865/json.v7i3.9637

Abstract

The swift expansion of social media in Indonesia has led to a significant rise in hate speech, highlighting the urgent need for effective automated detection techniques. This research evaluates the performance of the proposed FastText-Long Short-Term Memory with Easy Data Augmentation (FastText-LSTM-WE) compared with the baseline model, FastText-Convolutional Neural Network with Easy Data Augmentation (FastText-CNN-WE). To further investigate the impact of data augmentation, the effectiveness of both FastText-Long Short-Term Memory without Easy Data Augmentation (FastText-LSTM-WO) and FastText-Convolutional Neural Network without Easy Data Augmentation (FastText-CNN-WO) was also assessed. Bayesian Optimization was employed to identify the best hyperparameter configurations for each model. The experiments were carried out on a dataset comprising 14,306 samples while maintaining consistent experimental conditions. Model performance was measured using precision, recall, F1-score, and accuracy derived from the confusion matrix. The results indicate that FastText-LSTM-WE achieved the highest performance, with precision, recall, F1-score, and accuracy of 84.02%, 83.16%, 83.59%, and 81.37%, respectively. These findings demonstrate that the proposed model provides a robust and reliable solution for detecting hate speech within the Indonesian context, thereby improving automated content moderation systems in practical applications.
EVALUATING MACHINE LEARNING MODELS FOR PREDICTING SLEEP DISORDERS IN A LIFESTYLE AND HEALTH DATA CONTEXT Gregorius Airlangga
JIKO (Jurnal Informatika dan Komputer) Vol 7 No 1 (2024)
Publisher : Program Studi Teknik Informatika Universitas Khairun

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33387/jiko.v7i1.7870

Abstract

Sleep disorders significantly impact public health, but their detection is often complicated by the multifaceted nature of causative factors. This study investigates the efficacy of various machine learning (ML) models in identifying sleep disorders based on comprehensive lifestyle and health data. We employed a dataset comprising 400 individual records with features including demographic information, sleep metrics, lifestyle factors, and health parameters. The dataset distinguished between individuals with no sleep disorder, insomnia, and sleep apnea. We evaluated a broad spectrum of ML models including logistic regression, decision trees, ensemble methods like RandomForest and GradientBoosting, support vector machines, and neural networks. The models' performances were assessed using accuracy, precision, recall, and F1 score metrics. Results indicated that ensemble methods, particularly RandomForest and XGBClassifier, outperformed other models in terms of accuracy, precision, and F1 scores, achieving values as high as 0.93. These methods proved effective in managing the complexity and variability of the dataset, thereby suggesting their robustness in clinical predictive analytics. The study's findings advocate for the use of advanced ensemble techniques in developing diagnostic tools for sleep disorders, highlighting their potential to enhance predictive accuracy and reliability in real-world healthcare settings. Further research is recommended to optimize these models and explore their integration into clinical practice.
EVALUATING HYBRID NEURAL NETWORK ARCHITECTURES FOR PREDICTING SLEEP DISORDERS FROM STRUCTURED DATA Gregorius Airlangga
JIKO (Jurnal Informatika dan Komputer) Vol 7 No 1 (2024)
Publisher : Program Studi Teknik Informatika Universitas Khairun

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33387/jiko.v7i1.7873

Abstract

The accurate diagnosis of sleep disorders is crucial for effective treatment and management, yet current methods often rely on subjective assessments and are not always reliable. This research examines the efficacy of various neural network architectures, including dense networks, convolutional neural networks (CNNs), recurrent neural networks (RNNs), and innovative hybrid models, in predicting sleep disorders from structured health data. Our study focuses on comparing the performance of these models using metrics such as accuracy, precision, recall, and F1 score across a dataset comprising 400 individuals with detailed sleep and lifestyle data. Our findings demonstrate that while traditional models like dense networks and CNNs for structured data yield robust results, hybrid models, particularly the CNN-Transformer, significantly outperform others. This model effectively integrates convolutional layers with Transformer’s attention mechanisms, excelling in handling complex data interactions and providing superior predictive accuracy with an F1 score and accuracy reaching as high as 0.91. Conversely, RNN models, designed to capture temporal data dependencies, showed less efficacy, underscoring the importance of model selection aligned with data characteristics. This suggests that for datasets not exhibiting strong temporal features, models leveraging spatial relationships or advanced attention mechanisms are more suitable. This study not only advances our understanding of neural network applications in medical diagnostics but also highlights the potential of hybrid models in enhancing diagnostic accuracy. These insights could lead to significant improvements in the early detection and treatment of sleep disorders, thereby enhancing patient outcomes and contributing to the broader field of medical informatics.
Evaluating Deep Learning Models for HIV/AIDS Classification: A Comparative Study Using Clinical and Laboratory Data Gregorius Airlangga
Journal of Information System Research (JOSH) Vol 6 No 2 (2025): January 2025
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/josh.v6i2.6783

Abstract

The accurate classification of HIV/AIDS status is critical for effective diagnosis, treatment planning, and disease management. This study evaluates the performance of four deep learning models: Multilayer Perceptron (MLP), Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU) on a comprehensive clinical and laboratory dataset derived from the AIDS Clinical Trials Group Study 175. The dataset includes features such as demographic information, treatment history, and immune markers like CD4 and CD8 counts. To address class imbalance, the Synthetic Minority Oversampling Technique (SMOTE) was applied, followed by stratified 10-fold cross-validation to ensure robust evaluation. Each model's performance was assessed using metrics including accuracy, precision, recall, F1-score, and ROC-AUC. GRU emerged as the most effective model, achieving the highest accuracy (71.04%) and ROC-AUC (57.72%), demonstrating its robustness in handling sequential data. CNN and LSTM showed competitive performance, particularly in balancing precision and recall. However, all models faced challenges in recall, highlighting difficulties in identifying minority-class samples. The findings underscore the potential of GRU for HIV/AIDS classification while identifying limitations in current approaches to handling class imbalance. Future work will explore advanced architectures, such as attention mechanisms and hybrid models, to further improve sensitivity and robustness. This study contributes to the growing body of research on applying deep learning to healthcare, with implications for improving diagnostic accuracy and patient outcomes.
Performance Evaluation of Machine Learning Models for HIV/AIDS Classification Gregorius Airlangga
Journal of Information System Research (JOSH) Vol 6 No 2 (2025): January 2025
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/josh.v6i2.6790

Abstract

Accurate and early diagnosis of HIV/AIDS is critical for effective treatment and reducing disease transmission. This study evaluates the performance of several machine learning models, including Logistic Regression, Decision Tree, Random Forest, Gradient Boosting, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Naive Bayes, for classifying HIV/AIDS infection status. A dataset comprising 50,000 samples was used, and models were assessed based on accuracy, precision, recall, and F1 score using stratified ten-fold cross-validation to ensure robust evaluation. The results reveal significant trade-offs between sensitivity and specificity across the models. Gradient Boosting achieved the highest accuracy (70.85%) and precision (57.81%), making it suitable for confirmatory testing where minimizing false positives is critical. Conversely, Naive Bayes demonstrated the highest recall (57.99%) and F1 score (51.04%), emphasizing its effectiveness in early-stage diagnostics where sensitivity is paramount. SVM exhibited the highest precision (59.87%) but struggled with recall (11.28%), reflecting its conservative nature in classifying positive cases. These findings underscore the importance of selecting models tailored to specific diagnostic objectives. While Naive Bayes is ideal for comprehensive screening programs, Gradient Boosting and SVM are better suited for confirmatory testing. This research provides valuable insights into the strengths and limitations of machine learning models for medical diagnostics, paving the way for developing more robust, hybrid approaches to optimize sensitivity and specificity in HIV/AIDS classification.
Co-Authors Afandi Nur Aziz Thohari, Afandi Nur Aziz Alamsyah - Amalia, Andika Elok Basuki, Widodo Widjaja Bata, Julius Victor Manuel Christanto, Henoch Juli Christiany Suwartono Denny Jean Cross Sihombing Denny Jean Cross Sihombing Eugenius Kau Suni Henoch Juli Christanto Julius Bata Julius Victor Manuel Bata Khairul Huda Kristian, Yoel Lenson, Abraham K. S. Lim, Bobi Hartanto Pramudita Makin, See Jong Nugroho, Oskar Ika Adi Pedro Manuel Lamberto Buu Sada Philia, Gregoria Piolo, Samuel Pujo Hari Saputro Rahmananta, Radyan Ronald Sukwadi Setiawan, Yohanes Evan Sidhunata, Billy Macarius Stephen Aprius Sutresno Sugianto, Lai Ferry Suni, Eugenius Kau Wicaksana, Hilman Singgih