Claim Missing Document
Check
Articles

Found 37 Documents
Search

Optimalisasi Keamanan IoT dan Edge Computing Menggunakan Model Machine Learning Leny Margaretha Huizen; Roy Rudolf Huizen
Jurnal Sistem dan Informatika (JSI) Vol 17 No 2 (2023): Jurnal Sistem dan Informatika (JSI)
Publisher : Direktorat Penelitian,Pengabdian Masyarakat dan HKI - Institut Teknologi dan Bisnis (ITB) STIKOM Bali

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30864/jsi.v17i2.543

Abstract

Penggunaan teknologi berbasis Internet of Things (IoT) telah meningkat pesat berkat revolusi digital dan membawa tantangan keamanan yang signifikan. Pengoptimalan keamanan IoT pada edge computing dengan menerapkan model berbasis machine learning, untuk deteksi dan identifikasi. Metodologi yang digunakan meliputi pengumpulan data dari sensor IoT dan log aktifitas sebagai data, pra-pemrosesan data, serta pelatihan dan validasi model machine learning. Pada penelitian ini, deteksi dan identifikasi serangan menggunakan empat algoritma, yaitu K-Nearest Neighbors (KNN), Support Vector Machines (SVM), Random Forest (RF), dan Decision Trees (DT). Hasil penelitian menunjukkan bahwa model Random Forest (RF) dan Decision Tree (DT) memiliki kinerja terbaik dalam mendeteksi serangan siber, dengan nilai True Positive (TP) yang tinggi dan tingkat kesalahan yang rendah. Evaluasi kinerja berdasarkan metrik Akurasi, Presisi, Recall, dan F1-Score mengonfirmasi bahwa RF dan DT mampu memberikan hasil yang akurat dan andal dalam mendeteksi ancaman. Model Random Forest menunjukkan Akurasi 98,4%, Presisi 98,4%, Recall 83,9%, dan F1-Score 90,5%, sedangkan Decision Tree menunjukkan Akurasi 98,1%, Presisi 90,5%, Recall 83,9%, dan F1-Score 87,1%. Implementasi model machine learning dalam sistem keamanan IoT dan edge computing terbukti tidak hanya meningkatkan keamanan data dan perangkat, tetapi juga memaksimalkan efisiensi operasional dengan kemampuan untuk mempelajari dan beradaptasi dengan pola serangan baru.
Optimalisasi Rekayasa Lalu Lintas Melalui Teknologi Deteksi Objek Huizen, Roy Rudolf
Jurnal Sistem dan Informatika (JSI) Vol 18 No 2 (2024): Jurnal Sistem dan Informatika (JSI)
Publisher : Direktorat Penelitian,Pengabdian Masyarakat dan HKI - Institut Teknologi dan Bisnis (ITB) STIKOM Bali

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30864/jsi.v18i2.605

Abstract

Pertumbuhan populasi dan urbanisasi yang pesat telah menyebabkan peningkatan signifikan dalam jumlah kendaraan, terutama di daerah perkotaan. Kondisi ini memicu berbagai masalah lalu lintas, seperti kemacetan dan polusi udara, yang berdampak negatif pada kehidupan sosial dan ekonomi masyarakat. Untuk mengatasi tantangan ini, diperlukan pendekatan rekayasa lalu lintas berbasis teknologi yang cerdas dan efisien. Penelitian ini membandingkan tiga metode utama—Support Vector Machine (SVM) dengan kernel Radial Basis Function (RBF), Decision Tree, dan Random Forest—dalam memprediksi kemungkinan terjadinya kemacetan pada suatu jalan. Menggunakan dataset lalu lintas yang mencakup faktor-faktor seperti volume kendaraan, kecepatan rata-rata, dan kondisi cuaca, setiap metode dilatih dan diuji untuk mengklasifikasikan data lalu lintas menjadi kategori kemacetan atau tidak. Hasil penelitian menunjukkan bahwa Random Forest memiliki performa terbaik, dengan akurasi mencapai 91,06%, precision hingga 83,04%, recall sebesar 91,06%, dan F1-score tertinggi di antara metode yang diuji. Untuk SVM menunjukkan akurasi antara 89,52% hingga 90,04%, dan Decision Tree menunjukkan akurasi antara 87,03% hingga 87,39%. Random Forest menunjukkan keunggulan dalam memprediksi kemacetan lalu lintas dan dapat menjadi solusi andal untuk diterapkan dalam sistem rekayasa lalu lintas berbasis teknologi.
Classification of Lung Diseases in X-Ray Images Using Transformer-Based Deep Learning Models Mahajaya, Nyoman Sarasuartha; Putu Desiana Wulaning Ayu; Roy Rudolf Huizen
Jurnal Nasional Pendidikan Teknik Informatika : JANAPATI Vol. 13 No. 3 (2024)
Publisher : Prodi Pendidikan Teknik Informatika Universitas Pendidikan Ganesha

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23887/janapati.v13i3.81425

Abstract

This research evaluates the performance of two Transformer models, the Vision Transformer (ViT) and Swin Transformer, in the analysis of thoracic X-ray images. The study's objective is to determine whether Transformer models can enhance diagnostic accuracy for lung diseases, considering challenges such as early symptom variability and similar radiological signs. The dataset includes 21,165 X-ray images, featuring 3,616 COVID-19 cases, 10,192 normal images, 6,012 images of Lung Opacity, and 1,345 pneumonia images. Model development involved tuning hyperparameters such as epoch numbers and optimizer choice. The results indicate that using the AdamW and Adamax optimizers achieves an optimal balance between computational efficiency and accuracy. The Swin Transformer model, using the Adamax optimizer, reached the highest testing accuracy of 96.10% in 33,802.70 seconds, while the Vision Transformer achieved a testing accuracy of 95.10% in 33,503.10 seconds.
The object detection model uses combined extraction with KNN and RF classification Kurniati, Florentina Tatrin; Manongga, Daniel HF; Sembiring, Irwan; Wijono, Sutarto; Huizen, Roy Rudolf
Indonesian Journal of Electrical Engineering and Computer Science Vol 35, No 1: July 2024
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/ijeecs.v35.i1.pp436-445

Abstract

Object detection plays an important role in various fields. Developing detection models for 2D objects that experience rotation and texture variations is a challenge. In this research, the initial stage of the proposed model integrates the gray-level co-occurrence matrix (GLCM) and local binary patterns (LBP) texture feature extraction to obtain feature vectors. The next stage is classifying features using k-nearest neighbors (KNN) and random forest (RF), as well as voting ensemble (VE). System testing used a dataset of 4,437 2D images, the results for KNN accuracy were 92.7% and F1-score 92.5%, while RF performance was lower. Although GLCM features improve performance on both algorithms, KNN is more consistent. The VE approach provides the best performance with an accuracy of 93.9% and an F1-score of 93.8%, this shows the effectiveness of the ensemble technique in increasing object detection accuracy. This study contributes to the field of object detection with a new approach combining GLCM and LBP as feature vectors as well as VE for classification.
Analysis of the Impact of Data Oversampling on the Support Vector Machine Method for Stroke Disease Classification Luh Ayu Martini; Pradipta, Gede Angga; Huizen, Roy Rudolf
Journal of Electronics, Electromedical Engineering, and Medical Informatics Vol 7 No 2 (2025): April
Publisher : Department of Electromedical Engineering, POLTEKKES KEMENKES SURABAYA

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35882/jeeemi.v7i2.698

Abstract

Data imbalance is a critical challenge in the classification of medical data, particularly in stroke disease prediction, a life-threatening condition requiring immediate intervention. This imbalance arises due to the disproportionate number of non-stroke cases compared to stroke cases, which can lead to biased models favoring the majority class. Consequently, the model may struggle to correctly identify stroke cases, resulting in lower recall and an increased risk of misdiagnosis. This study evaluates the impact of various oversampling techniques, including Synthetic Minority Over-sampling Technique (SMOTE), Borderline-SMOTE, SMOTE-Edited Nearest Neighbor (SMOTE-ENN), and SMOTE-Instance Prototypes Filtering (SMOTE-IPF), along with feature selection using Information Gain and Chi-Square, to assess their influence on model performance. Oversampling is utilized to address class imbalance by generating synthetic samples, thereby improving the representation of the minority class. Feature selection is employed to eliminate irrelevant or redundant features, enhancing both interpretability and computational efficiency. The dataset obtained from Kaggle, consists of 5,110 records and 12 features. Support Vector Machine (SVM) is used as the classification algorithm, with evaluations conducted on Linear, Radial Basis Function (RBF), and Polynomial kernels. Experimental results indicate that the highest performance is achieved by the combination of Borderline-SMOTE and the RBF kernel, yielding an accuracy of 96.86%, precision of 98.65%, recall of 94.99%, and an F1-score of 96.79%. This model outperforms others in stroke disease classification, demonstrating that the integration of oversampling techniques can effectively enhance prediction accuracy. Future research could focus on implementing deep learning-based models to further optimize stroke classification in the case of imbalanced data. These advancements are expected to enhance model performance, leading to a more effective and efficient approach for medical datasets.
Classification Model for Bot-IoT Attack Detection Using Correlation and Analysis of Variance Firgiawan Faira; Dandy Pramana Hostiadi; Roy Rudolf Huizen
Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi) Vol 9 No 2 (2025): April 2025
Publisher : Ikatan Ahli Informatika Indonesia (IAII)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29207/resti.v9i2.6332

Abstract

Industry 4.0 requires secure networks as the advancements in IoT and AI exacerbate the challenges and vulnerabilities in data security. This research focuses on detecting Bot-IoT activity using the Bot-IoT UNSW Canberra 2018 dataset. The dataset initially showed a significant imbalance, with 2,934,447 entries of attack activity and only 370 entries of normal activity. To address this imbalance, an innovative data aggregation technique was applied, effectively reducing similar patterns and trends. This approach resulted in a balanced dataset consisting of 8 attack activity points and 80 normal activity points. Feature selection using the ANOVA method identified 10 key features from a total of 17: seq, stddev, N_IN_Conn_P_SrcIP, min, state_number, mean, N_IN_Conn_P_DstIP, drate, srate, and max. The classification process utilized Random Forest, k-NN, Naïve Bayes, and Decision Tree algorithms, with 100 iterations and an 80:20 training-testing split. Random Forest showed superior performance, achieving 97.5% accuracy, 97.4% precision, and 97.4% recall, with a total computation time of 11.54 seconds. Pearson correlation analysis revealed a strong positive correlation (+0.937) between N_IN_Conn_P_DstIP and seq, as well as a weak negative correlation (-0.224) between N_IN_Conn_P_SrcIP and state_number. The novelty of this research lies in the application of a data aggregation technique to address class imbalance, significantly improving machine learning model performance and optimizing training time. These findings contribute to the development of robust cybersecurity systems to effectively detect IoT-related threats.
Sentiment Analysis for Hotel Reviews Using Snowball and VADER Rustamaji, Abdullah; Huizen, Roy Rudolf; Hostiadi, Dandy Pramana
Jurnal Nasional Pendidikan Teknik Informatika: JANAPATI Vol. 14 No. 2 (2025)
Publisher : Prodi Pendidikan Teknik Informatika Universitas Pendidikan Ganesha

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23887/janapati.v14i2.82556

Abstract

In the hotel industry, the role of hotel guests is very influential in the development and sustainability of business. Therefore, hotels need to provide services that can satisfy guests. However, many hotels still do not have an analysis system for guest comments. Hotels still manually conduct analysis by discussing with operational leaders to determine whether incoming guest comments contain positive, negative, or neutral sentiments. Previous research introduced guest sentiment analysis but has yet to have optimal accuracy. This paper proposes sentiment analysis using a combination of VADER and Snowball stemmer algorithms, which are tested using real datasets. The goal is to get accurate sentiment analysis results. The experimental results show that the VADER method combined with SnowBall Stemmer has better accuracy than other sentiment analysis methods, with an accuracy of 96.21%. The sentiment analysis model can be used as a basis for decision-making for hotel business owners.
Combination of CNN and SMOTE-IPF for Early Detection of Diabetes Patients in Thermogram Images W Mega Adhi Agam Pradhana; Gede Angga Pradipta; Roy Rudolf Huizen
Jurnal Nasional Pendidikan Teknik Informatika: JANAPATI Vol. 14 No. 2 (2025)
Publisher : Prodi Pendidikan Teknik Informatika Universitas Pendidikan Ganesha

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23887/janapati.v14i2.83145

Abstract

This study aims to enhance the early detection of diabetic complications through the analysis of plantar foot thermogram images using deep learning techniques. A total of 334 thermographic images were utilized, comprising 244 images from 122 diabetic patients (DM class) and 90 images from 45 non-diabetic individuals (control group, CG class). To address the dataset’s imbalance (ratio of 2.64), the Synthetic Minority Over-sampling Technique with Iterative-Partitioning Filter (SMOTE-IPF) was applied both before and after feature extraction. Image quality was further enhanced using Adaptive Histogram Equalization (AHE) and Gamma Correction preprocessing techniques. A Convolutional Neural Network (CNN) model was trained and evaluated on an independent test set of 54 images. The model achieved outstanding results: 99.37% accuracy, 99.37% precision, 100% recall, and a 99.68% F1-score for AHE-processed images. Gamma-corrected images achieved 98.50% accuracy, while original images reached 97.20%. These findings demonstrate the combined value of data balancing and preprocessing in improving non-invasive diabetic foot ulcer detection, offering a promising diagnostic aid for clinical settings.
A Machine Learning-Based Approach for Retail Demand Forecasting: The Impact of Spending Score and Algorithm Optimization Adriani, Ni Putu Erica Puspita; Huizen, Roy Rudolf; Hermawan, Dadang
Jurnal Ilmiah Teknik Elektro Komputer dan Informatika Vol. 11 No. 2 (2025): June
Publisher : Universitas Ahmad Dahlan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26555/jiteki.v11i2.30630

Abstract

Demand forecasting in the retail industry remains a critical challenge, with inaccurate predictions leading to substantial inventory inefficiencies, financial losses, and reduced customer satisfaction. Traditional forecasting methods, primarily reliant on historical sales data, often lack the capacity to effectively model the complexities of dynamic consumer behavior and rapid market fluctuations. To address this, this study proposes a refined demand forecasting approach through the introduction of the Spending Score, a novel synthetic feature that synthesizes customer purchase frequency and total spending to augment predictive accuracy. We implement and optimize machine learning algorithms, specifically Random Forest, Decision Tree, and Support Vector Machine (SVM), using rigorous hyperparameter tuning techniques to determine the most effective model for retail demand prediction. Utilizing detailed customer transaction data, this research aims to identify key purchasing patterns that significantly influence demand variability. By integrating the Spending Score into our predictive models, we provide a data-driven framework enabling retailers to optimize inventory management, enhance targeted marketing strategies, and minimize operational inefficiencies. Empirical results demonstrate that the inclusion of the Spending Score leads to more stable and accurate demand forecasts, facilitating improved alignment between supply and market demand. While acknowledging potential limitations, including data scalability issues and the risk of feature-induced bias, future research will explore the integration of real-time data streams, advanced deep learning methodologies, and expanded datasets to further improve predictive capabilities and model adaptability in the continuously evolving retail landscape.
Comparative Analysis of Augmentation and Filtering Methods in VGG19 and DenseNet121 for Breast Cancer Classification Seneng, I Kadek; Ayu, Putu Desiana Wulaning; Huizen, Roy Rudolf
Jurnal Teknik Informatika (Jutif) Vol. 6 No. 3 (2025): JUTIF Volume 6, Number 3, Juni 2025
Publisher : Informatika, Universitas Jenderal Soedirman

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52436/1.jutif.2025.6.3.4397

Abstract

Breast cancer is one of the most prevalent malignancies and a leading cause of mortality among women worldwide. Mammography plays a crucial role in early detection, yet challenges in manual interpretation have led to the adoption of Convolutional Neural Networks (CNNs) to improve classification accuracy. This study evaluates the performance of Visual Geometry Group (VGG19) and Densely Connected Convolutional Networks (DenseNet121) in mammogram classification. It examines the impact of data augmentation and image enhancement techniques, including Contrast-Limited Adaptive Histogram Equalization (CLAHE), Median Filtering, and Discrete Wavelet Transform (DWT), as well as the influence of varying epochs and learning rates. A novel approach is introduced by assessing data augmentation effectiveness and exploring model adaptations, such as layer incorporation and freezing during training. Classification performance is enhanced through fine-tuning strategies combined with image enhancement techniques, reducing reliance on data augmentation. These findings contribute to medical imaging and computer science by demonstrating how CNN modifications and enhancement methods improve mammogram classification, providing insights for developing robust deep learning-based diagnostic models. The highest performance was achieved using VGG19 with DWT, a learning rate of 0.0001, and 20 epochs, yielding 98.04% accuracy, 98.11% precision, 98% recall, and a 97.99% F1-score. Data augmentation did not consistently enhance results, particularly in clean datasets. Increasing epochs from 10 to 20 improved accuracy, but performance declined at 30 epochs. The confusion matrix showed high accuracy for Benign (100%) and Cancer (99.5%), with more misclassifications in the Normal class (94.5%).