Article Per Year (5 Year)
p-Index From 2021 - 2026
8.08
P-Index
This Author published in this journals
All Journal Indonesian Journal of Electronics and Instrumentation Systems ELKHA : Jurnal Teknik Elektro JUITA : Jurnal Informatika Annual Research Seminar Sinkron : Jurnal dan Penelitian Teknik Informatika Jusikom : Jurnal Sistem Komputer Musirawas SELAPARANG: Jurnal Pengabdian Masyarakat Berkemajuan JURNAL TEKNIK INFORMATIKA DAN SISTEM INFORMASI Petir KOMPUTIKA - Jurnal Sistem Komputer Jurnal Teknologi Terpadu EDUMATIC: Jurnal Pendidikan Informatika Jutisi: Jurnal Ilmiah Teknik Informatika dan Sistem Informasi Progresif: Jurnal Ilmiah Komputer Jurnal Ilmiah Betrik : Besemah Teknologi Informasi dan Komputer JATI (Jurnal Mahasiswa Teknik Informatika) Syntax: Journal of Software Engineering, Computer Science and Information Technology Algoritme Jurnal Mahasiswa Teknik Informatika Informatics and Enginering Dedication DEVICE : JOURNAL OF INFORMATION SYSTEM, COMPUTER SCIENCE AND INFORMATION TECHNOLOGY Jurnal Informatika dan Multimedia MDP Student Conference JRIIN :Jurnal Riset Informatika dan Inovasi Jurnal Ilmiah Betrik : Besemah Teknologi Informasi dan Komputer Applied Information Technology and Computer Science (AICOMS) INOVTEK Polbeng - Seri Informatika JuTISI (Jurnal Teknik Informatika dan Sistem Informasi)
Yohannes, Yohannes
Universitas Multi Data Palembang
Agriculture, Biological Sciences & Forestry Chemical Engineering, Chemistry & Bioengineering 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 Medicine & Pharmacology Nursing Public Health Social Sciences Other

Published : 73 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 73 Documents
Search

 4   5   6   7   8 
Feature-Level Fusion of DenseNet121 and EfficientNetV2 with XGBoost for Multi-Class Retinal Classification Laksana, Jovansa Putra; Yohannes
Sinkron : jurnal dan penelitian teknik informatika Vol. 10 No. 1 (2026): Article Research January 2026
Publisher : Politeknik Ganesha Medan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33395/sinkron.v10i1.15670

Abstract

Accurate and efficient classification of retinal fundus images plays a critical role in supporting the early diagnosis of ocular diseases. However, models relying on a single deep learning backbone often struggle to capture the multi-scale and heterogeneous characteristics of retinal lesions, leading to unstable performance across visually similar disease classes. To address this limitation, this study proposes a novelty feature-level fusion framework that integrates complementary representations from DenseNet121 and EfficientNetV2-s, followed by classification using XGBoost. The fusion pipeline extracts 1024-dimensional features from DenseNet121 and 1280-dimensional features from EfficientNetV2-s, which are concatenated into a unified 2304-dimensional feature vector. Experiments were conducted on a dataset of 10,247 retinal fundus images spanning six categories: Central Serous Chorioretinopathy, Diabetic Retinopathy, Macular Scar, Retinitis Pigmentosa, Retinal Detachment, and Healthy. The proposed fusion model achieved an accuracy of 91.60%, outperforming DenseNet121 XGBoost (91.31%) and EfficientNetV2-s XGBoost (89.70%). Moreover, the fusion strategy demonstrated improved class-level stability, particularly for visually similar retinal disorders where single-backbone models exhibited higher misclassification rates. This study contributes a lightweight yet effective multi-backbone feature-level fusion approach that enhances discriminative representation and classification stability without increasing model complexity. In addition, the use of XGBoost introduces a tree-based decision mechanism that is inherently more interpretable than conventional fully connected layers, offering potential advantages for clinical analysis. Overall, the results highlight the effectiveness of multi-backbone feature fusion as a reliable strategy for automated retinal disease classification.
Performance Analysis of YOLOv11 Integrated with Lightweight Backbones (MobileNetV2, GhostNet, ShuffleNet V2) for Cigarette Detection Andreas, Kevin; Yohannes; Meiriyama
INOVTEK Polbeng - Seri Informatika Vol. 11 No. 1 (2026): February
Publisher : P3M Politeknik Negeri Bengkalis

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35314/0gjq1j10

Abstract

Cigarette object detection in indoor environments plays a vital role for enforcing smoke-free zone regulations and protecting public health from secondhand smoke exposure. This study investigates the performance of YOLOv11n architecture integrated with three lightweight backbone modifications (MobileNetV2, GhostNet, and ShuffleNet V2) for real-time cigarette detection with the aim of achieving efficiency suitable for potential deployment on resource-constrained edge devices. Comprehensive experiments were conducted using the Cigar Detection Dataset comprising 5,333 images, augmented to 8,890 samples through horizontal flipping and brightness adjustment techniques. All models were trained for 100 epochs using the SGD optimizer on an NVIDIA Tesla T4 GPU. The evaluation metrics included detection accuracy (mAP@0.5, mAP@0.5:0.95, precision, recall, and F1-score) and computational efficiency (parameters, model size, GFLOPs, and FPS). Experimental results demonstrate that the pretrained YOLOv11n baseline achieves the highest detection accuracy with mAP@0.5 of 0.8072 and precision of 0.8688. Among lightweight backbone variants, ShuffleNet V2 (0.5x) provides the most compact solution with only 2.28M parameters and a 4.73 MB model size, while ShuffleNet V2 (0.75x) offers an optimal balance between accuracy (mAP@0.5: 0.7430) and efficiency with only 0.95% accuracy degradation compared to the 1.0x variant. These findings provide practical guidance for selecting appropriate model configurations based on deployment constraints in smoke-free area monitoring systems.
Klasifikasi Motif Kain Batik Nitik Menggunakan Support Vector Machine dengan Ekstraksi Fitur EfficientNet-B0 Saputra, Dika; Yohannes, Yohannes
Progresif: Jurnal Ilmiah Komputer Vol 22, No 1 (2026): Januari
Publisher : STMIK Banjarbaru

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35889/progresif.v22i1.3507

Abstract

Nitik Batik is an Indonesian cultural heritage with complex geometric dot patterns, yet its digitalization and preservation efforts remain limited. This study aims to develop an automatic classification system for 60 Nitik Batik motifs using a combination of EfficientNet-B0 as a feature extractor and Support Vector Machine (SVM) as a classifier. The Batik Nitik 960 Dataset was expanded from 960 to 1,920 images with rotation augmentation. Experiments were conducted with 10-fold cross-validation and evaluation on separate test data. Results show that the model without augmentation achieved 55.71% accuracy, 90.06% macro precision, 55.71% recall, and 65.29% F1-score. Blur augmentation with 30% probability reduced accuracy to 49.29% although it decreased overfitting by 6.63%. SVM parameters were set to C=0.3 and gamma=0.01 to improve regularization. This study concludes that the combination of EfficientNet-B0 and SVM is effective for multi-class batik classification, but blur augmentation is unsuitable for detail-rich textile data. Future research recommendations include exploring geometric augmentation and more advanced feature extractor architectures.Kata kunci: Nitik Batik; Image classification; EfficientNet-B0; Support Vector Machine; augmentation. AbstrakBatik Nitik merupakan warisan budaya Indonesia dengan motif geometris berbentuk titik yang kompleks, namun upaya digitalisasi dan pelestariannya masih terbatas. Penelitian ini bertujuan mengembangkan sistem klasifikasi otomatis untuk 60 motif Batik Nitik menggunakan kombinasi EfficientNet-B0 sebagai ekstraktor fitur dan Support Vector Machine (SVM) sebagai klasifikator. Dataset Batik Nitik 960 Dataset diperluas dari 960 menjadi 1.920 citra dengan augmentasi rotasi. Eksperimen dilakukan dengan skema 10-fold cross-validation dan evaluasi pada data uji terpisah. Hasil menunjukkan bahwa model tanpa augmentasi mencapai akurasi 55,71%, precision macro 90,06%, recall 55,71%, dan F1-score 65,29%. Augmentasi blur 30% justru menurunkan akurasi menjadi 49,29% meskipun mengurangi overfitting sebesar 6,63%. Parameter SVM diatur C=0,3 dan gamma=0,01 untuk meningkatkan regularisasi. Penelitian ini menyimpulkan bahwa kombinasi EfficientNet-B0 dan SVM efektif untuk klasifikasi batik multikelas, namun augmentasi blur tidak sesuai untuk data tekstur kaya detail. Rekomendasi penelitian selanjutnya adalah eksplorasi augmentasi geometris dan arsitektur feature extractor yang lebih advance.Kata kunci: Batik Nitik; Klasifikasi citra; EfficientNet-B0; Support Vector Machine; Augmentasi.
KLASIFIKASI MAMALIA MENGGUNAKAN EXTREME GRADIENT BOOSTING BERDASARKAN FITUR HISTOGRAM OF ORIENTED GRADIENT Yohannes; Johannes Petrus
BETRIK Vol. 13 No. 03 (2022): Jurnal Ilmiah BETRIK : Besemah Teknologi Informasi dan Komputer
Publisher : PPPM Institut Teknologi Pagar Alam

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.36050/e2t7t733

Abstract

Mammals are one type of animal that has many characteristics and characteristics.The shape of the face in each type of mammal has a similar shape. The faces of mammals in theform of frontal images are a challenge in image classification. In this study, the Histogram ofOriented Gradient (HOG) is used as a feature of the facial shape of mammals. HOG is used as astrengthening feature in the classification process using the eXtreme Gradient Boosting(XGBoost) method. The test was carried out using a dataset of frontal facial imagery ofmammals consisting of 15 species. The results of the tests show that the XGBoost method with theHOG feature is able to provide better classification results for mammals than without the HOGfeature. This is indicated by an increase in the precision value of 0.61; recall of 0.62; and an f1-score of 0.60 on XGBoost with HOG feature which is almost double that of XGBoost withoutHOG feature.
Penerapan Metode Branch and Bound untuk Optimalisasi Rute Wisata Terdekat di Kota Palembang Jaysen Stephanus; Felix Gunawan; Yohannes Yohannes
Applied Information Technology and Computer Science (AICOMS) Vol 5 No 1 (2026): AICOMS
Publisher : Pengelola Jurnal Politeknik Negeri Ketapang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.58466/eqadem96

Abstract

This study discusses the application of the Branch and Bound method to optimize the nearest tourist route in Palembang City using the Traveling Salesman Problem (TSP) approach. The problem raised is how to determine the most efficient tourist route from several tourist destinations with minimum travel distance. The study utilizes geographic coordinate data of tourist destinations obtained through OpenStreetMap, then the distance between locations is calculated using the Haversine Formula to obtain an accurate distance estimate based on latitude and longitude. Furthermore, the Branch and Bound Algorithm is used to find the optimal route solution through the process of branching, bounding, and pruning so that the solution search becomes more efficient than the brute force method. The results show that the system successfully produces an optimal circular tourist route with a total minimum distance of 40.47 km and an execution time of 12.84 seconds. The integration of the Haversine Formula and Branch and Bound is proven to be able to provide efficient, accurate, and adaptive tourist route recommendations to help tourists save travel time and transportation costs in Palembang City.
Perbandingan Algoritma Greedy dan Dynamic Programming Pada Optimasi Playlist Spotify Untuk Jogging Fadhel Muhammad; Muhammad Radja Juang Jamemiko; Yohannes Yohannes
Applied Information Technology and Computer Science (AICOMS) Vol 5 No 1 (2026): AICOMS
Publisher : Pengelola Jurnal Politeknik Negeri Ketapang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.58466/1htfcz49

Abstract

Spotify provides audio metadata that can be utilized to support physical activities such as jogging. This study compares the performance of Greedy and Dynamic Programming algorithms for Spotify playlist optimization modeled as a 0/1 Knapsack Problem. Song duration is treated as weight, while a score derived from popularity and energy is used as value. The dataset was obtained from Spotify Wrapped 2025 Top 50 Songs and Spotify All-Time Top 100 Songs, resulting in 31 candidate songs after preprocessing and filtering. Experiments were conducted on playlist durations of 30, 45, 60, 75, and 90 minutes. The results show that Dynamic Programming consistently achieved higher total scores than Greedy across all scenarios. For the 60-minute playlist, Dynamic Programming obtained a total score of 1897 compared to 1894 achieved by Greedy. However, Greedy required a lower execution time (4.244 ms) than Dynamic Programming (16.196 ms). The average optimality gap between the two methods was 1.89%, indicating that Greedy produced solutions that were close to the optimal solutions generated by Dynamic Programming while requiring less computation time.
Perbandingan Kinerja Algoritma Greedy dan Dynamic Programming dalam Optimasi Diskon Keranjang Belanja E-Commerce Menggunakan Dataset Online Retail UCI Jonathan Tanujaya; Daffa Yudha Musyaffa; Yohannes Yohannes
Applied Information Technology and Computer Science (AICOMS) Vol 5 No 1 (2026): AICOMS
Publisher : Pengelola Jurnal Politeknik Negeri Ketapang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.58466/f0fdve41

Abstract

E-commerce platforms heavily rely on automated promotional strategies, such as tiered discounts, to enhance customer loyalty. Therefore, this study aims to analyze the performance of computational algorithms in determining item priorities within a shopping cart under promotional budget constraints. The 0/1 Knapsack Problem was addressed by comparing two computational approaches: Dynamic Programming (DP) and the Greedy Algorithm. Transaction data from the UCI Online Retail dataset were cleaned and aggregated into 3,746 unique product catalogs, then simulated using a promotional budget limit of £499.40 with a 10% discount policy. Computational experiments revealed contrasting trade-off characteristics between the two approaches. The DP algorithm guaranteed an absolute optimal solution with a total profit of £2,725,575.77 but required 28.10 seconds of computation time. In contrast, the Greedy algorithm completed the selection process in a fraction of a second (0.17 seconds) while incurring only a marginal profit deficit of 0.01%. The Greedy heuristic approach proved to be highly practical and efficient for integration into real-time user interface systems, whereas the superior accuracy of DP makes it more suitable for offline database processing and inventory analytics research.
Optimasi Strategi Repeat Buyer pada E-commerce Indonesia Melalui Pendekatan Dynamic Programming untuk Bundling Product Multi-Kategori Siti Fatimah Az Zahrah; Yeremia Agung Chandra; Yohannes Yohannes
DEVICE : JOURNAL OF INFORMATION SYSTEM, COMPUTER SCIENCE AND INFORMATION TECHNOLOGY Vol 7, No 1: JUNI 2026
Publisher : Universitas Dharmawangsa

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.46576/device.v7i1.8956

Abstract

Penelitian ini bertujuan untuk mengoptimalkan strategi peningkatan repeat buyer pada e-commerce di Indonesia melalui penyusunan rekomendasi bundling product multi-kategori berbasis pendekatan komputasional. Pendekatan yang digunakan adalah Dynamic Programming melalui model optimasi Knapsack yang dikombinasikan dengan analisis Threshold Standard Deviation untuk menyaring kategori produk berdasarkan kedekatan demografis pelanggan. Proses penelitian meliputi tahap preprocessing data, pemodelan parameter bobot dan profit, optimasi kombinatorial, serta penentuan prioritas rekomendasi berbasis customer profiling. Hasil penelitian menunjukkan bahwa sistem mampu menghasilkan rekomendasi bundling yang relevan dan terpersonalisasi berdasarkan usia dan riwayat transaksi pelanggan. Dynamic Programming menunjukkan performa yang lebih stabil dan efisien pada kompleksitas data yang lebih tinggi, meskipun pada dataset kecil Brute Force memiliki waktu eksekusi lebih cepat. Secara keseluruhan, pendekatan yang diusulkan dinilai mampu meningkatkan akurasi rekomendasi serta mendukung strategi pemasaran untuk mendorong loyalitas pelanggan.
Penyelesaian Capacitated Vehicle Routing Problem with Time Windows Menggunakan Algoritma Greedy dan Tabu Search pada Distribusi Pengiriman Farmasi Migel Orvin Febryan; Siska Amelia; Yohannes Yohannes
DEVICE : JOURNAL OF INFORMATION SYSTEM, COMPUTER SCIENCE AND INFORMATION TECHNOLOGY Vol 7, No 1: JUNI 2026
Publisher : Universitas Dharmawangsa

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.46576/device.v7i1.8804

Abstract

Distribusi farmasi merupakan salah satu rantai pasok kritis yang menuntut ketepatan waktu dan efisiensi operasional tinggi. Kompleksitas distribusi farmasi muncul dari ketatnya batasan waktu pengiriman, beragamnya jenis produk dengan karakteristik berat dan volume berbeda, serta keterbatasan kapasitas armada kendaraan yang harus melayani puluhan hingga ratusan titik pengiriman dalam satu hari operasional. Kegagalan memenuhi jendela waktu pelayanan dapat berdampak langsung pada ketersediaan obat di fasilitas kesehatan dan berpotensi membahayakan keselamatan pasien. Penelitian ini bertujuan menyelesaikan permasalahan Capacitated Vehicle Routing Problem with Time Windows (CVRPTW) pada sistem distribusi farmasi menggunakan kombinasi algoritma Greedy dan Tabu Search. Algoritma Greedy dengan strategi nearest neighbor digunakan untuk membentuk solusi awal, sedangkan Tabu Search digunakan untuk mengoptimasi solusi tersebut melalui mekanisme relocate inter-route dan local search 2-opt. Dataset yang digunakan memuat 78 titik pengiriman dengan batasan kapasitas kendaraan sebesar 400 kg berat dan 3 m³ volume. Fungsi objektif yang diminimalkan mencakup total jarak tempuh, total waktu, pelanggaran time window, serta kelebihan kapasitas. Hasil eksperimen menunjukkan bahwa Tabu Search mampu menghasilkan perbaikan nilai fungsi objektif dibandingkan solusi awal Greedy, dengan tetap mempertahankan feasibilitas seluruh rute. Penelitian ini membuktikan efektivitas kombinasi metaheuristik berbasis memori dengan konstruksi heuristik sederhana untuk permasalahan optimasi rute kendaraan berskala nyata.
Convolutional Block Attention Module Integration into YOLO11 Architecture for MRI Image-based Brain Tumor Detection Jendraja Husin Kotan; Yohannes; Hafiz Irsyad
Journal of Innovation and Technology Polbeng Series on Informatics (INOVTEK Polbeng - Seri Informatika) Vol. 11 No. 2 (2026): May
Publisher : P3M Politeknik Negeri Bengkalis

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35314/n4nrvj87

Abstract

Brain tumor is one of the deadly diseases in the world that can affect anyone, this disease is characterized by the growth of abnormal cells or tissues in the brain, medically it can be life-threatening if not treated properly. Most tumor detection tasks are done by manual assessment by radiologists or pathologists where this work is time-consuming, so accurate and reliable detection is needed in the medical field in diagnosing brain tumors. The purpose of this study is to integrate CBAM on the YOLO11 architecture in detecting brain tumors and determine the performance of the brain tumor detection model using the YOLO11 architecture with CBAM integration. The method used to detect brain tumors is the YOLO11 architecture with CBAM integration. The dataset used is an image in the form of brain MRI. The results of this study indicate that the precision is 86.9%, recall is 86.2%, mAP50 is 91%, and mAP50-95 is 64% in the validation data and precision is 89.1%, recall is 92%, mAP50 is 79%, mAP50-95 is 51.6%, and F1 score is 90.5% in the test data which can be used to help medical personnel in detecting and treating brain tumors considering that this model has outstanding results, especially in the recall metric section which reaches 92% in the test data.
Co-Authors Ade Hendri Pandrean Adhytio Mahendra Adrian Chandra Albert Cahayadi Andreas, Kevin Ariel Sudarsono Azarya, Philips Denny Beni Anthony Bobby Jaya Saputra Cahyati, Imelia Dwinora Calvin Oliver Saputra Candra Candra Celvine Adi Putra Cendy Prakarsah Daffa Yudha Musyaffa Dafid Dafid Dandy, Dandy Daniel Udjulawa Daniel Udjulawa Devella, Siska Dody, Muhammad Fadhel Muhammad Famerdi, Farhan Agung Farhan Agung Famerdi Farisi, Ahmad Febbiola Febbiola Felix Gunawan Femmy Johan Feristyani, Indah Firda Novia Rahmawati Gerry Jeven Timoti Glen, Billy Hafiz Irsyad Hafiz Irsyad Hartati, Ery Inayatullah Inayatullah Indah Feristyani Jaysen Stephanus Jendraja Husin Kotan Jennifer Verty Jericho Jericho Jerry Setiawan Jimmy Aprilyanto Johannes Petrus Jonathan Tanujaya Joseph Eduard Uly Loni Julian Rusli Tee Baldi Juliana Nasution Kelvin Arianto Kevin Andreas Klaudius Audie Irsansaputra Laksana, Jovansa Putra Leo Chandra Leonardo Leonardo M Dhafa Adjie Saputra M. Zaky Naufal Farisky Marcella, Dewi Meiriyama Meiriyama, Meiriyama Migel Orvin Febryan Molavi Arman Muhammad Ezar Al Rivan Muhammad Farid Athar Muhammad Radja Juang Jamemiko Muhammad Rizky Pribadi Muhammad Yudha Setiawan Muhdhor, Umar Novan Wijaya Nur Rachmat Pandi Pandi Pandi Pandi, Pandi Pandrean, Ade Hendri Philips Denny Azarya Prabowo, Adrianus Prasthio, Rial Putra, Lipi Amanda Raphael Lee Ricky Wijaya RR. Ella Evrita Hestiandari Sahpira, Mulia Saputra, Dika Sari, Yulya Puspita Selvie Selvie Serenity Devina Suryanto Setiawan, Jerry Siska Amelia Siska Devella Siti Fatimah Az Zahrah Sonia Sonia, Sonia Tanuwijaya, William Timoteus Ivan Sariyo Veraldo Verrino Adityya Wijang Widhiarso William Hadisaputra William Tanuwijaya Yeremia Agung Chandra Yoannita Yoannita Yoannita Yoannita, Yoannita Yulya Puspita Sari