Article Per Year (5 Year)
p-Index From 2021 - 2026
0.408
P-Index
This Author published in this journals
All Journal Jurnal Penelitian Pendidikan IPA (JPPIPA) International Journal Software Engineering and Computer Science (IJSECS)
Khusnul Fajri Rhomadon
Universitas Esa Unggul
Agriculture, Biological Sciences & Forestry Biochemistry, Genetics & Molecular Biology Chemical Engineering, Chemistry & Bioengineering Chemistry Computer Science & IT Education Materials Science & Nanotechnology Physics

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

 1 
Comparative Analysis of Convolutional Neural Network (ResNet-50) and Vision Transformer (ViT-B/16) for Histopathological Image Classification of Colorectal Cancer Muhammad Fazly Qusyairy; Habibullah Akbar; Wahyu Purnama Magribi; Khusnul Fajri Rhomadon
Jurnal Penelitian Pendidikan IPA Vol 12 No 7 (2026)
Publisher : Postgraduate, University of Mataram

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29303/jppipa.v12i7.14880

Abstract

The diagnosis of colorectal cancer (CRC) through histopathological images requires high accuracy to support appropriate clinical decisions. Although Convolutional Neural Networks (CNN) have become the gold standard in medical image analysis, the emergence of Vision Transformer (ViT) architecture offers a new paradigm based on global attention mechanisms (self-attention) that is claimed to be superior on large-scale datasets. However, the effectiveness of ViT-B/16 on medical datasets with limited sample sizes and high texture variation remains debatable. This study aims to comprehensively evaluate the performance of the ViT-B/16 architecture compared to ResNet-50 on the NCT-CRC-HE-100K histopathology dataset, which consists of 9 network classes. The performance of both models was tested using equivalent training scenarios. The evaluation was conducted multidimensionally, covering classification metrics (Accuracy, F1-Score), training stability, feature space separability (t-SNE), visual interpretability (Grad-CAM), and computational efficiency. The experimental results show that ResNet-50 significantly outperforms ViT-B/16 with a test accuracy of 93.24%, compared to ViT-B/16 which only achieves 57.11%. The t-SNE analysis revealed that ViT-B/16 failed to form well-separated feature clusters due to a lack of inductive bias to recognize local features such as cell membrane edges. Failure analysis shows that ViT-B/16 often misclassifies adipose cells as mucus and smooth muscle as tumors. In terms of efficiency, ResNet-50 is 5.8 times lighter in storage size and has lower inference latency. This study concludes that CNN-based architecture (ResNet-50) is still far superior, more stable, and more feasible for clinical implementation than ViT-B/16 in the context of medium-scale histopathological image classification.
Integration of Static Knowledge and Telemetry Data for an IoT-Based Customer Support AI Agent Jauhar Maknun Adib; Wahyu Purnama Magribi; Muhammad Fazly Qusyairy; Eric Julianto; Khusnul Fajri Rhomadon
International Journal Software Engineering and Computer Science (IJSECS) Vol. 6 No. 3 (2026): DECEMBER 2026
Publisher : Lembaga Komunitas Informasi Teknologi Aceh (KITA), Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35870/ijsecs.v6i3.7875

Abstract

The operational effectiveness of artificial intelligence agents in customer support is frequently attributed to the generative sophistication of underlying language models, although answer reliability primarily depends on the structure and quality of the referenced knowledge base. In complex technical domains such as the Internet of Things (IoT), static documentation alone often fails to resolve customer inquiries that involve real-time device states, sensor readings, and connectivity logs. This study investigates how progressive tiers of knowledge integration affect the response quality of an AI agent within an IoT customer support context. Employing a mixed-method pilot design combining literature synthesis, comparative analysis, experimental testing, and conceptual exploration, thirty representative questions were classified into informational, status, and diagnostic categories. These queries were evaluated across four operational scenarios: without retrieval-augmented generation (RAG), article-based RAG, article plus device metadata records, and a comprehensive hybrid framework integrating static articles, device records, and telemetry streams. Answer quality was measured using Cosine Similarity, BERTScore (F1), and a randomized blind human evaluation assessing relevance, correctness, and usefulness. The results demonstrate that while curated static documentation adequately addresses procedural informational queries, resolving status and diagnostic issues necessitates real-time operational context. The hybrid integration delivered the highest overall performance, achieving a human-evaluation score of 4.40 compared with 2.53 for article-based RAG (p = 0.0039). These empirical findings confirm that robust IoT support agents require dynamic knowledge governance that bridges versioned documentation, synchronized asset metadata, and live telemetry data within a unified retrieval framework.
Co-Authors Akbar, Habibullah Eric Julianto Jauhar Maknun Adib Muhammad Fazly Qusyairy Muhammad Fazly Qusyairy Wahyu Purnama Magribi