Article Per Year (5 Year)
p-Index From 2021 - 2026
0.882
P-Index
This Author published in this journals
All Journal Dinamik Jurnal Media Computer Science CHAIN: Journal of Computer Technology, Computer Engineering and Informatics Jurnal Ilmu Komputer dan Informatika
Ridwan Mahenra
Universitas Teknokrat Indonesia
Computer Science & IT Control & Systems Engineering Electrical & Electronics Engineering Engineering Mathematics

Published : 5 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 5 Documents
Search

 1 
Pengembangan Model Hybrid untuk Identifikasi Tuberkulosis Pada Gambar Rontgen Dada Ridwan Mahenra
CHAIN: Journal of Computer Technology, Computer Engineering, and Informatics Vol. 3 No. 1 (2025): Volume 3 Number 1 January 2025
Publisher : PT. Tech Cart Press

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.58602/chain.v3i1.166

Abstract

Tuberkulosis (TB), penyakit menular parah yang berdampak pada jutaan orang di seluruh dunia, umumnya didiagnosis dengan menggunakan rontgen dada. Untuk memastikan diagnosis yang akurat, terutama pada tahap awal, para profesional di bidang kesehatan mengandalkan dukungan teknologi canggih. Tidak seperti model yang sudah ada yang terutama berfokus pada deteksi TB pada gambar sinar-X, penelitian ini bertujuan untuk mengklasifikasikan gambar yang berhubungan dengan TB untuk memfasilitasi pemilihan metode yang tepat untuk deteksi TB yang tepat. Pendekatan yang diusulkan menggabungkan kemampuan yang kuat dari arsitektur VGG16 dengan jaringan syaraf tiruan (CNN) untuk tujuan klasifikasi. Memanfaatkan keefektifan VGG16 dalam menangkap fitur gambar yang penting, kami memodifikasinya untuk ekstraksi fitur untuk mengidentifikasi tanda-tanda TB pada gambar sinar-X. Untuk klasifikasi, CNN digunakan untuk mengkategorikan gambar yang terkena TB. Metode yang diusulkan ini dievaluasi menggunakan dataset standar, yang menunjukkan kinerja yang unggul dalam hal akurasi, recall, dan presisi dibandingkan dengan teknik yang ada saat ini.
Analisis Komparatif Teknik Quantization (INT4 vs INT8 vs FP16) terhadap Kualitas Output Large Language Model Berbasis Instruksi Ridwan Mahenra; Mahendra Dewantoro
Jurnal Ilmu Komputer dan Informatika | E-ISSN : 3063-9026 Vol. 3 No. 1 (2026): Juli - September
Publisher : GLOBAL SCIENTS PUBLISHER

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Quantization is a model compression technique that reduces the numerical precision of Large Language Model (LLM) weights to lower memory requirements and accelerate inference. This study presents a comparative analysis of three widely used quantization schemes—INT4, INT8, and FP16—using the Mistral-7B-Instruct-v0.2 model as the subject of study. Evaluation was conducted across five output quality dimensions: BLEU score, ROUGE-L score, perplexity, BERTScore, and average inference latency. Testing utilized 200 instruction prompts covering four task categories: text summarization, factual question answering, code generation, and logical reasoning. The analysis results indicate that INT8 offers the best balance between computational efficiency and output quality, showing an average performance degradation of 2.3% compared to FP16 while achieving a 48% reduction in memory requirements. INT4 exhibited a more significant degradation of 7.8% in logical reasoning tasks, despite successfully reducing memory usage by 74%. These findings provide practical guidance for researchers and practitioners in selecting a quantization scheme suited to available computational resources.
Evaluasi Kualitas Rangkuman Teks Otomatis Menggunakan Algoritma ROUGE Studi Komparatif Model Google Gemini dan OpenAI ChatGPT Ridwan Mahenra; Reza Pajriansyah
Dinamik Vol 31 No 2 (2026)
Publisher : Universitas Stikubank

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35315/dinamik.v31i2.10485

Abstract

The rapid development of large language models has significantly improved the performance of automatic text summarization systems. However, each model may demonstrate different levels of effectiveness depending on its architectural characteristics and training approach. This study aims to evaluate the quality of Indonesian news text summarization generated by Google Gemini and OpenAI ChatGPT. A total of fifty news articles were used as the dataset, each accompanied by a manually written summary to serve as the ground truth. The research procedure involved web scraping, text preprocessing, automated summarization using both models, and performance evaluation through ROUGE metrics, including ROUGE-1, ROUGE-2, and ROUGE-L. The results show that ChatGPT consistently achieves higher ROUGE-1 and ROUGE-2 scores compared with Gemini, indicating a better ability to preserve key terms and lexical relations within the text. Meanwhile, ROUGE-L scores for both models are relatively close, suggesting that Gemini remains competitive in maintaining overall summary structure. Distribution analysis also reveals that ChatGPT demonstrates more stable performance across various text types. Overall, this study concludes that ChatGPT provides more accurate and consistent automatic summarization results than Gemini for Indonesian news texts.
Analisis Komparatif Performa Algoritma Random Forest dan Gradient Boosting dalam Klasifikasi Data Finansial Ridwan Mahenra; Hutriatmo Ilham Dito Armando
Dinamik Vol 31 No 2 (2026)
Publisher : Universitas Stikubank

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35315/dinamik.v31i2.10489

Abstract

This study aims to analyze and compare the performance of the Random Forest and Gradient Boosting algorithms in classifying credit card default risk using the Default of Credit Card Clients dataset. The dataset consists of 30,000 entries with 24 financial and demographic variables representing customers’ payment behavior over the past six months. The research procedure includes data acquisition and exploration, preprocessing through feature standardization, stratified data splitting, model construction, and performance evaluation using accuracy, precision, recall, and F1-score metrics. The experimental results indicate that both algorithms successfully capture complex financial patterns; however, Gradient Boosting demonstrates superior performance, particularly in recall and F1-score, highlighting its better sensitivity to default cases. The feature importance analysis confirms that payment history variables, especially PAY_0, play a major role in influencing model predictions. Overall, this study recommends the use of boosting-based models for credit risk prediction, particularly when dealing with imbalanced datasets, due to their ability to learn minority patterns and iteratively reduce prediction errors.
Optimizing Multi-Class Imbalance in Employee Attrition Prediction Using Non-dominated Sorting Genetic Algorithm II with Fairness-Driven Fitness Function Erick Yoga Res Hendra; Ridwan Mahenra
Jurnal Media Computer Science Vol 5 No 1 (2026): Januari
Publisher : LPPJPHKI Universitas Dehasen Bengkulu

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37676/jmcs.v5i1.9230

Abstract

Class imbalance in multi-class employee attrition prediction is a major challenge in human resource analytics, causing bias towards the majority class and poor performance on the minority class. This study proposes a multi-objective genetic algorithm (GA) framework using NSGA-II to address multi-class imbalance in the IBM HR Analytics Employee Attrition & Performance dataset. By optimising precision, recall, and fairness (normalised Demographic Parity Difference), this framework generates synthetic samples for minority classes (Resign, Retire_Termination) through two-class adaptive clustering and a weighted fitness function (β=0.3). Experiments were conducted with the XGBoost classifier, comparing GA with the SMOTE baseline. Results show that GA achieves a macro F1-score of 0.65 ± 0.02, surpassing SMOTE (0.56 ± 0.03), with significant improvements in Resign (F1-score 0.59 vs. 0.51) and Retire_Termination (F1-score 0.42 vs. 0.24). The fairness value of GA (0.82 ± 0.02) was higher than that of SMOTE (0.75 ± 0.03), indicating fairer predictions. Visualisation of the Pareto front and convergence of GA illustrates the trade-off between objectives and algorithm robustness. Key contributions include a GA framework that integrates fairness, advantages over SMOTE, and flexibility for HR applications.
Co-Authors Erick Yoga Res Hendra Hutriatmo Ilham Dito Armando Mahendra Dewantoro Reza Pajriansyah