Article Per Year (5 Year)
p-Index From 2021 - 2026
1.066
P-Index
This Author published in this journals
All Journal Jurnal Sistem Komputer dan Informatika (JSON) TIN: TERAPAN INFORMATIKA NUSANTARA Bulletin of Computer Science Research
Muhammad Affandes
Universitas Islam Negeri Sultan Syarif Kasim Riau, Pekanbaru
Aerospace Engineering Arts Automotive Engineering Biochemistry, Genetics & Molecular Biology Chemical Engineering, Chemistry & Bioengineering Civil Engineering, Building, Construction & Architecture Computer Science & IT Control & Systems Engineering Decision Sciences, Operations Research & Management Dentistry Earth & Planetary Sciences Economics, Econometrics & Finance Energy Engineering Environmental Science Health Professions Immunology & microbiology Industrial & Manufacturing Engineering Law, Crime, Criminology & Criminal Justice Materials Science & Nanotechnology Mathematics Mechanical Engineering Medicine & Pharmacology Neuroscience Nursing Physics Social Sciences Veterinary Other

Published : 6 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 6 Documents
Search

 1 
Klasifikasi Citra Daging Babi dan Daging Sapi Menggunakan Deep Learning Arsitektur ResNet-50 dengan Augmentasi Citra Sarah Lasniari; Jasril Jasril; Suwanto Sanjaya; Febi Yanto; Muhammad Affandes
Jurnal Sistem Komputer dan Informatika (JSON) Vol. 3 No. 4 (2022): Juni 2022
Publisher : Universitas Budi Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30865/json.v3i4.4167

Abstract

Beef is an example of an animal protein-rich food. The consumption of meat in Indonesia is increasing year after year, in tandem with the country's growing population. Many traders purposefully combine beef and pork in order to maximize profits. With the naked eye, it's difficult to tell the difference between pork and beef. In Muslim-majority countries, the assurance of halal meat is crucial. This study uses Deep Learning with the Convolutional Neural Network (CNN) method and ResNet-50 with data augmentation to classify images of beef and pork. The original meat picture databases contain 457 images, however following the data augmentation process, there are 2742 images in total, divided into three classes. The distribution of training and test data is 90 percent:10 percent in the comparison test scenario between the two original data schemes and supplemented data. With an average of 87.64 % accuracy, 87.59 % recall, and 90.90 % precision, the Confusion Matrix is the best classification performance model. There was no evidence of overfitting based on observations from the visualization of the training and testing process.
Penerapan Saliency Maps dalam Explainable AI Untuk Deteksi Penyakit Paru-Paru pada Citra X-Ray Dada dengan Deep Learning Wahyu Reinaldy; Benny Sukma Negara; Muhammad Irsyad; Muhammad Affandes; Surya Agustian
TIN: Terapan Informatika Nusantara Vol 7 No 1 (2026): June 2026
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/tin.v7i1.9962

Abstract

Early identification of lung diseases is very important so that medical personnel can quickly provide first aid and further study the patient's condition. In this study, a model was developed to classify chest X-ray images of the lungs using the VGG16 architecture. These chest X-ray images were categorized into three groups: COVID-19, normal lungs, and pneumonia. A combination of hyperparameters, including a learning rate of 0.001, 50 epochs, and a batch size of 16, was used to train the model, achieved an accuracy of 96%. Several evaluation metrics, including precision, recall, f1-score, and confusion matrix, were used to assess the model. In addition, saliency map methods were used to visually interpret the model's prediction output and display the areas of the chest X-ray images that most influenced the model's decision-making. The saliency map visualization findings show that the model focuses its predictions on regions of the lungs associated with the disease, which helps in understanding the algorithm's decision-making process.
Analisis Komparatif Konfigurasi Multilayer Perceptron pada Classifier Head RoBERTa untuk Klasifikasi Ujaran Kebencian Ikhwan Habibi; Surya Agustian; Jasril Jasril; Muhammad Affandes
TIN: Terapan Informatika Nusantara Vol 7 No 2 (2026): July 2026
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/tin.v7i2.10512

Abstract

The widespread dissemination of hate speech and offensive language on social media has increased the demand for accurate automated text classification systems. Although RoBERTaForSequenceClassification has been widely used for various for text classification task, the effect of its default classifier head configuration on classification performance has not yet been systematically evaluated. As the main contribution, this study conducts a controlled evaluation of 32 multilayer perceptron (MLP)-based classifier head configurations, varying the number of hidden layers, activation functions, and dropout rates, against the default classifier head on the English HASOC 2021 dataset for two subtasks: binary and multiclass classification. Each configuration was evaluated using Stratified 5-Fold Cross-Validation with Macro-F1 as the evaluation metric, after which the best-performing configuration was further evaluated on an independent test set. For the binary task, the best configuration achieved a test Macro-F1 of 80.90%, about 0.3 percentage points higher than the baseline's 80.59%. For the multiclass task, the configuration with the highest validation performance instead achieved a test Macro-F1 of 65.68%, about 0.4 percentage points lower than the baseline's 66.11%, showing that an advantage observed during cross-validation does not always hold on the test set. Further analysis revealed that excessively deep hidden layers combined with aggressive dimensional compression can sharply degrade performance on the multiclass task. These findings indicate that the effect of classifier head configuration is small and task-dependent, so systematic evaluation remains necessary before adopting a given configuration in place of the default classifier head when fine-tuning RoBERTa-based models.
Implementasi Langchain dan Large Language Models Dalam Automatic Question Generation Untuk Computer Assisted Test Novri Rahman; Nazruddin Safaat Harahap; Muhammad Affandes; Pizaini Pizaini
Bulletin of Computer Science Research Vol. 5 No. 4 (2025): June 2025
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bulletincsr.v5i4.558

Abstract

The advancement of Artificial Intelligence (AI), particularly Large Language Models (LLM), presents new opportunities in transforming educational assessment systems. This study aims to implement the LangChain framework integrated with LLM for an Automatic Question Generation (AQG) system within a Computer Assisted Test (CAT) platform, using eleventh-grade Biology subject matter as a case study. The methodology includes data collection from PDF-based instructional materials, text embedding using Facebook AI Similarity Search (FAISS) as the knowledge base, and automatic question generation through the GPT-4o model. The system is developed using a microservices architecture comprising frontend and backend services built with the Next.js, FastAPI, and Express.js frameworks. System evaluation was conducted using the User Acceptance Test (UAT) and the DeepEval framework. The evaluation results show a teacher satisfaction rate of 92.7% and a positive response from students at 67.5%. Meanwhile, the DeepEval assessment reported average scores of 3,69% for hallucination, 97,44% for contextual precision, 83,30% for contextual relevancy, 70,63% for answer relevancy, and 92,47% for prompt alignment. These findings indicate that the integration of LangChain and LLM is effective in generating contextually accurate and relevant questions, although improvements are still needed in answer relevancy. This study is expected to provide an efficient solution for digital-based educational assessment and contribute to future developments in educational AI.
Optimasi Klasifikasi Hate Speech dan Offensive Language melalui Frozen RoBERTa Feature Extraction dan Random Forest Marsha Cahyani Dwisyakilla; Surya Agustian; Novriyanto Novriyanto; Muhammad Affandes
Bulletin of Computer Science Research Vol. 6 No. 4 (2026): June 2026
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bulletincsr.v6i4.1157

Abstract

Hate speech and offensive content detection on social media remains a significant challenge in Natural Language Processing (NLP) due to the characteristics of Twitter data, which are typically short, informal, and contain various elements such as mentions, URLs, hashtags, and emotional expressions that complicate the classification process. End-to-end Transformer fine-tuning approaches generally require substantial computational resources; therefore, this study explores a more computationally efficient approach by utilizing RoBERTa as a frozen feature extractor combined with Random Forest as the classifier. This approach enables the exploitation of contextual representations generated by Transformer models without requiring full model retraining.The study employs the HASOC 2021 English Track dataset, which consists of two classification tasks: Task A for binary classification (HOF and NOT) and Task B for multi-class classification (HATE, OFFN, PRFN, and NONE). The classification pipeline is optimized through the incorporation of handcrafted features, oversampling, Random Forest hyperparameter tuning, and threshold tuning in specific scenarios. Model performance is evaluated using accuracy, precision, recall, and F1-macro, with F1-macro serving as the primary metric due to class imbalance. The best-performing model achieved an F1-macro score of 0.80 on Task A and 0.64 on Task B. These results indicate that the combination of frozen RoBERTa representations and Random Forest provides strong performance for binary hate speech and offensive content classification. However, the performance on Task B highlights the difficulty of distinguishing linguistically similar categories, such as HATE, OFFN, and PRFN, suggesting that fine-grained multi-class classification remains a challenging task. Overall, the findings indicate that RoBERTa-based frozen feature extraction constitutes a computationally efficient alternative for hate speech detection on English Twitter data, although further improvements are required to enhance performance in multi-class classification settings.
Klasifikasi Hate Speech dan Offensive Language Menggunakan BERT dan Support Vector Machine Muhammad Tirta Syakban; Surya Agustian; Muhammad Fikry; Muhammad Affandes
Bulletin of Computer Science Research Vol. 6 No. 3 (2026): April 2026
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bulletincsr.v6i3.1061

Abstract

Hate speech and offensive language have become increasingly complex problems on social media, requiring classification approaches that can effectively capture linguistic context. While transformer-based models with end-to-end fine-tuning have become the dominant approach, the use of transformers as fixed feature extractors combined with classical machine learning algorithms remains relatively underexplored, particularly in benchmark settings such as HASOC 2021. This study aims to investigate the effectiveness of a feature-based transformer approach by combining embeddings from BERT and RoBERTa with Support Vector Machine (SVM) classifiers using multiple kernel configurations, including Linear, RBF, Polynomial, and LinearSVC. Experiments were conducted on Sub-task A and Sub-task B by comparing traditional feature-based methods (TF-IDF) with transformer-based embeddings. The experimental results show that RoBERTa embeddings consistently outperform other feature extraction methods. On the test dataset, the combination of RoBERTa and SVM achieves competitive performance compared to other systems in HASOC 2021. In Sub-task B, the optimal model achieves a Macro F1-score of 0.61, outperforming several BERT-based and classical baseline systems.These findings demonstrate that using transformer embeddings as fixed feature representations combined with optimized SVM classifiers can serve as an effective alternative to fine-tuning approaches, particularly in achieving more stable performance under class imbalance conditions. This study contributes by highlighting the potential of feature-based transformer methods as a flexible and competitive strategy for hate speech and offensive language detection.
Co-Authors Febi Yanto Ikhwan Habibi Jasril Jasril Marsha Cahyani Dwisyakilla Muhammad Fikry Muhammad Irsyad Muhammad Tirta Syakban Nazruddin Safaat H Negara, Benny Sukma Novri Rahman Novriyanto Novriyanto Pizaini Pizaini Sarah Lasniari Surya Agustian Suwanto Sanjaya Wahyu Reinaldy