Claim Missing Document
Check
Articles

Found 16 Documents
Search

Optimizing EfficientNet for imbalanced medical image classification using grey wolf optimization Khusnul Khotimah; Sugiyarto Surono; Aris Thobirin
Computer Science and Information Technologies Vol 6, No 2: July 2025
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/csit.v6i2.p112-121

Abstract

The advancement of deep learning in computer vision has result in substantial progress, particularly in image classification tasks. However, challenges arise when the model is applied to small and unbalanced datasets, such as X-ray data in medical applications. This study aims to improve the classification performance of fracture X-ray images using the EfficientNet architecture optimized with grey wolf optimization (GWO). EfficientNet was chosen for its efficiency in handling small datasets, while GWO was applied to optimize hyperparameters, including learning rate, weight decay, and dropout to improve model accuracy. Random cropping, rotation, flipping, color jittering, and random erasing, were used to expand the diversity of the dataset, and class weighting is applied to overcome class imbalance. The evaluation uses accuracy, precision, recall, and F1-score metrics. The combination of EfficientNetB0 and GWO resulted in an average 4.5% improvement in model performance over baseline methods. This approach provides benefits in developing deep learning methods for medical image classification, especially in dealing with small and imbalanced datasets.
The Integration of Hybrid and Game-Based Learning in Physics Education: A Systematic Review of Their Effects on Critical Thinking and Problem-Solving Skills (2018 - 2024) Efi Kurniasari; Sugiyarto Surono; Ishafit Ishafit; Dian Artha Kusumaningtyas; Dwi Sulisworo
Indonesian Review of Physics Vol. 8 No. 2 (2025)
Publisher : Universitas Ahmad Dahlan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12928/irip.v8i2.15888

Abstract

This study aims to analyze the impact of integrating hybrid and game-based learning on critical thinking and problem-solving skills in physics education and to identify research trends from 2018 to 2024. The method used was a systematic literature review following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. The literature search was conducted through Scopus, Web of Science, and Accredited National Journals databases using a combination of relevant keywords. Inclusion criteria included empirical research articles published in English or Indonesian between 2018 and 2024. The selection process involved the stages of identification, screening, eligibility assessment, and inclusion. A total of 16 articles were analyzed further. Data were analyzed using thematic synthesis techniques to identify patterns, methodological trends, and key findings from the reviewed studies. The results indicate that hybrid learning has a more consistent impact on enhancing critical thinking and problem-solving skills, particularly through problem-based approaches and technological support. Meanwhile, game-based learning contributes more to improving motivation, engagement, and understanding of physics concepts. This study contributes a thematic synthesis and comparative analysis of the integration of hybrid and game-based learning and serves as a conceptual foundation for the development of technology-based physics learning models to enhance critical thinking and problem-solving.
Enhancing Multi-Class Classification of Non-Functional Requirements Using a BERT-DBN Hybrid Model Badzliana Aqmar Suris; Aris Thobirin; Sugiyarto Surono; Mohamed Naeem Antharathara Abdulnazar
INTENSIF: Jurnal Ilmiah Penelitian dan Penerapan Teknologi Sistem Informasi Vol 9 No 2 (2025): August 2025
Publisher : Universitas Nusantara PGRI Kediri

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29407/intensif.v9i2.24637

Abstract

Background: Software requirements classification is essential to group Non-Functional Requirements (NFR) into several aspects, such as security, usability, performance, and operability. The main challenges in NFR classification are data limitations, text complexity, and high generalization needs. Objective: This research seeks to create a classification model using a hybrid of BERT and DBN, optimize hyperparameters, and improve data representation. Methods: A BERT and DBN-based approach is used, where DBN enhances BERT's ability to extract hierarchical features. Bayesian Optimization determines the optimal hyperparameters and data augmentation is applied to enrich the dataset variation. The model is tested on the PROMISE dataset consisting of 625 data. Results: The BERT-DBN model achieves 95% accuracy on the baseline configuration and 94% on the extensive configuration, better than the previous model, BERT-CNN. The model shows stability without any indication of overfitting. Conclusion: The combination of data augmentation, hyperparameter optimization, and DBN's ability to capture hierarchical patterns improves the accuracy of NFR classification, making it more effective than existing methods, and is expected to enhance text-based classification for software requirements.
Karakteristik Sosiodemografi dan Klinis Pasien Osteoarthritis pada Berbagai Tingkat Keparahan Berdasarkan Klasifikasi Kellgren-Lawrence Muhammad Nuh; Endang Darmawan; Sugiyarto Surono; Abdul Fadlil; Adnan Adnan
Sinteza Vol. 6 No. 2 (2026): August
Publisher : Universitas Hamzanwadi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29408/sinteza.v6i2.35895

Abstract

Osteoarthritis (OA) is the most common degenerative joint disease, and its severity is commonly assessed using the Kellgren–Lawrence (KL) classification. Understanding patient characteristics across different severity levels is essential to support clinical evaluation and optimize patient management. This study aimed to analyze the sociodemographic, clinical, and laboratory characteristics of patients with OA according to KL severity. A retrospective analytical observational study was conducted using the medical records of 1,124 patients diagnosed with OA at Sultan Agung Islamic Hospital, Semarang, Indonesia, between January 2020 and January 2025. Sociodemographic, clinical, and laboratory data were collected from electronic medical records. Descriptive statistics were used to summarize patient characteristics, while differences across KL severity levels were analyzed using the Chi-square test with a significance level of p<0.05. Significant differences in OA severity were observed according to age and body mass index (p<0.05), whereas sex was not significantly associated with KL severity (p>0.05). Clinical characteristics, including OA type, comorbidity status, number and type of comorbidities, and pain intensity, differed significantly across KL grades (all p<0.05). Laboratory parameters, including blood pressure, blood glucose, renal function, and lipid profile, also demonstrated significant differences according to OA severity (all p<0.05). These findings indicate that OA severity is associated with distinct sociodemographic, clinical, and laboratory characteristics, providing valuable evidence to support risk stratification, clinical assessment, and individualized management of patients with OA.
Comparative Analysis of CNN Architectures and Optimizers for ALL Classification Sugiyarto Surono; Fanita Damayanti; Frenky Riski Gilang Pratama; Attarik Mohammad; Maretta Mia Audina; Marfungah Wati
Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi) Vol 10 No 4 (2026): August 2026
Publisher : Ikatan Ahli Informatika Indonesia (IAII)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29207/resti.v10i7506

Abstract

Deep learning has become an important approach for medical image analysis, particularly for automated image classification. Among its various architectures, Convolutional Neural Networks (CNNs) are frequently employed to extract and distinguish visual patterns from medical images. However, the performance of a CNN may vary depending on the optimization algorithm used during the training process. This study investigates the effect of four optimization methods on different CNN architectures for the classification of Peripheral Blood Smear (PBS) images. A publicly available Kaggle dataset comprising 3,242 images and four diagnostic categories was used in the experiments. Six architectures, namely ResNet50V2, Xception, EfficientNetB2, EfficientNetB4, EfficientNetV2-S, and EfficientNetV2-M, were evaluated in combination with Adam, AdamW, NAdam, and RMSprop. Before training, the images underwent resizing, normalization, and augmentation procedures. The dataset was subsequently partitioned into training, validation, and test subsets using proportions of 80%, 10%, and 10%, respectively. Classification performance was analyzed using accuracy, loss, and confusion matrix-based evaluation. Among all tested combinations, Xception with RMSprop produced the best result, obtaining 99.69% accuracy and a loss value of 0.0194. Conversely, the lowest result was observed for ResNet50V2 with RMSprop, which achieved 93.85% accuracy. The experimental findings indicate that the optimizer can substantially influence CNN classification performance, and the effectiveness of an optimization method is dependent on the architecture with which it is combined.
Machine Learning-Based Clustering for Optimizing Zakat, Infaq, and Sadaqah Fund Distribution Among Ahmad Dahlan University Students Safika Maranti; Sugiyarto Surono; Nurhidayati Kasrul Jalil; Khusnul Khotimah; Sintia Afriyani
Jurnal Internasional Ekonomi Islam Vol 8 No 02 (2026): International Journal of Islamic Economics
Publisher : Universitas Islam Negeri Jurai Siwo Lampung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32332/ijie.v8i02.13481

Abstract

Objective: This study aims to identify latent patterns in the distribution of zakat, infaq, and sadaqah funds among student beneficiaries at Universitas Ahmad Dahlan by grouping recipients according to their socio-economic, academic, and social characteristics. Method: This study adopts a quantitative approach using unsupervised machine learning techniques. The K-Means and Density-Based Spatial Clustering of Applications with Noise algorithms were applied to secondary data from Lazismu Ahmad Dahlan University for 2023–2026. After cleaning, the study retained 185 valid recipient records from an initial dataset of 194 observations. Data preprocessing included encoding and normalization. The study compares centroid-based (K-Means) and Density-Based Spatial Clustering of Applications with Noise clustering methods to identify recipient segments. Results: The findings show that Density-Based Spatial Clustering of Applications with Noise produces more compact and better-separated clusters than K-Means. The algorithm also identifies atypical recipients as noise, revealing unique beneficiary profiles. The results indicate substantial heterogeneity among recipients. Some economically vulnerable students demonstrate strong academic performance yet receive relatively lower assistance, while other groups receive higher levels of support despite less vulnerable conditions. Implications: The findings suggest that clustering analysis can support more objective, transparent, and adaptive ZIS fund distribution. Integrating such analysis into planning, implementation, and evaluation processes may improve targeting accuracy and promote equitable, evidence-based decision-making while still considering individual circumstances.