Sinkron : jurnal dan penelitian teknik informatika
Not a member yet
1261 research outputs found
Sort by
Metaheuristic-Optimized SVM for Stunting Risk Detection in Pregnancy
No full textStunting is a chronic growth disorder that originates during pregnancy, making early risk detection crucial for effective prevention and long-term child development. This study introduces a stunting risk prediction model based on urine testing, employing a Support Vector Machine (SVM) algorithm enhanced through metaheuristic optimization. Three metaheuristic algorithms—Grey Wolf Optimizer (GWO), Simulated Annealing (SA), and Firefly Algorithm (FA)—were utilized to fine-tune the SVM hyperparameters (C and gamma). Clinical urine samples collected from pregnant women served as the dataset for model training and validation. The results indicate that the SVM model optimized using GWO achieved the highest prediction accuracy at 94.15%, outperforming both the default SVM (88.46%) and the models optimized using SA (94.12%) and FA (85.71%). Additionally, significant improvements were observed in precision, recall, and F1-score metrics, affirming the effectiveness of metaheuristic tuning in enhancing classification performance. These findings highlight the potential of integrating metaheuristic algorithms with SVM for robust medical prediction tasks, especially in the early detection of stunting risks. The proposed model offers a promising and non-invasive diagnostic approach that can be implemented in prenatal care settings, enabling timely interventions to mitigate stunting and improve maternal and child health outcomes
LDA Topic Modeling: Twitter-Based Public Opinion on Indonesian Ministry of Finance
No full textPeople in the modern era use social media daily to exchange opinions regarding government policies, such as discussions related to the Indonesian Ministry of Finance (Kemenkeu). This study aims to analyze the topics of discussion about the Ministry of Finance on the Twitter platform, now known as 'X', and to determine the results of more effective preprocessing. The data in this study was taken from Twitter using the Tweet Harvest Tool with the keyword 'Ministry of Finance' from January 2024 to July 2024. The data is then processed through cleaning, preprocessing, calculation of coherence values, LDA modeling, and visualization. The preprocessing process includes several scenarios to compare the best results that are easy for the reader to understand. The highest coherence value obtained is 0.572250 by using stemming from NLTK library. The most effective preprocessing results are normalization, tokenization, stopwords, and stemming using Sastrawi. Modeling is done to find latent topics through LDA topic modeling techniques. Visualizing the intertopic distance map provides information on the distance between each topic. The results show that the distance between one topic and another has a variety of distance variations. This study shows that social media platforms can serve as a source of evaluation for the Indonesian government. The findings of these topics are helpful as insights for readers and the Kemenkeu. Finally, the analysis identified several key topics in public discussion, including fiscal policy, budget transparency, and the Ministry of Finance's performance in addressing current economic issues.
A Systematic Review of Retrieval-Augmented Generation for Enhancing Domain-Specific Knowledge in Large Language Models
No full textThis literature review examines the use of Retrieval-Augmented Generation (RAG) in enhancing Large Language Models (LLM) for domain-specific knowledge. RAG integrates retrieval techniques with generative models to access external knowledge sources, addressing the limitations of LLMs in handling specialized information. By leveraging external data, RAG improves the accuracy and relevance of generated content, making it particularly useful in fields that require detailed and up-to-date knowledge. This review highlights the effectiveness of RAG in overcoming challenges such as data sparsity and the dynamic nature of specialized knowledge. Furthermore, it discusses the potential of RAG to enhance LLM performance, scalability, and the ability to generate contextually accurate responses in knowledge-intensive applications. Key challenges and future research directions in the implementation of RAG for domain-specific knowledge are also identified
Convolutional Neural Network Algorithm Implementation for Classifying Traditional Wood Carving Motifs of Patra Bali
No full textThis research develops an automatic classification system to recognize Balinese Patra carving motifs using deep learning method based on Convolutional Neural Network (CNN). The data used are images of Cina Patra, Mesir Patra, Punggel Patra, and Sari Patra motifs, which have gone through preprocessing stages such as cropping, resizing, and augmentation in the form of flip and rotation to increase data variation. Three pre-trained CNN models were used in testing, namely DenseNet169, InceptionResNetV2, and MobileNetV2. The training process was performed with Adam optimization, batch size 32, and 100 epochs. Model performance evaluation was performed using accuracy and confusion matrix metrics. The results show that all three models were able to achieve 100% accuracy on the test data, with MobileNetV2 recording the lowest loss of 0.75%, followed by DenseNet169 (1.14%) and InceptionResNetV2 (1.18%). Based on the confusion matrix, all motifs were recognized very well, although there was a slight misclassification of the Patra Sari motif by the InceptionResNetV2 model. These findings prove that CNN is effectively used in the recognition of traditional carving motifs and has the potential to support cultural preservation through interactive visual technology
Hybrid Genetic Algorithm for Dynamic Portfolio Optimization Problems
No full textDynamic portfolio optimization is a complex problem due to continuous changes in market conditions, demanding algorithms capable of effective adaptation. Genetic Algorithms (GA) are often used for optimization problems but may face limitations in convergence speed and solution precision. This research aims to develop and evaluate a Hybrid Genetic Algorithm (HGA) that integrates GA with the Hill Climbing local search method, and to compare its performance against standard GA in solving dynamic portfolio optimization problems with the objective of maximizing the Sharpe Ratio. A series of simulation-based experiments were conducted by varying key algorithmic and dynamic environment parameters. Simulation results indicate that HGA generally has significant potential to improve performance compared to standard GA. Consistently, HGA successfully achieved superior solution quality, both in terms of Offline Performance Solution Quality and Overall Best Fitness. Regarding robustness to dynamic changes, HGA also demonstrated a smaller impact from performance degradation and a more promising recovery capability after market environment changes. Although HGA's superiority in convergence speed is not always absolute and the implementation of Hill Climbing adds to the computational time per generation, the improvement in solution quality and robustness offered in many configurations can be considered a worthwhile trade-off, especially for complex dynamic portfolio optimization problems. These findings support the hypo that hybridizing GA with local search can provide a positive contribution, noting that careful parameter tuning is crucial for maximizing HGA's potential
MLP Model Optimization for Heart Attack Risk Prediction: A Systematic Literature Review
No full textHeart disease remains a leading cause of global mortality, making the development of accurate predictive models a clinical priority. While Multilayer Perceptron (MLP) models offer significant potential, their application is hindered by challenges in optimization, data imbalance, and interpretability. This systematic literature review aims to address these issues by synthesizing current research on MLP model optimization for heart disease prediction, focusing on strategies for handling class imbalance and achieving model transparency with SHapley Additive exPlanations (SHAP). Following PRISMA guidelines, a structured search of major scientific databases resulted in the in-depth analysis of 30 peer-reviewed studies. The findings indicate that MLP optimization is increasingly sophisticated, employing automated hyperparameter tuning and novel architectures. For class imbalance, the Synthetic Minority Over-sampling Technique (SMOTE) is the predominant data-level solution, though a trend towards advanced algorithm-level techniques is emerging. The application of SHAP has successfully validated models by confirming the importance of known clinical risk factors like age and chest pain type, while also demonstrating potential for new discovery. This review concludes by providing a comprehensive roadmap for researchers, highlighting a critical need for comparative studies on imbalance techniques, deeper applications of explainable AI for local-level analysis, and a stronger focus on validation using large-scale, real-world clinical data to develop truly robust and trustworthy predictive systems
Creditworthiness Classification Utilizing AHP-SVM Based on 5C Criteria
No full textCredit risk occurs when borrowers fail to meet loan repayment obligations, posing significant challenges to the financial stability of lending institutions. Accurate classification of creditworthiness is essential to mitigate such risks. This study proposes a hybrid approach that integrates the Analytical Hierarchy Process (AHP) and Support Vector Machine (SVM) to evaluate borrower eligibility based on the 5C model: Character, Capacity, Capital, Collateral, and Condition. The AHP method is used to assign weights to credit attributes based on expert judgment, while SVM performs the classification. Three experiments were conducted to compare the effectiveness of different feature selection strategies: (1) expert-defined 5C attributes, (2) AHP weighting conducted by experts, and (3) AHP weighting conducted by non-experts. Experimental results show that the 5C-SVM model achieved the highest performance with 96% accuracy, followed by AHP-SVM (expert) with 95% and AHP-SVM (non-expert) with 93%. The findings indicate that expert involvement in the feature selection process significantly improves model performance. This study demonstrates the effectiveness of combining domain knowledge with machine learning in building intelligent decision support systems for credit risk analysis. The proposed approach offers practical value for financial institutions seeking more objective, accurate, and consistent credit evaluation processes. Furthermore, it opens new opportunities for integrating expert-based reasoning with automated analytics in financial decision-making.
Analyzing User Acceptance of NFJuara Mobile Application Using TAM and D&M IS Success Model
No full textThis study purposes to know how NFJuara application is accepted by the users in Nurul Fikri Lampung using the Technology Acceptance Model (TAM) Integrated with D&M IS Success Model. Data was collected by a validated questionnaire with inner model and outer model testing using PLS-SEM software SmartPLS. The type of data in this study is a quantitative approach. The number of samples collected was 143 respondents. Results of this research show that one of the hypotheses is rejected, that is, Service Quality (SEQ) does not affect Perceived Usefulness (PU) significantly. Besides that, this study shows that Perceived Usefulness (PU) and Perceived Ease of Use (PEU) affect as significant Acceptance of IT (AI) with R2=0.59 (Moderate) and β=0,36 (PUàAI), β=0,46 (PEUàAI). These findings imply that developers of NFJuara applications need to improve the service quality to increase acceptance, although overall NFJuara application is accepted by the user because they still feel the benefits and usefulness of the application. The contribution of this study lies in testing the technology acceptance model in the context of mobile learning, which enriches the literature on the adoption of application-based e-learning, as well as providing practical recommendations for application developers to enhance user experience.
Implementation of a Hybrid Cryptosystem Using ChaCha20 and ECC for Image Encryption in an Android Application
No full textThis study aims to develop an Android application capable of securely encrypting and decrypting images using a hybrid cryptographic method. The system combines the ChaCha20 algorithm as symmetric cryptography to encrypt image files, and Elliptic Curve Cryptography (ECC) as asymmetric cryptography to encrypt the ChaCha20 key. The key used is temporary (ephemeral), ensuring that only the intended recipient who possesses the appropriate ECC private key can decrypt the file. The application was developed using the Kotlin programming language in Android Studio, with a PHP-based backend and MySQL database. Testing was conducted using the black-box method and involved 15 beta testers to evaluate functionality, security, and usability aspects. The results show that all features of the application run properly, and the encryption and decryption processes can be performed efficiently and securely. Beta testers gave an average rating of 4.6 out of 5 and stated that the application is easy to use and provides sufficient protection for personal data. Therefore, the developed application successfully meets the objectives of the study and offers an alternative solution for securing image file transfers between users via Android devices
Comparative Analysis of DNA Sequence Alignment Algorithms in SARS-CoV-2
No full textSequence alignment is fundamental in bioinformatics, with Smith-Waterman (local) and Needleman-Wunsch (global) algorithms widely applied. However, comparative analyses on highly similar viral genomes such as SARS-CoV-2 remain scarce. This study systematically evaluated both algorithms using the first 5,000 nucleotides of two SARS-CoV-2 genomes (29,903 and 29,684 nt) under four parameter configurations: standard, low gap penalty, high gap penalty, and high match reward. Performance was assessed through alignment score, sequence identity, gap distribution, execution time, and parameter sensitivity. Both algorithms produced identical sequence identity (97.80%), with 4,943 matches out of 5,054 positions. Smith-Waterman consistently yielded higher alignment scores (12.6-112 points advantage), while Needleman-Wunsch was substantially faster (0.7752 vs 3.9014 s), showing 5.03 times greater computational efficiency. These findings indicate that both methods are reliable for highly similar viral sequences, with a trade-off between scoring precision and computational speed. This study provides the first parameter-sensitive comparison for full SARS-CoV02 genomes, emphasizing how parameter tuning can influence performance outcomes. A key limitation is that the analysis was restricted to the first 5,000 nucleotides, which may not capture variability across the complete genome