Indonesian Journal of Electrical Engineering and Computer Science
Not a member yet
9109 research outputs found
Sort by
MODIS-NDVI and wheat yield patterns and predictions in Taounate, Morocco
Get PDFThis study is devoted to the use of varied analytical methods to elucidate the complex relationship between normalized difference vegetation index (NDVI) and wheat production in Taounate, Morocco based on MODIS Satellite data. Linear regression (LR), with a coefficient of determination (R²) of 0.93, provided a solid basis, while the decision tree (DT) showed significant performance with an R² of 0.81. Support vector regression (SVR) performed well with an R² of 0.96, highlighting its ability to capture the non-linear nuances of the data. Given the complexity inherent in the observed relationships, characterized by non-linear variations, we opted for a combined approach. K-means, closely linked to SVR, was integrated for its ability to identify homogeneous subgroups in the data (R2 up to 0.98). This combination made it possible to circumvent the limits of strictly linear methods, thus reinforcing the robustness of our analysis. These results underline the capacity of the chosen methodology to decode the interactions between NDVI and wheat production in the complex context of Taounate. By providing clear and nuanced perspectives, this study helps inform agricultural decisions and build resilience to climate challenges in the region
Enhanced SMS spam classification using machine learning with optimized hyperparameters
Get PDFShort message service (SMS) text messages are indispensable, but they face a significant issue with spam. Therefore, there is a need for robust models capable of classifying SMS messages as spam or non-spam. Machine learning offers a promising approach for this classification, based on existing datasets. This study explores a comparison of several techniques, including logistic regression (LR), support vector machines (SVM), gradient boosting (GB), and neural networks (NN). Hyperparameters play a crucial role in the performance of these models, and their optimization is essential for achieving high accuracy. To this end, we employ an evolutionary programming approach for hyperparameter optimization. This approach evaluates the performance of these models before and after hyperparameter optimization, aiming to identify the most effective model for SMS spam classification
Improving the MSMEs data quality assurance comprehensive framework with deep learning technique
Get PDFIn the year of 2022 the ministry of cooperatives and small and medium enterprises (SMEs) executed a complete data collection program for the cooperatives and micro small and medium enterprises (MSMEs) profile. As the complexity of the process and the uniqueness of the data characteristics, plenty of risks must be mitigated. The most challenging risk is the possibility of reduced data quality. This study is performed to validate the proposed comprehensive framework to ensure the quality data of cooperatives and MSME. The proposed framework aims to prevent, detect, repair, and recover dirty data to achieve the required data quality minimum standard. We investigated many techniques namely rule-based, selection-based, and deep learning-based. By applying the framework, 6,850,000 missing values are found and corrected, whereas the number of instant data containing attribute values that do not follow the domain constraints or integrity rule is 4,082,630. The first deep learning task applied in the framework is MSME activity image description (image captioning) generated by the convolutional neural network-recurrent neural network (CNN-RNN) model. By using 1000 MSME images as data training, the model’s performance is quite good, achieving the average BLEU score of Culinary 0,3149, Fashion 0,4868, and creative products 0,5086. So far, the proposed framework can contribute to supporting MSME one data as the Indonesian government program
ADALINE-based synchronous detection for enhanced shunt APF performance
Get PDFPower quality issues caused by current harmonics from nonlinear and unbalanced loads are a growing concern. This paper presents a novel control strategy for four-wire shunt active power filters (SAPF) that surpasses existing conventional methods in mitigating harmonics and power factor correction. The strategy employs an improved synchronous detection method (SDM) enhanced by an adaptive linear neural network (ADALINE) trained using the least mean square (LMS) algorithm. This approach accurately estimates harmonic frequencies, enabling the SAPF to generate precise compensation currents. The effectiveness of the proposed method is validated through MATLAB-Simulink simulations under balanced supply conditions, encompassing diverse load scenarios. These simulation results are compared with those obtained using instantaneous power theory (IPT). They demonstrate the ability of the proposed method to achieve excellent harmonic identification and elimination, to comply with IEEE 519 harmonic limits, to ensure sinusoidal and balanced line currents, and to compensate for reactive power and neutral current. Furthermore, its simple architecture and noise robustness make it a promising solution for enhancing power quality
Implementing zero-knowledge proof authentication on Hyperledger fabric to enhance patient privacy and access control
Get PDFIn recent years, the healthcare sector has encountered significant challenges in authenticating identities for online medical services. A predominant reliance on centralized identity management systems (IDMs) has presented obstacles to the seamless exchange of patient identities among various healthcare institutions, often resulting in data isolation within individual silos. Of paramount concern are the potential privacy breaches associated with centralized IDMs, which may compromise patient confidentiality. In response to these challenges, we propose a novel approach to securely sharing patient details across multiple hospitals utilizing the zero-knowledge access protocol (MediCrypt-ZKAP) within the Hyperledger Fabric blockchain framework. By adopting MediCrypt-ZKAP, hospitals can effectively verify the identities of requesting entities without disclosing sensitive patient information, thereby ensuring the highest levels of confidentiality and privacy protection. The proposed system represents a proactive step towards addressing the critical need for secure and interoperable patient data exchange within the healthcare sector. Through the integration of MediCrypt-ZKAP into existing blockchain infrastructure, our solution aims to enhance data security and privacy while promoting seamless collaboration among healthcare institutions
Cluster based water leakage detection frame work for the improvement of water management using WSN
Get PDFWater is a vital resource that is essential for human survival and economic development. In any case, water shortage and wastage have become significant difficulties that undermine economical turn of distribution network. A critical reason for water wastage is water leakage in the distribution system, which prompts an extensive loss of water assets and energy. Conventional manual techniques for identifying water leakage are tedious, work serious, and frequently ineffectual. Hence, there is a need for an automated system that can efficiently detect, control and monitor water leakage to improve water management. In this paper, cluster-based water leakage detection (CBWLAD) algorithm for the improvement of water management using wireless sensor network (WSN) is proposed. The design system contains sensor nodes which are conveyed all through the water distribution networks and associated with central control unit. The sensor nodes can distinguish changes in the water pressure and flow rate, which are indicative of water leakage. The real time monitoring feature also enables timely maintenance and repair of the network of water distribution prolongs the lifespan of infrastructure. Further, the research and development are required to optimize the system’s performance and adapt it to real-world scenarios
Improved Bi-GRU for parkinson’s disease severity analysis
Get PDFParkinson’s disease (PD) is a common neuro-degenerative issue, evaluated via the continuous deterioration of motor functions over time. This condition leads to a gradual decline in movement capabilities. For diagnosing clinical set of PDs, medical experts utilize medical observations. These observations are highly based on the expert’s experience and can vary among clinicians due to its subjective nature, leading to differences in evaluation. The gait patterns of individuals with PD typically exhibit distinctions from those of adults. Evaluating these gait malformations not only aids in diagnosing PD but can also enable the categorization of severity stages with respect to symptoms of motor movement. Therefore, this paper introduces a classification of gait model based on the optimized deep learning (DL) model bidirectional gated recurrent unit-artificial hummingbird optimizer (BI-GRU-AHO). The training and testing involved the sequential segmentation of the right and left instances from the signals of vertical ground reaction force (VGRF) based on the identified gait cycle. The outcomes of the proposed BI-GRU-AHO exhibits reliable and accurate assessment of PD and achieved better accuracy of 98.7 %. The proposed model is trained and tested satisfactorily; hence it can be implemented in a real-time environment by integrating the model into a software application or system capable of receiving real-time data from PD patients
Detect and envision of pandemic disease exposure using CNN
Get PDFCOVID-19 has emerged as a pandemic, affecting millions globally with its high transmission rate, especially in colder climates. The virus's multiple mutations have made it progressively harder to detect and manage. Despite widespread awareness of preventive measures such as masks and sanitizers, early detection remains critical. Traditional methods like blood tests are time-consuming, and existing studies utilizing fuzzy K-means clustering, principal component analysis (PCA), stochastic discriminant analysis (SDA), decision trees (DT), and support vector machines (SVM) have faced limitations, including small datasets, insufficient accuracy, inadequate medical data, weak methodologies, and failure to consider primary symptoms. This work proposes a deep learning (DL) convolutional neural network (CNN) architecture utilizing CT scan images of the lungs for the rapid and accurate identification of COVID-19 infections. The approach leverages the Visual Geometry Group 16 (VGG16) model to extract significant features, such as size and color differences, from computed tomography (CT) scan images, facilitating a swift and precise diagnosis. The VGG16 model, implemented using the Keras library on top of TensorFlow, processes the preprocessed images through neural network layers to classify the images as COVID-19 positive or negative. The proposed model demonstrates a high accuracy rate of 94.12%, indicating that this method is both efficient and reliable for detecting COVID-19, offering a significant improvement over conventional diagnostic techniques and existing studies
Implementation of augmented reality as a revolutionary approach in computer stores
Get PDFThe adoption of augmented reality (AR) spans various fields, from education to business. Currently, many businesses utilize AR to boost customer engagement and enhance product understanding. This research focuses on developing and examining an AR application on a Solo computer store’s website to improve customer engagement. Results indicate that AR significantly enhances the shopping experience, deepens product comprehension, and increases website interactions. Features like 3D product visualization and detailed information enable customers to make more informed purchasing decisions. A questionnaire with 25 respondents revealed a high acceptance rate of the AR application, averaging 92%. Additionally, AR was shown to increase customer engagement, potentially boosting sales by up to 35%, reducing operating costs by 20%, and enhancing productivity by 15%. The study also found differing preferences across age groups: older respondents (40-70 years) favored traditional website features without AR and were less comfortable with markerless technology, whereas younger consumers (18-39 years) were more attracted to AR-enhanced websites. These insights offer valuable guidance for the Solo computer store to craft innovative marketing strategies tailored to the diverse preferences and needs of their customers
Tree-based models and hyperparameter optimization for assessing employee performance
Get PDFThe Palembang city fire and rescue service (FRS) is encountering challenges in adhering to national standards for fire response time. Hence, the Palembang city FRS is committed to enhancing employee performance through quarterly performance assessments based on various criteria such as attendance, work targets, behavior, education, and performance reports. This study proposes tree-based models in machine learning (ML) and hyperparameter optimization to assess the performance of Palembang city FRS employees. Tree-based models encompass decision trees (DT), random forests (RF), and extreme gradient boosting (XGB). The predictive performance of each model was evaluated using the confusion matrix (CM), the area under the receiver operating characteristic (AUROC), and the kappa coefficient (KC). The results indicate that RF performs better than DT and XGB in the sensitivity, AUROC, and KC metrics by 1.0000, 0.9874, and 0.8584, respectively