Indonesian Journal of Electrical Engineering and Computer Science
Not a member yet
9109 research outputs found
Sort by
A novel approach for generating physiological interpretations through machine learning
Get PDFPredicting blood glucose trends and implementing suitable interventions are crucial for managing diabetes. Modern sensor technologies enable the collection of continuous glucose monitoring (CGM) data along with diet and activity records. However, machine learning (ML) techniques are often used for glucose level predictions without explicit physiological interpretation. This study introduces a method to extract physiological insights from ML-based glucose forecasts using constrained programming. A feed-forward neural network (FFNN) is trained for glucose prediction using CGM data, diet, and activity logs. Additionally, a physiological model of glucose dynamics is optimized in tandem with FFNN forecasts using sequential quadratic programming and individualized constraints. Comparisons between the constrained response and ML predictions show higher root mean square error (RMSE) in certain intervals for the constrained approach. Nevertheless, Clarke error grid (CEG) analysis indicates acceptable accuracy for the constrained method. This combined approach merges the generalization capabilities of ML with physiological insights through constrained optimization
Deep learning approaches, platforms, datasets for behaviorbased recognition: a survey
Get PDFVideo surveillance is an extensively used tool due to the high rate of atypical behavior and many cameras that enable video capture and storage. Unfortunately, most of these cameras are operator dependent for stored content analysis. This limitation necessitates the provision of an automatic behavior identification system. This behavior identification can be achieved using unsupervised (generative) computer vision methods. Deep learning makes it possible to model human behavior regardless of where they could be. We attempt to classify current research work to report the ongoing trends in human behavior recognition using deep learning algorithms. This paper reviews various aspects, like the ones associated with machine learning and deep learning models, human activity recognition (HAR), deep learning frameworks/tools, abnormal behavior datasets, and a variety of other current trends in the field of automatic learning. All these are to give the researcher a sense of direction in this area
Exploring the impact of artificial intelligence driven solutions on early detection of cardiac arrest
Get PDFThe advancement of medical science and technology has yet not evolved up with a concrete solution towards early detection of cardiac arrest from practical deployment. It is noted that artificial intelligence (AI) has been proving a potential contributor to address this state of diagnosis emergency. In current era of research work, there has been various implementation model and review work has been carried out towards advocating AI for determining early onset of cardiac arrest; however, there are various contradiction and shortcoming which is quite challenging to be extracted. Hence, the current manuscript presents a review of existing methodology by presenting core taxonomies of recent AI-methods towards early detection of cardiac arrest. Various standard dataset has been studied too to find associated advantages and limitation that restrict the actual potential of AI to prediction. The outcome presents novel highlights of research gap, trade-off, and crisp highlights of effectiveness of existing AI approaches as a study contribution
Classification model for infectious lung diseases using convolutional neural networks on web and mobile applications
Get PDFAccurate lung disease diagnosis in infected patients is critical for effective treatment. Tuberculosis, COVID-19, pneumonia, and lung opacity are infectious lung diseases with visually similar chest X-ray presentations. Human expertise can be susceptible to errors due to fatigue or emotional factors. This research proposes a real-time deep learning-based classification system for lung diseases. Three models of convolutional neural networks (CNNs) were deployed to classify lung illnesses from chest X-ray images: MobileNetV3, ResNet-50, and InceptionV3. To evaluate the effect of high interclass similarity, the models were evaluated in 3-class (Tuberculosis, COVID-19, normal), 4-class (lung opacity, tuberculosis, COVID-19, normal), and 5-class (tuberculosis, lung opacity, pneumonia, COVID-19, normal) modes. The best classification accuracy was attained by retraining MobileNetV3, which obtained 94% and 93.5% for 5-class and 4-class, respectively. InceptionV3 had the lowest accuracy (90%, 89%, 93% for 5-, 4-, and 3-class), while ResNet-50 performed best for the 3-class setting. These findings suggest MobileNetV3's potential for accurate lung disease diagnosis from chest X-rays despite the interclass similarity, supporting the adoption of computer-aided detection systems for lung disease classification
High-efficiency multimode charging interface for Li-Ion battery with renewable energy sources in 180 nm CMOS
Get PDFThe high-efficiency multi-source Lithium-Ion battery charger with multiple renewable energy sources described in the present paper is based on supply voltage management and a variable current source. The goal of charging the battery in a constant current (CC) mode and controlling the supply voltage of the charging circuit are both made achievable using a variable current source, which may improve the battery charger’s energy efficiency. The battery must be charged with a degraded current by switching from the CC state for the constant voltage (CV) state to prevent harming the Li-Ion battery. The Cadence Virtuoso simulator was utilized to obtain simulation results for the charging circuit, which is constructed in 0.18 μm CMOS technology. The simulation results obtained using the Cadence Virtuoso simulator, provide a holding current trickle charge (TC) of approximately 250 mA, a maximum charging current (LC) of approximately 1.3 A and a maximum battery voltage of 4.2 V, and takes only 29 minutes to charge
Spatial-temporal data imputation for predictive modeling in intelligent transportation systems
Get PDFData imputation is necessary to overcome data loss in intelligent transportation systems (ITS) due to the many sensors used to monitor traffic conditions. Sensor malfunction, hardware limitations, and technical glitches can lead to incomplete data, potentially leading to errors in traffic data analysis. This analysis investigated spatial-temporal data imputation approaches applied for predictive modeling in ITS. Each approach's strengths, weaknesses, and applicability in the context of ITS are evaluated. We analyzed various imputation approaches involving statistical, machine learning, and combined methods. Statistical methods are more straightforward but could effectively handle modern traffic's complexity. On the other hand, machine learning and combined approaches, such as hybrid convolutional neural network (CNN)- long short-term memory (LSTM), offer more robust capabilities in capturing non-linear patterns present in spatio-temporal data. This research aims to investigate the effectiveness of each approach in overcoming data incompleteness and the accuracy of predicting future traffic conditions with the widespread adoption of IoT, electric vehicles, and autonomous vehicles. The results of this investigation provide an understanding of the most suitable approaches to address the challenges of spatio-temporal data imputation and provide practical guidance for predictive modeling in ITS
Sentiment analysis resource of Libyan dialect for Libyan Airlines
Get PDFArabic lacks extensive corpora for natural language processing (NLP) when compared to other languages, namely in the Libyan dialect (LD). Therefore, this study proposes the first corpus of Arabic sentiment analysis (ASA) of the Libyan Dialect for the Airline Industry (ASALDA). It comprises 9,350 comments and tweets, annotating them manually depending on text polarity into three labels: positive, negative, and neutral, and utilized aspect-based sentiment analysis (SA) to annotate opinions regarding fifteen aspects. Also constructs a simple sentiment lexicon of the LD. The solution is based on the idea that the corpus and lexicon can be helpful models to improve classification for the LD. The approach has notable merits, namely creating a corpus and sentiment lexicon for the LD from comments and tweets of airline companies. A comprehensive verification using a statistical technique called the chi-square test is carried out with the corpus to determine if two aspects are related to one another. Based on the statistical work, we found that airlines should focus on improving their services in aspects where they are performing poorly, such as late flights, customer service, or price. The corpus and lexicon that we proposed can be utilized to perform many opinion mining and SA experimentations using machine learning and deep learning
Influences from SiO2 particles on optical properties of white diodes verified through computer simulation
Get PDFFor typical white-illumination diodes (WLEDs) based on phosphor conversion, heat consistency would be an essential element in phosphor samples, which is based on consistent discharge intensity, apex profile, as well as location as the samples function under different heat levels. With the goal of attaining desirable heat consistency, the study herein concerns the thermic mechanism in different phosphor samples singularly or dualincorporated with Ce3+ and Eu2+. Based on our acquired data, the luminescent features for the samples exhibit copious alterations when subject to different heat levels, primarily decided by phosphor bases’ crystalline formation. The assessment of the interaction among the thermic mechanism and base latticework in the samples suggest that a merger between firm crystalline formations and symmetrical locations would result in desirable thermic consistency in samples. As such, the study herein also assesses a number of formations possessing firm foundations as well as specific approaches for avoiding thermic irregularities in phosphor samples, aiming to identify reliable samples as well as approaches for augmenting heat consistency for said samples
Unveiling the influence of back-translation on sentiment analysis of Indonesian hotel reviews
Get PDFThis study aims to conduct sentiment analysis on hotel reviews in Indonesian using several machine learning classification algorithms, namely multinomial naive bayes (MNB), support vector machine (SVM), and random forest (RF). The back translation method is employed to generate synthetic data variations that are used as additional data variations in building classification models. This research tests three scenarios based on the datasets used: the original dataset, the dataset resulting from back translation, and the combined dataset of both. The experimental results show that the use of combined data yields better results, with the random forest algorithm standing out as the best performer. Back translation significantly improves model evaluation in sentiment analysis for several reasons, including enriching the dataset with new variations, enhancing model robustness, and increasing dataset complexity. However, the differences in the number of word features among scenarios indicate that back translation also significantly influences the dataset's characteristics
Enhancing energy efficiency and reliability in wireless sensor networks using BioGAT optimization
Get PDFThe BioGAT model, as proposed, presents a novel methodology for enhancing the efficiency of wireless sensor networks (WSNs), which are essential elements of contemporary communication and sensing systems. For real-time monitoring and data analysis, WSNs are comprised of autonomous sensor nodes that are outfitted with processing, wireless communication, and sensing capabilities. These nodes are deployed in a variety of environments. By means of an advanced optimization model, this work aims to address critical challenges in WSNs, specifically in the areas of node placement, energy efficiency, and network reliability. By utilizing biogeography-based optimization (BBO) and graph attention networks (GAT), the BioGAT model endeavors to dynamically adapt to network changes while achieving a balance between efficient coverage and energy consumption. Cluster heads (CHs), which are essential for the aggregation of data, have a significant impact on improvements in energy efficiency and the longevity of networks. By means of comprehensive simulations and evaluation, this study presents exceptional outcomes. The BioGAT model outperforms prior approaches by attaining a 95% packet delivery ratio and an enhanced throughput. In addition, the model effectively decreases mean energy consumption, underscoring its capacity to improve the sustainability and dependability of networks in a variety of WSN applications