Bulletin of Electrical Engineering and Informatics
Not a member yet
2885 research outputs found
Sort by
Multimodal speech emotion recognition optimization using genetic algorithm
Get PDFSpeech emotion recognition (SER) is a technology that can detect emotions in speech. Various methods have been used in developing SER, such as convolutional neural networks (CNNs), long short-term memory (LSTM), and multilayer perceptron. However, sometimes in addition to model selection, other techniques are still needed to improve SER performance, namely optimization methods. This paper compares manual hyperparameter tuning using grid search (GS) and hyperparameter tuning using genetic algorithm (GA) on the LSTM model to prove the performance increase in the multimodal SER model after optimization. The accuracy, precision, recall, and F1 score improvement obtained by hyperparameter tuning using GA (HTGA) is 2.83%, 0.02, 0.05, and 0.04, respectively. Thus, HTGA obtains better results than the baseline hyperparameter tuning method using a GS
MyPharmaceutical: an interactive proof of concept
Get PDFWith the rise of health awareness, pharmaceutical and cosmetic products should be verified to protect ourselves from health risks. MyPharmaceutical is a proof-of-concept proposed to provide a mobile application for users to carry out product verification and reporting and a web application for administrative purposes. The data on the registered pharmaceutical and cosmetic products were extracted from national pharmaceutical regulatory agency (NPRA) website. MyPharmaceutical mobile application provides functionalities such as searching the registered product, bookmarking products, reporting products, and tracking report status. The mobile application also implemented a barcode scanner feature to provide ease of product verification. A named entity recognition algorithm is applied with the NLP.js library to provide an improved product search feature for the users, where products can be searched with multiple search criteria in a single input. The web application is proposed to support the mobile application, where the NPRA data admins and officers can manage reported products, publish announcements, verify product data, and utilize the analytic dashboard. The system proposed is expected to provide ease of product verification and reporting to assist the public in choosing safe registered products and a platform for NPRA to manage data and deliver information to the users
Barium titanate–silicon elastomer based body coupled antenna for wearable microwave head imaging applications
Get PDFThis paper presents a flexible monopole antenna fed by a coplanar waveguide (CPW) feeding line with a barium titanate (BaTiO3) silicon-elastomer impedance matching layer for microwave head imaging applications. The operating frequency bandwidth of the proposed antenna is 614 MHz which is from 0.475 GHz to 1.089 GHz. In biomedical microwave sensing and imaging applications, the major challenge is the high power loss due to reflection between the BaTiO3 and the antenna due to impedance mismatch. Therefore, the proposed BaTiO3 silicon-elastomer composite is designed to have dielectric property of 20 which acts as an impedance matching layer for the monopole antenna. The proposed antenna has dimensions of 70×30×6 mm. The flexibility of the antenna is provided by the use of the silicon elastomer. It has been shown that the power radiated into an artificial head phantom improved by almost 160% as compared to antenna without impedance matching layer. Moreover, the SAR level is 0.0286 W/kg when 1 mW of power is transmitted, which is well below the limit set by the regulation. This makes the antenna suitable for wearable biomedical applications due to its wideband characteristic and improved power penetration into human head
Early prediction of COVID-19 infection using data mining and multi machine learning algorithms
Get PDFThe fields of artificial intelligence (AI) and machine learning (ML) have attracted significant interest and investment from a diverse range of industries, especially during the last several years. Despite the fact that AI methods have been used extensively and put through extensive testing in the healthcare industry, the recently discovered coronavirus disease (COVID-19) necessitates the use of these methods in order to prevent the emergence of the disease. The proposed system is based on six ML algorithms to predict COVID-19 infection as random forest (RF) algorithm, naive bayes (NB) algorithm, support vector machine (SVM) algorithm, decision tree (DT) algorithm, multi-layer perceptron (MLP), and k-nearest neighbor (KNN). It is based on two steps: first, we uploaded the dataset to train the model. Then, we test our model on those cases to work directly after making a trained classifier so it can directly discover with automatic COVID-19 prediction state of a patient suspected or not. The proposed system results showed the high accuracy of NB, DT, and SVM as 98.646%. Besides the better time to build the model and early predict the state of patients is 31 ms of the NB algorithm
Enhance the accuracy of malicious uniform resource locator detection based on effective machine learning approach
Get PDFPhishing attacks are increasing with the rise in web users. Addressing them requires understanding the techniques and employing effective response strategies. Phishing websites mimic authentic ones to deceive users into divulging personal information like bank account details, national insurance numbers, and passwords. Therefore, victims face financial loss from breached information security, constituting high-level internet fraud. Detecting phishing websites necessitates an intelligent model capable of recognizing suspicious features. To that purpose, this paper examines three classification methods for detecting phishing website attacks. This analysis allows to reconsider our awareness of phishing attacks and prevent the damage caused by phishing attempts in advance. Phishing website detection algorithm using three classification algorithms is proposed in this paper. It achieves high phishing website detecting accuracy, because three classification algorithms random forest (RF), support vector machine (SVM), and Bagging are combined in one system. The result of this research is found accuracy on validation set is 92.33%, the precision on validation set is 92.13%, the recall is 92.09% and F1 score is 92.10%. That prove that the result obtained in this research is more accurate than all the results of all the algorithms were applied in the same dataset that was train the proposed algorithm on it
Classification of human grasp forces in activities of daily living using a deep neural network
Get PDFThe study of human grasp forces is fundamental for the development of rehabilitation programs and the design of prosthetic hands in order to restore hand function. The purpose of this work was to classify multiple grasp types used in activities of daily living (ADLs) based on finger force data. For this purpose, we developed a deep neural network (DNN) model using finger forces obtained during the performance of six tests through a novelty force sensing resistor (FSR) glove system. A study was carried out with 25 healthy subjects (mean age: 35.4±11.6) all right handed. The DNN classifier showed high overall performance, obtaining an accuracy of 93.19%, a precision of 93.33%, and a F1-score of 91.23%. Therefore, the DNN classifier in combination with the FSR glove system is an important tool for physiotherapists and health professionals to determine and identify finger grasp forces patterns. The DNN model will facilitate the development of tailored and personalized rehabilitation programs for subjects recovering of hand injurie and other hand diseases. In future work, prosthetic hand devices can be optimized to more accurately reproduce natural grasping patterns
System dynamics modeling for predicting the impact of tutoring on student retention in the school of engineering
Get PDFStudent retention is a persistent problem in many educational institutions, and we seek to address this issue through the implementation of tutoring programs. To achieve this objective, system dynamics (SD) modeling is proposed as a method. This analytical tool allows simulating and predicting the behavior of a complex system over time, considering the interactions between its components. The main objective of this research is to perform SD modeling to improve student retention through tutoring. It seeks to design more effective and personalized tutoring programs, adapted to the specific needs and challenges of the institution's students. The results obtained show that, in the period between 2022 and 2026, research degrees will be encouraged, reaching 50% participation. This increase is considered a positive indicator that encourages universities to become research protagonists. In conclusion, SD modeling makes it possible to forecast and strategically plan the expected results in terms of student retention. This method provides tools to more effectively address the problem of retention, ensuring the academic success of students and promoting the participation of universities in research
A cost-effective ECG monitoring in rural areas: leveraging artificial neural networks for efficient healthcare solutions
Get PDFCardiovascular diseases engender serious public health concerns in developing nations since access to specialized medical equipment is often limited and standard treatment expenses can be prohibitive. This study proposes an efficient and relatively affordable electrocardiogram (ECG) monitoring system that reads and analyzes a person's electrocardiogram data to provide affordable and quality healthcare solutions. The device initially extracts features from electrocardiogram records by reading electrical signals in the heart. Extracted data are then analyzed by a trained deep learning model to determine precisely if the heart is in a healthy state or undergoing complexities. Experimental results showed that the fine-tuned ANN architecture outperformed the state-of-the-art architectures in this field with an accuracy of 98.95%. The data can also be sent to specialists through an MQTT server if necessary, allowing for remote diagnosis and treatment. The system is intended to be deployed in countries where rural regions lack access to specialized healthcare equipment and professionals. Additionally, the device is inexpensive and, hence can be made accessible to people with limited affordability
Effects of DBDS and DBPC antioxidants on the corrosion of copper strips immersed in transformer oil
Get PDFThis paper presents the experimental findings on the effects of antioxidants on the corrosion of copper strips immersed in mineral oil (MO)–antioxidant samples. First, the uninhibited MO was mixed thoroughly with dibenzyl disulphide (DBDS) or 2,6-di-tert-butyl-p-cresol (DBPC) at different concentrations (5, 15, 25, 50, and 200 mg/kg) using a hot plate magnetic stirrer set at a temperature of 73 °C and stirring speed of 750 rpm for 15 min. Following this, the MO–antioxidant samples were poured into separate test vessels and copper strips were added into the vessels. Next, each MO–antioxidant sample was blanketed with nitrogen gas, sealed, and placed in a forced convection laboratory oven. The MO–antioxidant samples were then thermally aged at 150 °C for 48 h. The results showed that the acidity of the MO–antioxidant sample decreased with an increase in the antioxidant concentration, regardless whether the antioxidant was DBDS or DBPC. However, the corrosion of the copper strip worsened with an increase in the antioxidant concentration, where DBDS had a higher relative degree of corrositivity to copper compared with DBPC. In addition, the results showed that a DBPC concentration of 25 mg/kg reduced the the acidity of the MO–antioxidant sample and resulted in a moderate tarnish of the copper strip
Improving frequency regulation for future low inertia power grids: a review
Get PDFThe modern power system is witnessing an unprecedented increase in the penetration of renewable variable generation (VG) sources. Increased uptake of converter interfaced VG like solar PV and wind power while replacing conventional synchronous generators (SGs) introduces new challenges to grid operators in terms of dynamically handling frequency stability and regulation. Reducing the number of SGs while increasing non-synchronous, inertia-less converter interfaced VG reduces grid natural inertia, which is critical for maintaining frequency stability. To cure inertia deficiency, researchers, broadly, have proposed implementing supplemental control strategies to VG sources or energy storage systems to emulate natural inertia (virtual inertia (VI)). Alternatively, VG sources can be operated below their maximum power point (deloaded mode), making available a reserve margin which can rapidly be deployed in case of a contingency with the help of power electronic devices, to provide fast frequency response. This paper reviews recent solutions proposed in literature to address the low inertia problem to improve frequency stability. Additionally, it highlights the formulation of an optimization problem for VI sizing and placement as well as techniques applied in solving the optimization problem. Finally, gaps in literature that require further research were identifie