Bulletin of Electrical Engineering and Informatics
Not a member yet
2885 research outputs found
Sort by
Explainable deep learning for diagnosing acute lymphocytic leukemia using blood smear images
Full text linkAcute lymphocytic leukemia (ALL) is a rapidly progressing blood cancer that affects the lymphocytes. The diagnosis of ALL typically entails the examination of blood smears under a microscope, processes that are both time-consuming and susceptible to errors. Deep learning (DL) approaches have shown significant promise in automating the classification of ALL from microscopic images. However, the lack of transparency in these models hinders their widespread adoption in clinical settings. This study addresses this challenge by employing fine-tuned EfficientNetV2B3, a DL model, in conjunction with local interpretable model-agnostic explanations (LIME), a technique for explainable artificial intelligence (XAI) technique, to classify microscopic images of ALL. The C-NMC 2019 dataset, which has been augmented to ensure class balance, was utilized for training and evaluation. The proposed approach achieved impressive results, with an average recall, F1-score, and accuracy of 0.9795 and precision of 0.9796. The use of LIME effectively highlights relevant areas for prediction, accurately corresponding to the cell characteristics. The integration of DL and XAI techniques enhances the interpretability of ALL classification models, potentially increasing their trustworthiness and adoption in clinical practice. This study aims to further the development of diagnostic tools that are both precise and transparent for ALL
Designing electric braking system for brushless direct current motor as an electric bicycle propulsion
Full text linkOne of the problems arising in the conversion of pedal-based bicycles into electric bicycles using brushless direct current (BLDC) motors is how to provide an electric braking system for the BLDC motor. Published research on braking systems for electric bicycles from a practical perspective was still limited. The objective of the article is to develop an electric braking system for BLDC motor as a propulsion of electric bicycle converted from pedal-based bicycle from the empirical point of view. The pedal-based bicycle was converted to the electric bicycle by fully replacing its rear wheel including its chain system with the BLDC motor. The braking system was developed by adding a DC motor as a load for BLDC in the braking mode. Test was conducted by turning on the acceleration handle and then the braking action was applied. The stoppage time was recorded from the start of braking action until the wheel was fully stopped. The test results showed that the addition of a direct current (DC) motor as a load can shorten stoppage time of the electric bicycle dramatically, i.e., needs 1.65 seconds compared to 10.14 seconds and 3.09 seconds for without braking and with braking but without DC motor as a load respectively
Accurate brain tumor classification with STN-NAM in ResNet50 using MRI
Full text linkBrain tumor is an abnormal cell growth that contains malignant and benign cells emerging from numerous cell types within brain. Magnetic resonance imaging (MRI) is utilized for brain tumor classification which provides high-resolution images. However, tumors exhibit different characteristics like shape, location, and size which make it challenging to accurately distinguish among different tumor types and accurately classify them. In this research, spatial transformer network and non-local attention mechanism (STN-NAM) is proposed in ResNet50 to accurately classify tumors. STN transforms spatial information while NAM identifies relationships among normal and lesion areas, which together accurately classify tumors. Initially, images are obtained from Figshare, Brats 2019, and Brats 2020 datasets. These images are pre-processed using a normalized median filter (NMF) to reduce salt and pepper noise. Then, normalization is performed to resize original image to a standard size which assists uniformity in image dimension. U-Net is employed to segment tumor regions and STN-NAM is performed to accurately classify tumors. In comparison to the existing techniques namely, multi-level attention network (MANet), mathematical model with 3D attention U-Net, and convolutional neural network (CNN), the STN-NAM achieves superior accuracy of 98.06%, 99.05%, and 98.66% in Figshare, Brats 2019, and Brats 2020 datasets, respectively
Wideband and high gain mmWave antenna with phase gradient metasurface
Full text linkThis study presents a phase gradient metasurface (PGM) measuring 30 mm on each side and discusses its development. The use of multiple unit cell sizes is critical to the design of the PGM that was produced. A sand-timer-shaped monopole antenna was designed specifically for wideband millimetre wave (mmWave) applications. An antenna structure is reinforced with a PGM that was specifically designed to increase antenna gain over its bandwidth. This is done to increase the antenna gain overall. The antenna has a bandwidth of 9.18 GHz, which includes mmWave frequencies ranging from 20.78 GHz to 29.96 GHz. For a wideband response, the ground plane must contain flaws. These flaws must exist on the surface of the ground plane. This study presents an in-depth examination of the antenna, PGM design, and operating principles, backed up by experimental verification. The use of PGM results in a 5 dB increase in antenna gain, with an average improvement of 2 dBi across all frequencies
Optimized convolutional neural network deep learning for Arabian handwritten text recognition
Full text linkIn general, the term handwritten character recognition (HCR) refers to the process of recognizing handwritten characters in any form, whereas handwritten text recognition (HTR) refers to the process of reading scanned document images that include text lines and converting those text lines into editable text. The identification of recurring structures and configurations in data is the primary focus of the field of machine learning known as pattern recognition. Optical character recognition, often known as OCR, is a challenging issue to solve when it comes to the field of pattern recognition. This article presents machine learning enabled framework for accurate identification of Arabian handwriting. This framework has provisions for image processing, image segmentation, feature extraction and classification of handwritten images. Images are enhanced using contrast limited adaptive histogram equalization (CLAHE) algorithm. Image segmentation is performed by k-means algorithm. Classification is performed using convolutional neural network (CNN) VGG 16 and support vector machine (SVM) algorithm. Classification accuracy of CNN VGG 16 is 99.33%
A multicriteria comparison of end-to-end and cascade speech-to-text translation models
Full text linkThis paper presents a thorough examination of two prominent speech-to-text translation (STT) models: the end-to-end (E2E) model and the cascade model. STT is a critical technology in today’s multilingual society, facilitating communication across language barriers. The study focuses on comparing these models using a multicriteria approach to evaluate their effectiveness in translating speech to text. The E2E model represents a unified architecture that directly translates speech into text, while the cascade model involves separate modules for speech recognition and machine translation (MT). Both models have distinct advantages and challenges, which are explored in detail. Through a multicriteria comparison, this research assesses various performance metrics and criteria to determine the strengths and weaknesses of each model. The weighted sum method is employed to assign weights to evaluation criteria, providing a systematic evaluation framework. The findings have implications for researchers and developers in STT. By understanding the comparative performance of E2E and cascade models, researchers can make informed decisions regarding model selection based on criteria such as accuracy, speed, robustness, and resource requirements. This research advances the understanding of speech translation technologies and provides a foundation for future studies to refine evaluation methodologies, explore hybrid models, and enhance translation quality
Wireless charging and monitoring system utilizing internet of things technology for electric vehicle application
Full text linkInternal combustion engine (ICE) vehicles are major contributors to climate change and pollution, driving the transition to electric vehicles (EVs) as a cleaner alternative. However, EVs encounter challenges with charging infrastructure, notably the need for physical cables and issues with alignment for efficient charging. To address these problems, a wireless EV charging system has been developed using internet of things (IoT) technology for real-time monitoring and control. This system incorporates ESP32 and ESP8266 microcontrollers, infrared sensors, inductive coils, an OLED display, an ESP32-CAM module, relay modules, an AC to DC converter, a TP4056 charging module, a DC voltage sensor, and lithium-ion batteries. It employs a 20-turn coil for inductive coupled wireless power transfer (WPT), enabling the full charging of two lithium-ion batteries within 60 minutes. The system can detect an EV’s presence, display battery status on an OLED screen, and provide real-time images of the vehicle’s position through the SWEVCS mobile app. Infrared sensors ensure proper and precise alignment for effective charging. This advanced wireless charging solution enhances EV charging efficiency and convenience while supporting a more sustainable energy approach
Elliptic curve cryptography based light weight technique for information security
Full text linkRecent breakthroughs in cryptographic technology are being thoroughly scrutinized due to their emphasis on innovative approaches to design, implementation, and attacks. Lightweight cryptography (LWC) is a technological advancement that utilizes a cryptographic algorithm capable of being adjusted to function effectively in various constrained environments. This study provides an in-depth analysis of elliptic curve cryptography (ECC), which is a type of asymmetric cryptographic method known as LWC. This cryptographic approach operates over elliptic curves and has two applications: key exchange and digital signature authentication. Next, we will implement asymmetric cryptographic algorithms and evaluate their efficiency. Elliptic curve elgamal algorithms are implemented for encryption and decryption of data. Elliptic curve Diffie-Hellman key exchange is used for sharing keys. Experimental results have shown that ECC needs small size keys to provide similar security. ECC takes less time in key generation, encryption and decryption of plain text. Time taken by ECC to generate a 2,048 bit long key is 1,653 milliseconds in comparison to 4,258 millisecond taken by Rivest-Shamir-Adleman (RSA) technique
Improved half-maximal inhibitory concentration regression model using amyotrophic lateral sclerosis data
Full text linkThe current research addresses the critical need for precise half-maximal inhibitory concentration regression in the neurodegenerative condition amyotrophic lateral sclerosis (ALS). Unavailable drug-induced gene expressions and irrelevant molecular descriptors have yielded regression models with less accuracy using traditional machine learning (ML). Drugs can be converted to graph format and integrated with gene expressions to learn drug-gene interactions better thereby producing precise half-maximal inhibitory concentration regression models. To accomplish this, three variants of graph neural networks (GNN) namely graph attention networks (GAT), message passing neural networks, and graph isomorphism networks are utilized in the proposed work. The gene expression profiles of ALS drugrelated genes were retrieved from the DepMap PRISM drug repurposing hub, and the drug graphs with their accompanying half-maximal inhibitory concentration values were obtained from the ChEMBL databases. The graph is constructed for ninety approved drugs connected to 32 key protein targets of ALS and its related conditions. The half-maximal inhibitory concentration regression model trained with optimized hyperparameters in GAT performs well with an R2 score of 0.92, a mean absolute error (MAE) of 0.20, and a root mean square error (RMSE) of 0.17. This model produced better results than other ML and deep learning models
Task scheduling algorithm using grey wolf optimization technique in cloud computing environment
Full text linkScheduling refers to the process of allocating cloud resources to several users according to a schedule that has been established in advance. It is not possible to get acceptable performance in settings that are distributed without proper planning for simultaneous processes. When developing productive schedules in the cloud, it is necessary for work scheduling to take a variety of constraints and goals into consideration.When dealing with activities that have performance optimization limits, resource allocation is a very important aspect to consider. When it comes to cloud computing, the only way to achieve great performance, high profits, high scalability, efficient provisioning, and cost savings is with an exceptional task scheduling system. This article presents a grey wolf optimization (GWO) based framework for efficient task scheduling in cloud computing environment. The proposed algorithm is compared with particle swarm optimization (PSO) and flower pollination algorithm (FPA) and GWO is performing task scheduling in less execution time and cost in comparison with PSO and FPA techniques. Execution time taken by GWO to finish 200 task in 120.2 ms. It is less than the time taken by PSO and FPA algorithm to finish same number of tasks