Majed Alsafyani

Majed Alsafyani

Auto-generated profile, based on publicly available information

Majed Alsafyani

Department of Computer Science, College of Computers and Information Technology, Taif University, Taif, Saudi Arabia

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
15 May 2024 in IEEE Access
Verified
Journal Article
Open access
Abdullah Ayub KhanAbdullah Ayub Khan
Abdullah Ayub Khan
Asif Ali LaghariAsif Ali Laghari
Asif Ali Laghari
Roobaea AlroobaeaRoobaea Alroobaea
Roobaea Alroobaea
ABAbdullah M. Baqasah
Abdullah M. Baqasah
MAMajed Alsafyani
Majed Alsafyani
JAJamil Alsayaydeh
Jamil Alsayaydeh
Particularly in the context of smart cities, remote sensing data (RSD) has emerged as one of the hottest study topics in information and communication technology (ICT) today. The development of machine learning (ML) and artificial intelligence (AI) has made it possible to solve a number of issues, including automation, control access, optimization, monitoring, and management. Simultaneously, there are significant issues with the design and development of the process hierarchy, including inadequate training records, centralized architecture, data privacy protection, and overall resource consumption restrictions. The development of Distributed Ledger Technology (DLT), on the other hand, provides a decentralized infrastructure that allows systems to eliminate centralized data-sharing procedures of smart cities while transferring from network node to network node, and third-party access control solves machine learning issues. To process an ideal data delivery mechanism for the smart cities analytical model, the paper employs Partial Swam Optimization (POS) in conjunction with a secure blockchain distributed consortium network. This work makes three contributions. Firstly, it offers a safe transmission method that combines blockchain and machine learning to optimize the path for reliable data delivery across secure channels. Second, neighborhood encryption sequences are carried out using NuCypher proxy re-encryption-enabled value encryption, a public key cryptographic approach that avoids cypher conversion. Third, Artificial Neural Networks (ANNs) can solve the data deliverance classification problem in smart cities by optimizing record management and preservation.

12 September 2023 in IEEE Access
Verified
Journal Article
Open access
OSOumaima Saidani
Oumaima Saidani
MAMajed Alsafyani
Majed Alsafyani
Roobaea AlroobaeaRoobaea Alroobaea
Roobaea Alroobaea
NANazik Alturki
Nazik Alturki
Rashid JahangirRashid Jahangir
Rashid Jahangir
LMLeila Jamel Menzli
Leila Jamel Menzli
Human activity recognition is a challenging and active research topic in computer science due to its applications in video surveillance, health monitoring, rehabilitation, human-robot interaction, robotics, gesture and posture analysis, and sports. In the past, various studies have utilized manual features to identify human activities and obtained good accuracy. Nonetheless, the performance of such features degraded in complex situations. Therefore, recent research used deep learning (DL) techniques to capture the local features automatically from given activity instances. Though automatic feature extraction overcomes the problems of manual features, there is still a need to enhance the efficiency and accuracy of existing techniques. The motivation behind this research is to improve the efficiency and accuracy of HAR systems. This research proposed a HAR system, which applies data enhancement techniques before capturing robust and discriminative features set from each activity instance. The captured feature set is given to the transformer model for activities recognition using the PAMAP2, UCI HAR, and WISDM datasets. The achieved results revealed that the proposed HAR model outperformed the baseline methods. Specifically, the proposed HAR achieved 98.2% accuracy for PAMAP2 with all instances in 12 activities, 98.6% accuracy for UCI HAR with all instances in 6 activities, 97.3% for WISDM with all instances in 6 activities. The advantage of the proposed hybrid features is the capability to capture both low-level and high-level information from the sensor data, potentially enhancing the discriminative power of the system. In addition, this study employed a transformer a model due to its ability to capture long-range dependencies, which are beneficial in recognizing complex human activities patterns.

8 August 2024 in IEEE Access
Verified
Journal Article
Open access
GGGautam Gupta
Gautam Gupta
RSR. Sudeep
R. Sudeep
Bragadeshwaran AshokBragadeshwaran Ashok
Bragadeshwaran Ashok
RVR. Vignesh
R. Vignesh
Kannan ChidambaramKannan Chidambaram
Kannan Chidambaram
AEAhmed Mohamed Emara
Ahmed Mohamed Emara
Regenerative braking is of paramount importance both in terms of energy recovery and from the perspective of hybrid electric vehicles (HEVs). It allows HEVs to recover energy during braking which in turn enhances the range, efficiency and sustainability by reducing electricity demand and environmental impacts. Upon considering the demand for high-performance regenerative braking in HEVs, the development and optimization of advanced control strategies become mandatory. The study presents a model-based regenerative braking system that integrates tire-road coefficients with an intelligent control algorithm to enhance the efficiency and sustainability of HEVs. The literature review provides insights into existing hybrid power systems, tire-road friction estimation, braking force distribution, and regenerative braking optimization strategies. Different control strategies including lookup table, fuzzy logic control (FLC) and adaptive neuro-fuzzy inference system (ANFIS) are analyzed for their impact on the efficiency of regenerative braking systems under various driving conditions and their results have been presented. The study reveals the supremacy of the ANFIS control algorithm across multiple driving cycles, demonstrating its potential to enhance fuel efficiency and state of charge (SOC). The study involves several input and output parameters of a vehicle to arrive at the desired outcome, and this also showcases ANFIS’s capabilities in providing a better and smoother driving behaviour for the driver, indicating its adaptability and proficiency in energy recovery. FLC strategy excelled in μ- estimation with lower estimation errors of about 0.15%, 0.16% and 0.13% across various driving cycles when compared to the look-up table approach. ANFIS exhibited fuel economy, improvements of about 0.282%, 0.437%, and 0.345% over the look-up table approach in the same cycles. In terms of SOC optimization, ANFIS achieved the highest final SOC values, surpassing the look-up table strategy by 2.082%, 6.799%, and 0.702% in the same cycles.

24 November 2025 in IEEE Access
Verified
Journal Article
Open access
NDN. Deluxni
N. Deluxni
PSPradeep Sudhakaran
Pradeep Sudhakaran
MAMajed Alsafyani
Majed Alsafyani
AYAmr Yousef
Amr Yousef
Accurate object detection and instance segmentation of underwater trash is funda-mental for autonomous underwater vehicles (AUV) navigation, marine ecosystem preservation, and environmental monitoring. However, the real time implementation of the deep learning models and effective classification of the objects are the challenging tasks. This paper presents a deep learning framework for underwater debris segmentation using YOLOv12s architecture in combination with Recursive Efficient Layer Aggregation Networks (R-ELAN) and FlashAttention to increase feature extraction and computational efficiency. The proposed model leverages R-ELAN to maximize multiscale feature representation, hence enhancing hierarchical information flow. Flash Attention lowers latency without compromising accuracy by accelerating the attention mechanism. On a fresh dataset of underwater garbage including both synthetic and real-world images, the algorithm is trained and tested. Experimental data demonstrate better segmentation accuracy, mean Intersection over Union (mIoU), and inference speed when compared to state-of- the-art segmentation models. The proposed approach offers a practical and real-time means of tracking seabed debris, therefore promoting advancements in environmental monitoring and self-sufficient marine uses.

1 January 2026 in Complexity
Verified
Journal Article
Open access
ASAnto Lourdu Xavier Raj Arockia Selvarathinam
Anto Lourdu Xavier Raj Arockia Selvarathinam
SSSarita Simaiya
Sarita Simaiya
ABAbdullah M. Baqasah
Abdullah M. Baqasah
Roobaea AlroobaeaRoobaea Alroobaea
Roobaea Alroobaea
MAMajed Alsafyani
Majed Alsafyani
TTT. Hiyab Tekale
T. Hiyab Tekale
Cardiovascular diseases remain a major cause of morbidity and mortality worldwide, creating a need for accurate and interpretable automated electrocardiogram (ECG) analysis. Although deep learning has demonstrated strong performance in ECG interpretation, many existing approaches primarily rely on data‐driven representations and provide limited integration of structured clinical knowledge and model‐level interpretability. This study presents MedXpert‐AI, a context‐aware and explainable hybrid framework for 27‐label multilabel ECG classification. The framework integrates CNN‐, BiLSTM‐, and transformer‐based multilevel feature learning with clinical knowledge representation comprising a medical knowledge graph and rule‐based clinical reasoning. A graph neural network (GNN) is employed to learn knowledge‐guided representations, while adaptive attention feature fusion (AAFF) adaptively integrates ECG‐derived and clinical representations for final diagnostic prediction. SHAP, Grad‐CAM, and transformer attention heatmaps are incorporated to provide complementary model‐level explanations. The framework was evaluated using 43,101 12‐lead ECG recordings from the PhysioNet/Computing in Cardiology (CinC) 2020 dataset, comprising 27 scored diagnostic labels. For the primary multilabel evaluation, MedXpert‐AI achieved 98.85% AUROC, 97.64% AUPRC, 97.20% Macro F1, 97.46% Micro F1, and 0.94 MCC. Controlled ablation experiments showed performance reductions following removal of the GNN, transformer, and clinical knowledge representation modules, as well as replacement of AAFF with simple feature concatenation. Computational analysis further demonstrated a relatively low resource requirement under the evaluated experimental conditions, with approximately 2.3 billion FLOPs and 1.5 GB GPU memory usage. The explainability analysis provided complementary visualizations of feature attribution, activation localization, and temporal attention, although formal clinician‐based validation of explanation quality was not performed. These findings support MedXpert‐AI as a knowledge‐aware and explainable framework for automated multilabel ECG analysis, while external multicenter and prospective clinical validation remain necessary to establish generalizability and clinical utility.

29 November 2024 in IEEE Access
Verified
Journal Article
Open access
JRJ. Rahulkumar
J. Rahulkumar
NRNarayanamoorthi Rajamanickam
Narayanamoorthi Rajamanickam
MAMajed Alsafyani
Majed Alsafyani
Waleed Mohammed AbdelfattahWaleed Mohammed Abdelfattah
Waleed Mohammed Abdelfattah
This research article proposes Model-Free Predictive Control and a Modified Multi-Phase shift Pulse Width Modulation technique for a semi-dynamic Wireless Resonant Inductive Power Transfer for an Electric Vehicle charging system. The semi-dynamic wireless Electric Vehicle charging system recently gained considerable attention in facilitating the user-friendly Electric Vehicle charging infrastructure. The airgap coupling factor and load resistance are significant parameters of the WRIPT system and highly influence power transfer efficiency. By operating the system in constant voltage mode, and realizing the high-power factor, the life span of the equipment is enhanced. The Model-Free Predictive Control scheme addresses the dynamic variations in the system power transfer, by automatic resonant frequency optimization control technique. The input impedance (imaginary part) is considered to introduce three mega parameters into the control functions, it makes the system control independent. The objective is to operate the receiver coil of the EV online at a resonant frequency with the transmitter coil, even when the receiver coil is in a semi-dynamic moving state. Especially, when the receiver coil shifts to align between two transmitter coils on the e-lane, to achieve continuous charging. The group-based frequency control technique is used in the proposed Model-Free Predictive Control, it reduces the dynamic computation burden, enhances the system accuracy, and response speed, and ensures safe operation. The proposed control technique is experimented with to investigate its effectiveness.