United Arab Emirates University
United Arab Emirates University: Scholarworks@UAEU / جامعة الاماراتNot a member yet
5830 research outputs found
Sort by
أحكام المسؤولية الجنائية وأجهزة الذكاء الاصطناعي
No full textThe Criminal Liability Provisions of Artificial Intelligence (AI) Actions
This study aims to define the concept and levels of artificial intelligence (AI) to understand its legal nature, personality, and impact on criminal liability. It analyzes the current legal frameworks and examines existing legislation related to AI, assessing their ability to address issues concerning criminal responsibility for the actions of these systems. The study begins by defining the precise concept of criminal liability to understand how it can be applied in a technological environment. It then defines AI and explores its legal nature, to analyze current frameworks and how effectively they deal with questions of criminal accountability for AI’s actions. The study also discusses the adequacy of traditional concepts of criminal liability in addressing crimes that may be committed by AI systems .
The research concluded with several key findings, most notably that the mental element (mensrea) of traditional crimes does not align with the nature of autonomous AI systems capable of self-learning and independent decision-making. It also found that distributing responsibility among different parties (developers, manufacturers, owners, and users) is challenging, particularly when fault is unclear or complex. Accordingly, the study recommends that legislators establish a specific legal classification for AI systems that considers their dual nature—both tangible and non-human software-based—so that they are neither treated as mere “objects” nor granted full legal personality. Instead, a special legal status should be created to reflect their technical characteristics and define the scope of liability arising from their use. Additionally, new national legislation should be enacted to regulate the criminal responsibility of AI systems, particularly self-driving vehicles, including clear definitions of such systems and a classification of possible types of liability
باستخدام مجموعة بيانات متعددة من الطائرات بدون طيار YOLO البحث والإنقاذ البحري القائم على في ظروف الطقس الصعبة
No full textObject detection models, powered by deep learning and computer vision, are revolutionizing marine search and rescue (SAR). By analyzing aerial imagery and live drone footage, these systems automatically identify critical targets like survivors, life rafts, and debris across vast and treacherous ocean areas. This capability enhances operational efficiency by reducing human workload and accelerating response times, even in challenging conditions such as poor light, high seas, or cluttered backgrounds. The result is continuous monitoring, faster decision-making, and a significantly improved probability of successful rescue.
Departing from prior methodologies, YOLO introduced a paradigm shift through its single-shot architecture, which concurrently predicts bounding boxes and class probabilities in a single forward pass. This foundational design is responsible for YOLO\u27s defining trait: remarkable inference speed that made real-time object detection feasible. The ongoing evolution of the YOLO lineage reflects persistent architectural enhancements, yielding models that achieve a robust equilibrium between speed, precision, and size, cementing their effectiveness across a wide spectrum of tasks.
This study benchmarks the performance of YOLO models for maritime SAR under challenging weather conditions using the SeaDronesSee and AFO datasets. The results show that while YOLOv7 achieved the highest mAP@50, it struggled with detecting small objects. In contrast, YOLOv10 and YOLOv11 deliver faster inference speeds but compromise slightly on precision. This study also discusses enhancing the dataset using synthetic data to the existing dataset, which showed slightly higher experimental performance than the standalone dataset. The key challenges discussed include environmental variability, sensor limitations, and scarce annotated data, which can be addressed by such techniques as attention modules and multimodal data fusion. Overall, the research results provide practical guidance for deploying efficient deep learning models in marine SAR, emphasizing specialized datasets and lightweight architectures for edge devices
ADAPTATION OF EXOME SEQUENCING METHODS FOR LOW-QUALITY GENOMIC DNA EXTRACTED FROM AGED FTA MICRO-CARDS IN SUDDEN DEATH INVESTIGATIONS
No full textDNA degradaon and insufficient gDNA quanty and purity, preserving genomic DNA on FTA Micro-Cards presents substanal difficules for forensic invesgaons of unexpected fatalies. Through the use of lowquality genomic DNA recovered from four-year-old dried blood using FTA Micro-Cards, this effort seeks to enhance and opmize exome sequencing. Samples were sequenced using Next-Generaon Sequencing (NGS) with Illumina DNA Prep with Exome 2.0 Plus Enrichment with mitochondrial panel, starng at 0.1 ng/μl of gDNA.To improve the coverage of mitochondrial genes, a mitochondrial DNA spike-in was included, and the samples were sequenced using the NextSeq 1000/2000 plaorm, employing the onboard DRAGEN Enrichment Pipeline for analysis.
Unusual A260/A280 and A260/A230 purity raos were noted during the quality control (QC) of the extracted gDNA, indicang that the gDNA amounts were inadequate and failed to fulfill the specified standards for exome methods. Notwithstanding these drawbacks, the research employed 16 plex methodology to proficiently augment sequencing using modified protocols for library preparaon and enrichment. High coverage through enre exome sequencing at depths of 30X and above revealed that dependable genomic data may be acquired from low-quality inputs from samples with restricted recollectability.
These results highlight the ability to extract valuable genec informaon from contaminated materials, parcularly for forensic purposes and the detecon of genec variaons linked to unexpected mortality. The importance of older FTA samples is increased by this work, which also highlights the need for specialized opmizaon methods when handling biological specimens that have been degraded.
These results highlight the ability to discover genec variaons linked to sudden death and to retrieve crucial genec informaon from degraded materials, especially in forensic circumstances. This study enhances the suitability of legacy FTA Micro-Cards in complex forensic invesgaons and emphasizes the significance of customized opmizaon techniques for handling deteriorated biological specimens. This study involves the opmizaon of aged FTA cards in forensic challenging invesgaons, highlighng the need of customized improved techniques for the management of challenging biological specimens
TRUSTWORTHY FEDERATED LEARNING FRAMEWORK FOR SECURE, EFFICIENT, AND QUALITY-AWARE DISTRIBUTED AI
No full textFederated Learning (FL) emerged as a significant advancement in the field of Artificial Intelligence (AI), enabling collaborative model training across distributed devices while maintaining data privacy. As the importance of FL and its application in various areas increased, addressing trustworthiness issues in its various aspects became crucial. In FL process, not all client data may be relevant to the learning objective and incorporating updates from irrelevant data can harm the model\u27s performance. The selection of training samples significantly impacts model performance, as datasets with errors, skewed distributions, or low diversity can lead to inaccurate and unstable models. To address these issues, a data quality evaluation model has been introduced to assess the quality of datasets in FL systems. This model dynamically selects high-quality data samples for FL training by utilizing intrinsic and contextual data quality dimensions. Additionally, an importance-based interpretable feature selection model and a data quality-based dynamic client selection model employing Nash equilibrium and joint differential privacy (DP) have been designed. Building upon this, it has been identified that FL faces a major challenge of high communication overhead due to frequent model updates between clients and the central server. To address this, a novel lightweight Hierarchical FL (HFL) framework is proposed that integrates adaptive model pruning, quantization, and FL model communication and aggregation frequency optimization. First, a joint model pruning and quantization approach is introduced that dynamically adjusts pruning ratios and quantization levels. Second, a fairness-aware Stackelberg game-based model communication and aggregation frequency optimization mechanism has been developed, where clients, edge servers, and the central server collaboratively determine optimal update frequencies to balance overhead and convergence speed. Third, privacy protection is enhanced using Selective Homomorphic Encryption (SHE), and a verifiable model trust assessment is introduced to ensure secure participation of edge devices. However, communication efficiency and quality selection alone cannot guarantee trustworthy FL, as the security of both client models and aggregation processes is very important. FL enables distributed training while preserving data privacy, but it is vulnerable to poisoning attacks due to heterogeneous, non-IID client data and limited participation. To address this, a multi-layered defense is proposed that combines game-theoretic aggregation, incentive-aware client regularization, and model-side verification with important parameter selection and SHE. Extensive experiments have been conducted to validate the proposed approaches, demonstrating their effectiveness in improving data quality driven data sample and client selection, optimizing communication efficiency, and defending against adversarial poisoning in Trustworthy FL paradigm. We are hopeful that the results and discussions in this dissertation will help researchers to further improve trustworthy and secure FL systems
JET IMPINGEMENT AND VORTEX/SWIRL COOLING OF DIFFERENT INLET AND OUTLET GEOMETRICAL CONFIGURATIONS FOR TURBINE BLADE LEADING EDGE COOLING
No full textGas turbine blades operate in extreme environments, exposed directly to high-temperature combustion gases that cause severe thermal stresses, weaken material integrity, and may lead to structural failure. Proper cooling is crucial to lower blade temperatures, reduce thermal stresses, prevent failure, and improve overall engine efficiency. This work presents a detailed numerical study of various cooling configurations by applying two advanced leading-edge cooling methods, jet impingement and swirl cooling, across different inlet mass flow rates and jet Reynolds numbers (Rej) ranging from 1,000 to 20,000 to evaluate their cooling performance.
Several advanced leading-edge cooling configurations are proposed and compared with the conventional single-outlet design. A multiple outlets impingement configuration with a return channel addresses the limitations of the single outlet approach by reducing jet interference, improving flow distribution, and enhancing both heat transfer and overall cooling effectiveness. The maximum surface temperature decreases from 345 K and 460 K (single outlet) to 338 K and 425 K (multiple outlets) at Rej = 10.47×10³ and 1.047×10³, respectively. At Rej = 10.47×10³, the maximum Nusselt number (Nu) on the surface with multiple outlets exceeds that of the single outlet case by 56% near the center of the leading edge, 44% in the mid-downstream region, and 28% near the top region.
To address practical installation considerations, the investigation also examines how inlet adapters in leading-edge configurations impact cooling performance. Two inlet designs, one with a rectangular entrance without an adapter and the other similar one with an adapter, are evaluated under identical conditions. The circular adapter, while beneficial for alignment and sealing, causes uneven local heat transfer and broadens the distribution of Nu, whereas the rectangular inlet provides more uniform cooling. With the adapter, only 4–5% of the total mass flow enters the first jet at the start of the leading edge, while 12–13% enters the next jet at both Rej. Additionally, at Rej = 1.047×10⁴, the adapter increases the maximum surface temperature by 7% (from 339.5 K to 364 K) and decreases the maximum and average Nu by 1.6% and 1%, respectively. The inlet adapter study further examines how tapered and straight impingement jet geometries affect mass flow distribution, heat transfer, and internal flow behavior. The findings show that tapering has a more substantial impact at lower Rej, where it improves upstream jet utilization and local heat transfer by enhancing flow attachment. At Rej = 5×10³, the tapered configuration allowed 6.4% and 15% of the total mass flow through the first two jets, compared to 5.9% and 13.5% in the straight configuration, and achieved a 6.5% higher peak Nu for the tapered jet. These differences decrease at Rej = 10 × 10³, where both geometries demonstrate nearly identical results, indicating that tapering\u27s effect is less significant at higher flow momentum. The results highlight the importance of optimizing adapter configurations, with recommended angles between 10° and 20°, for effective momentum diffusion and uniform flow distribution. This novel target-based optimization concept, analogous to a targeted therapy approach, is introduced to enhance leading-edge cooling by adaptively modifying only the geometrically affected regions identified through flow and heat transfer analysis.
Building on these findings, the study demonstrates the advantages of multiple outlets with swirl cooling and evaluates their thermal performance. The research further extends the concept of multiple outlets to three swirl jet configurations: inline, staggered, and inline with separation walls. Results show that the first outlet is highly sensitive to geometry, with the staggered configuration reducing flow by up to 35.8% compared to the inline configuration. In contrast, the walled configuration increases flow by up to 50% relative to the inline configuration without walls. Meanwhile, the last outlet experiences almost no change in the staggered case but decreases by 16.4% in the walled case at Rej = 5×10³. At Rej = 20×10³, the increase is significant (23.71% in staggered, 4% in walled). These patterns suggest that staggered jets redistribute coolant toward the far downstream outlet, while the walls redirect the flow to maintain balance across the outlets.
These findings provide clear design guidelines for next-generation turbine blade leading-edge cooling, highlighting the benefits of multiple outlets configurations and optimized inlet geometries for enhanced durability and fuel efficiency in high-performance gas turbines
THE EFFECT OF THE JUDGMENT RENDERED BY THE FEDERAL SUPREME COURT DECLARING UNCONSTITUTIONALITY
No full textThis study aims to demonstrate that the constitutional judiciary constitutes the fundamental guarantee for the supremacy of the Constitution and the protection of rights and freedoms. It exercises subsequent control over legislative provisions to ensure their conformity with the Constitution. The judgments rendered by the constitutional court are objective in nature, directed at the challenged provision itself rather than the individuals concerned, which grants them absolute authority binding upon all legislative, executive, and judicial authorities, as well as individuals. Such judgments are not subject to any means of appeal. The importance of a judgment declaring unconstitutionality is reflected in its wide-ranging effects, especially since the relevant legislative provisions may have been in force and applied for many years. Therefore, it is necessary to carefully address the implications of such judgments in order to avoid disruption of legal relationships or the creation of a legislative vacuum. For this reason, the UAE legislator has paid special attention to regulating these judgments in a manner that balances the protection of rights and freedoms on one hand, and the preservation of legal stability and public order on the other. In the field of tax legislation, a judgment declaring unconstitutionality has a direct impact on the financial rights of individuals as well as on the State’s public treasury. Accordingly, the UAE legislator has adopted a specific approach to dealing with the effects of such judgments, seeking to achieve a balance between protecting individual interests and maintaining public revenues. As for criminal provisions, given the gravity of their consequences on individual liberty and the right to life, the general rule is that a judgment declaring unconstitutionality shall have retroactive effect, so that all persons to whom the unconstitutional provision was applied may benefit from it, in order to ensure equality and criminal justice
YOLO-BASED MARINE SEARCH AND RESCUE USING UAV MULTI-DATASET IN CHALLENGING WEATHER CONDITIONS
No full textObject detection models, powered by deep learning and computer vision, are revolutionizing marine search and rescue (SAR). By analyzing aerial imagery and live drone footage, these systems automatically identify critical targets like survivors, life rafts, and debris across vast and treacherous ocean areas. This capability enhances operational efficiency by reducing human workload and accelerating response times, even in challenging conditions such as poor light, high seas, or cluttered backgrounds. The result is continuous monitoring, faster decision-making, and a significantly improved probability of successful rescue.Departing from prior methodologies, YOLO introduced a paradigm shift through its single-shot architecture, which concurrently predicts bounding boxes and class probabilities in a single forward pass. This foundational design is responsible for YOLO\u27s defining trait: remarkable inference speed that made real-time object detection feasible. The ongoing evolution of the YOLO lineage reflects persistent architectural enhancements, yielding models that strike a robust balance between speed, precision, and size, thereby cementing their effectiveness across a wide spectrum of tasks.This study benchmarks the performance of YOLO models for maritime SAR under challenging weather conditions using the SeaDronesSee and AFO datasets. The results show that while YOLOv7 achieved the highest mAP@50, it struggled with detecting small objects. In contrast, YOLOv10 and YOLOv11 deliver faster inference speeds but compromise slightly on precision. This study also discusses enhancing the dataset using synthetic data to the existing dataset, which showed slightly higher experimental performance than the standalone dataset. The key challenges discussed include environmental variability, sensor limitations, and scarce annotated data, which can be addressed by such techniques as attention modules and multimodal data fusion. Overall, the research results provide practical guidance for deploying efficient deep learning models in marine SAR, emphasizing specialized datasets and lightweight architectures for edge devices
BIO-CORROSION STUDIES ON NOVEL Zr‑Co‑Ti BASED METALLIC GLASS ALLOYS FOR BIOMEDICAL IMPLANT APPLICATIONS
No full textThe relentless pursuit of advanced biomaterials that synergistically combine high mechanical strength, exceptional corrosion resistance, and inherent biocompatibility is critical for next-generation medical implants. This research addresses this challenge through the development and multi-faceted evaluation of a novel library of Zr-Co-Ti-based metallic glasses (MGs); Zr60Co30Ti10, Zr55Co35Ti10, and Zr50Co40Ti10, fabricated via melt-spinning. The alloys were confirmed to be fully amorphous by X-ray diffraction (XRD), a structure underpinned by exceptional thermal stability, with the Zr60 composition exhibiting a larger supercooled liquid region (Δ) of 149.68 °C, a hallmark of a strong glass former. This amorphous nature conferred outstanding mechanical properties, with nanoindentation revealing a high hardness of 9–9.5 GPa paired with a moderate elastic modulus (85–115 GPa), mitigating the risk of stress-shielding and positioning them favourably against bone. The electrochemical performance of these MGs was rigorously assessed across a physiologically relevant spectrum of environments, including acidic (HCl), neutral (NaCl), alkaline (NaOH), and simulated body fluids (SBFs) such as Artificial Saliva Solution (ASS), Artificial Blood Plasma (ABP), Phosphate-Buffered Saline (PBS), and Hank’s Balanced Salt Solution (HBSS). Potentiodynamic polarization (PDP) and electrochemical impedance spectroscopy (EIS) unveiled extraordinary corrosion resistance, with passive current densities on the order of 10-11 A/cm² and charge transfer resistance often exceeding 1006 Ω·cm². A compelling composition dependent hierarchy emerged. The Zr60Co30Ti10 (Zr60) alloy demonstrated consistently superior and versatile performance, particularly excelling in neutral and alkaline conditions. Intriguingly, while the Zr55Co35Ti10 (Zr55) and Zr50Co40Ti10 (Zr50) alloys showed enhanced resistance in highly alkaline environments (pH12), a reversal of the Zr60 trend, their overall performance remained secondary. This superior corrosion resistance was further validated against other melt-spun MG systems like Cu51Zr30Hf14Ag5 and Zr38Co34Al10Cu10Ti8. Post-corrosion analysis via SEM, EDS, and XPS confirmed the formation of stable, dense passive films rich in ZrO₂ and TiO₂, effectively hindering ion release. The biological efficacy of these alloys confirmed a crucial dual functionality. In vitro cytocompatibility assays using L-929 fibroblast cells confirmed excellent cell viability and proliferation, indicating minimal cytotoxic response. Furthermore, the MGs exhibited significant inherent antibacterial properties, drastically reducing biofilm formation against pathogenic strains of Listeria monocytogenes and Escherichia coli compared to conventional medical alloys. Molecular dynamics simulations provided profound atomistic insights, revealing a dominance of full icosahedral clusters via Voronoi tessellation and a split-second peak in the radial distribution function (RDF), which fundamentally justify the high glass-forming ability and the observed combination of properties. The common neighbour analysis (CNA) further confirmed the amorphous structure of the simulated Zr-Co-Ti based MG alloy (Zr60). The synergistic integration of exceptional corrosion resistance, optimal mechanical properties, dual functionality in promoting cell growth while inhibiting bacteria, and a robust amorphous structure establishes this novel Zr-Co-Ti based MG alloy system, particularly the Zr60Co30Ti10 composition, as a future candidate for demanding biomedical applications
Imaging in Sports Medicine: Frequency of Abnormal Findings and the Value of Point-of-Care Ultrasound
No full textMany musculoskeletal imaging modalities are considered the cornerstone of sports medicine, though the diagnostic yield is different across each modality. Radiographs often detect abnormalities in fewer than half of cases, with the MURA dataset reporting only 39% of abnormal extremity films (Rajpurkar et al., 2018). In contrast, musculoskeletal ultrasound has been shown to identify pathology in up to 83% of examinations (Chiavaras et al., 2008). Point-of-care ultrasound (POCUS) further enhances rapid bedside diagnosis, with emergency department studies showing pathology in approximately 79% of cases (Situ-LaCasse et al., 2018). This review aimed to describe the frequency of abnormal findings across imaging modalities, supplemented by local clinical data
VIRTUAL REALITY IN ESL TEACHING AND LEARNING IN A HYBRID LEARNING ENVIRONMENT IN THE UAE
No full textThis study investigates the impact of Virtual Reality (VR) on the oral proficiency of English as a Foreign Language (EFL) learners in the United Arab Emirates (UAE). It also explores the perceptions of students and their instructor regarding the use of VR as a tool for language learning. The research was conducted in an English Communications program at a higher education institution in the UAE, with 44 Emirati students divided into a control group and an experimental group. The control group practiced speaking using traditional classroom methods, while the experimental group engaged in communicative activities through VR environments. A mixed-methods design was employed: quantitative data were collected through pre- and post-speaking tests, while qualitative data were obtained from a student survey and a semi-structured teacher interview. The study lasted for 10 weeks. The findings revealed that, although both groups demonstrated improvement, mid- to high-proficiency learners in the experimental group showed stronger gains in oral proficiency compared to their peers in the control group. Lower-proficiency learners, however, required more structured scaffolding and guidance to benefit fully from VR-supported tasks. Students expressed positive perceptions of VR, describing it as engaging, motivating, and less anxiety-inducing than face-to-face communication. They emphasized its role in building confidence, practicing speaking, and fostering collaboration with peers. At the same time, students noted technical challenges such as audio clarity, connectivity, and the need for preparatory training. The teacher echoed these perspectives, highlighting VR’s potential to create immersive and interactive learning experiences, while also pointing to institutional and logistical barriers that limited its seamless integration. Theoretically, the study extends the boundaries of the Sociocultural Theory (SCT) by showing that scaffolding, mediation, and internalization can occur effectively in digitally immersive environments, not just face-to-face settings. It also expands Situated Learning Theory (SLT) by demonstrating that VR can replicate authentic, culturally relevant contexts for language use, allowing learners to participate meaningfully in simulated real-world settings. Pedagogically, the study highlights the importance of aligning VR activities with course outcomes, embedding scaffolding for lower-proficiency learners, and providing adequate institutional support for sustainable implementation. Importantly, this research represents one of the first empirical investigations of VR-supported oral proficiency development in EFL learners within the Gulf region, and the first within higher education in the UAE. While there is growing interest in immersive technologies across the UAE’s education sector, no empirical studies to date have directly examined the impact of VR on EFL oral proficiency in Gulf higher education contexts. Against this backdrop, the present study contributes original insights by addressing this gap and offering a localized perspective on how immersive technologies can be integrated into EFL classrooms to enhance oral proficiency