Onion cultivation is essential for food security and income generation, especially among smallholder farmers. However, production is severely affected by fungal and bacterial diseases that reduce both yield and quality. Traditional detection methods rely on manual inspection, which is often slow, subjective, and prone to human error. Although artificial intelligence has improved automated image-based detection, many existing systems focus only on visible symptoms and fail to integrate intelligent decision-making mechanisms. To address this gap, this study developed an Agent-Based Intelli-gent System for onion disease detection, integrating the YOLOv8 deep learning object detection model within a multi-agent framework. The model was implemented using Python and the Ultralytics YOLOv8 framework on Google Colab. Experimental results demonstrated outstanding performance, achieving a precision of 0.998, recall of 1.000, and mAP@0.5 of 0.995 on unseen test data, with only one misclassification observed. The system significantly outperformed benchmark models, confirming its robustness and generalization capability
Critical anomaly detection difficulties, such as false alarms during workload variations and delayed breach detection, have been brought about by the quick adoption of cloud-based storage systems. Data integrity and operational effectiveness are jeopardized by traditional static models' inability to adjust to the dynamic nature of cloud settings. In order to improve anomaly detection accuracy and resource optimization, this study created and verified an adaptive machine learning framework that makes use of real-time model updates and domain-specific cloud infrastructure information. CloudSim simulations of 1,000 cloudlets (10 runs, σ = 0.000), a quantitative survey of 51 IT specialists (92.7% response rate), and qualitative interviews with 13 infrastructure administrators were all included in the mixed-methods sequential explanatory design. A substantial importance-implementation gap in domain knowledge was found (Δ = 1.45, p <.001). With only 15% CPU overhead, the suggested framework, which is based on a domain-enhanced Random Forest, improved the F1-score by 64% and decreased false positives by 54% when compared to static thresholds. By bridging the gap between theoretical machine learning and the realities of cloud infrastructure in Africa, the study offers a deployable framework for improving cloud security in Kenya and other resource-constrained environments.
Machine Translation (MT) has achieved significant progress with the emergence of Transformer-based Neural Machine Translation (NMT) models. However, translating user-generated content (UGC) remains challenging due to the presence of noisy and informal linguistic features such as abbreviations, spelling errors, slang, emojis, code-switching, and inconsistent grammatical structures. These characteristics often degrade translation quality because conventional NMT systems are primarily trained on clean and well-structured corpora. This study proposes an Efficient Dual-BERT Adversarial Network (DBAN) to improve the translation of noisy UGC by integrating contextual representation learning with adversarial training. The proposed framework employs two pretrained BERT encoders to independently learn contextual representations of source and target sentences, while a cross-attention mechanism enhances semantic alignment between both languages. An adversarial discriminator is incorporated to distinguish authentic target representations from generated representations, encouraging the model to learn robust and domain-invariant semantic features. Furthermore, parameter sharing, knowledge distillation, and layer-freezing techniques are introduced to reduce computational complexity without compromising translation performance. The proposed model was evaluated against Standard Transformer NMT and BERT-enhanced NMT using parallel corpora containing informal user-generated text. Translation performance was assessed using Bilingual Evaluation Understudy (BLEU), Metric for Evaluation of Translation with Explicit Ordering (METEOR), Translation Error Rate (TER), and robustness under noisy input conditions. Experimental results demonstrate that the proposed DBAN consistently outperforms the baseline models by producing more accurate translations, preserving semantic meaning more effectively, and exhibiting greater robustness to noisy and domain-diverse user-generated text. The integration of dual contextual encoders and adversarial learning significantly improves contextual understanding and cross-lingual semantic alignment while maintaining computational efficiency. These findings demonstrate that the proposed framework provides a practical and scalable solution for enhancing machine translation of user-generated content and contributes to the development of more robust and context-aware multilingual translation systems suitable for real-world digital communication.
Cementitious Additive Manufacturing (CAM), commonly known as 3D Concrete Printing (3DCP), has emerged as a transformative digital fabrication technology within the paradigm of Construction 4.0. Despite its promise, most existing 3DCP platforms rely on monolithic gantry or robotic-arm configurations that are constrained by limited build volumes, high transportation and installation costs, low adaptability to complex construction sites, and vulnerability to single-point system failures. These limitations restrict scalability, reduce productivity gains, and impede widespread industrial adoption within the construction sector. Although significant advances have been made in cementitious material science and robotic motion control, a critical research gap remains in the development of scalable and reconfigurable modular 3D printing architectures capable of adaptive deployment across diverse construction environments. This study proposes a next-generation modular 3D printing system designed to address these challenges through distributed mechanical modules, plug-and-play structural interfaces, and a synchronized multi-unit control architecture. The research aims to design, develop, and experimentally validate a scalable modular CAM platform that improves geometric accuracy, structural integrity, operational resilience, and construction efficiency. The methodological framework adopts a systems engineering approach integrating computational modelling, digital twin simulation, prototype fabrication, and experimental validation of mechanical and dimensional performance metrics. The proposed system integrates scalable modular hardware with real-time distributed control, enabling dynamic expansion of build volume while maintaining deposition precision and structural stability. Ultimately, the platform seeks to enhance construction productivity, reduce material waste and embodied carbon, and support sustainable industrialized construction aligned with the United Kingdom’s Net Zero 2050 strategy.
Cementitious Additive Manufacturing (CAM), commonly known as 3D concrete printing, has emerged as a transformative digital construction technology that offers unprecedented design flexibility, eliminates the need for formwork, enhances material efficiency, and reduces construction time. However, the widespread adoption of CAM remains constrained by conventional monolithic and fixed-scale printing systems that lack the adaptability required for diverse project sizes, complex geometries, and varying site conditions. These limitations are further compounded by high capital investment, transportation constraints, and limited scalability, particularly in remote and resource-constrained environments. This study presents the design and development of a scalable modular 3D printing system tailored for CAM applications. Using a Design Science Research (DSR) methodology, the proposed platform integrates a modular mechanical architecture, adaptive extrusion mechanisms, distributed sensing, cyber-physical control, material–process coupling, and parametric toolpath planning to enable intelligent, flexible, and reconfigurable construction. Experimental results demonstrate that the proposed modular CAM platform achieves robust precision, reliable extrusion consistency, high dimensional accuracy, rapid reconfiguration, and excellent scalability across varying build volumes. The system also improves deployment flexibility, reduces material waste and operational costs, and supports efficient on-site and off-site fabrication. Overall, the proposed framework provides a practical, scalable, and sustainable solution that advances next-generation digital construction while addressing critical limitations of existing CAM systems.
Plants play a vital role in providing food on a global scale. Several environmental factors contribute to the occurrence of plant leaf diseases, leading to substantial reductions in crop yields. Nevertheless, the process of manually detecting plant leaf diseases is both time-consuming and detection to errors. However, despite these applications, several gaps in plant leaf disease research still need to be addressed for efficient disease detection. This study presents an exploring of enhanced sustainable agriculture for leaf disease detection using machine learning. This proposed system uses tomato leaf disease dataset from Kaggle. The dataset undergoes data pre-processing, image splitting, and data augmentation to enhance its detection. The discriminative attributes from leaf images are used to extract and selection of features. The extracted features are used to train and test tomato leaves classification into their respective disease categories by using machine learning classifier. The potential of machine learning as a sustainable tool for precision agriculture demonstrates efficiency and accuracy for identification of tomato leaf diseases. The results show proposed model can effectively support farmers in early disease diagnosis. It reduces dependence on manual inspection in promoting data-driven agricultural practices.
The rapid advancement of technology in the 21st century has influenced mosque management, particularly in optimizing zakat and alms administration. In the Society 5.0 era, mosque management is expected to innovate through digital solutions that ensure efficiency, transparency, and accountability while remaining aligned with Islamic values. However, challenges such as low digital literacy, inadequate infrastructure, and concerns regarding security and transparency often hinder public participation in digital zakat services. This study aims to design a user-friendly interface for a digital zakat application that can be accessed by diverse user groups, including those less familiar with digital technologies. The research employs the User-Centered Design (UCD) approach, following ISO 9241-210:2019, through iterative stages: understanding the context of use, specifying user requirements, developing user flows and low-fidelity wireframes, and creating high-fidelity prototypes using Figma. To evaluate the effectiveness of the design, usability testing was conducted using the System Usability Scale (SUS) method. The evaluation results show that the E-Zakat application achieved the category of Good Usability (Grade B), with a score of 76 from muzakki users and 79 from administrators. These findings indicate that the proposed interface effectively addresses user needs and enhances usability.
The growing need to integrate sustainability, technological innovation, and territorial development has driven new ways of conceiving geopark management. In this context, multimodal tourism 4.0 emerges as an integrative approach that combines advanced technologies such as artificial intelligence, data analytics, and augmented reality. The objective of this research is to propose a methodology for geopark design management based on this approach. A mixed-methods, quasi-experimental, cross-sectional study was conducted in the Moa-Baracoa region of Cuba. The sample consisted of 200 participants selected through simple random sampling (100 in Moa and 100 in Baracoa), representing 30% of potential users. The results highlight limitations in traditional management models and demonstrate the feasibility of incorporating multimodal technologies to optimize decision-making. It is concluded that the proposed methodology promotes intelligent, participatory, and sustainable geopark management.
The reliability of critical financial infrastructure has become a strategic imperative as digital payment platforms, central bank digital currencies (CBDCs), and real-time settlement systems expand globally. This article investigates the theoretical foundations and practical application of a multi-level Service Level Indicator (SLI) and Service Level Objective (SLO) model designed specifically for critical financial infrastructure environments. The study examines how technical observability metrics can be systematically mapped to business scenarios, regulatory compliance requirements, and operational continuity objectives. Using a comparative analytical framework and case-study evidence drawn from CBDC integration practice at a systemically important bank, the research demonstrates that hierarchically structured SLI/SLO contracts, combined with AI-driven observability, reduce mean time to recovery (MTTR) by an order of magnitude and enable proactive error budget management. The findings reveal that a four-layer SLI/SLO hierarchy covering business outcomes, service contracts, component metrics, and infrastructure signals is necessary and sufficient for aligning engineering reliability work with financial regulatory mandates. The article will be of interest to financial technology architects, reliability engineers, banking regulators, and researchers working at the intersection of distributed systems engineering and financial services compliance.
The accelerating deployment of artificial intelligence across global industries is producing a paradox of progress: while AI drives unprecedented gains in organizational productivity and decision-making quality, it is simultaneously dismantling the entry-level job market that has historically served as the primary on-ramp to professional careers. This paper argues that the most consequential and least adequately examined dimension of AI's workplace impact is not aggregate job displacement, but the selective erosion of junior and entry-level roles across white-collar sectors including finance, law, marketing, journalism, software development, and customer service. Drawing on labor market data, empirical studies, and organizational case studies, we document how generative AI tools are enabling organizations to compress or eliminate the early career tier, stranding a generation of young workers without the experiential foundation upon which professional competence is built. The paper further provides a comprehensive analysis of the ethical challenges raised by AI in the workplace including algorithmic bias, surveillance, accountability gaps, consent, and the concentration of economic power and argues that these ethical failures are structurally connected to the entry-level displacement crisis. We conclude with policy and organizational recommendations oriented toward preserving equitable pathways into the labor market.
Efficient energy scheduling in heterogeneous computing environments is a critical challenge, as task allocation decisions directly affect both energy consumption and execution performance. This work presents an energy aware scheduling framework based on a discretized grasshopper optimization algorithm (GOA), designed to balance energy reduction with acceptable makespan. The model formulates scheduling as a constrained objectives optimization problem, incorporating energy use, makespan, heterogeneous resource capacities, workflow precedence, and non preemptive execution. A constant aware representation and repair based decoding strategy enable GOA to generate feasible task to resources assignments. Implemented in Python, the framework is evaluated against HEFT, Min and Random scheduling under varying workload. Results show that the schedules based on GOA achieves lower energy consumption and improved performance delay energy while maintaining competitive makespan, with performance gains becoming more pronounced as workload complexity increases. These findings demonstrate the scalability and effectiveness of discretised GOA as a metaheuristic solution for energy aware scheduling in heterogeneous systems.
