Journal of Emerging Technology and Digital Transformation

INTELLIGENT FRUIT SORTING, SEGREGATION, AND QUALITY CONTROL FOR SMART FARMING SYSTEMS

*Muhammad Sajjad — 2025-07-27

Fruits are a mainstay of a healthy diet. They keep our bodies healthy because they contain minerals, vitamins, fiber, and water. Manpower is required for the segregation of the fruits to maintain their quality. A lot of time is wasted on segregating fruits to maintain quality. Due to the poor quality of fruits, farmers are facing a huge loss in their agricultural fields. Automation enhances the quality of the fruits and speeds up the segregation process by ensuring accuracy and efficiency. Many Algorithms have been developed by researchers for the segregation of fruits. The proposed Deep learning model YOLOv11 will segregate (Healthy or Rotten) the fruits into their specific classes, ensuring the quality by processing the images of the fruits, gaining validation Accuracy of 97.91%. This study fills the gap between agriculture and technology. It represents the potential of AI in food quality inspection processes.

Key points: Fruits Detection, Classification, Segregation, YOLO (You Only Look Once), Deep Learning, Computer Vision.

EDGE COMPUTING ADOPTION AND ITS EFFECT ON IOT SYSTEM PERFORMANCE IN PAKISTAN: MEDIATING ROLE OF LATENCY REDUCTION AND MODERATING ROLE OF NETWORK SCALABILITY

Geeta Kumari — 2025-08-25

The fast development of the Internet of Things (IoT) has imposed a tremendous stress with regard to latency, reliability and scalability especially in third-world countries like Pakistan. This paper has analyzed how the application of edge computing to the IoT systems influence its performance, where latency decrease is a mediator, and network scalability is a moderator. The research design was quantitative and the data was gathered by targeting professionals working in telecommunications, smart manufacturing, transport, and ICT infrastructure fields in key cities of Pakistan, i.e. Karachi, Lahore and Islamabad. There were a total of 220 valid responses of IT professionals, IoT system architects, cloud and edge engineers, and project managers having pertinent area experience. Partial Least Squares Structural Equation Modeling (PLS-SEM) was applied with SmartPLS. The model results indicated an important positive effect of edge computing adoption on performance of IoT system, which means that local data processing had a positive effect on the speed of responses, the efficiency of the system, and its reliability. The partial mediation of this relationship indicated that the decrease in latency was one of the main ways in which the introduction of edge computing contributed to a better performance as it reduced the time delays in data transmission. The moderating analysis also suggested that the positive impact of edge computing on performance of IoT systems was enhanced in the higher network scalability contexts, which points to the significance of scalable infrastructure on maximizing the benefit of edge computing application. This research added value to both the academics and practice because it offered empirical data on the impact of technological improvements in terms of edge-based computing to enhance the performance of smart systems in developing countries.

Keywords: Edge Computing, IoT System Performance, Latency Reduction, Network Scalability, Pakistan.

CLOUD-NATIVE ARCHITECTURES FOR LARGE-SCALE AI-BASED PREDICTIVE MODELING

Muhammad Talha Tahir Bajwa — 2025-08-22

The demand of adapted, expandable, efficient deployment techniques has become more acknowledged because of the accelerated growth of artificial intelligence (AI) initiatives and high intricacity of big forms of predictive modeling. Cloud-native architectures which are founded on concepts such as serverless computing, microservices, orchestration and containerization create a solid foundation in satisfying these needs. Dividing its emphasis between distributed model training, real-time inference, and automated lifecycle management, this paper explores how cloud-native technology acts to enable large-scale AI-based predictive modeling. By integrating MLOps practices with elastic cloud infrastructure, organizations will be able to realize better fault tolerance, faster deployment schedules, and the most efficient use of resources. The proposed methodology demonstrates that cloud-native ideas can help AI systems work with a vast amount of data, dynamically adapt to changing loads, and maintain high performance levels in the actual environment.

Keywords: Cloud-native architecture, predictive modeling, containerization, MLOps, microservices, real-time inference, serverless computing.

AI-POWERED BEHAVIORAL ACCESS CONTROL FRAMEWORK USING SMART CONTRACTS ACROSS SDN ENVIRONMENTS

Afzal Hussain — 2025-08-20

This research investigates the integration of artificial intelligence with blockchain-based smart contracts to create dynamic access control systems that adapt to evolving user behavior patterns. We propose a novel framework that leverages generative AI models to analyze user interactions across multi-domain Software-Defined Networking (SDN) environments and automatically adjust access permissions through blockchain smart contracts. Our approach addresses two critical research questions: (1) how can blockchain-based identity management scale effectively across multi-domain SDN environments? And (2) How accurate are generative AI models in modeling and predicting malicious insider behavior? Through empirical evaluation across three enterprise networks with 5,724 users, we demonstrate that our proposed system achieves 94.3% accuracy in anomaly detection while reducing administrative overhead by 76% compared to traditional role-based access control systems. The framework shows significant improvements in scalability with a throughput of 1,450 transactions per second while maintaining security posture across federated domains.

Keywords: Smart Contracts, Access Control, Behavioral Analysis, Artificial Intelligence, Blockchain, Software-Defined Networking, Zero-Trust Architecture, Insider Threat Detection

IDENTIFICATION OF FAKE NEWS ON SOCIAL MEDIA USING TEXT MINING

Areeba Razzaq — 2025-08-18

The fake news detection is one of the most pressing problems in online social media and it has major social and political implications. Several automated techniques for the filtering out fake news have been suggested earlier. This research focuses on the issue of detecting fake news on social media employing text mining approach. We propose an ensemble approach based on hard voting, combining the strengths of three machine learning models. So there are Logistic Regression, Random Forest, Decision Tree. Social media text data is cleaned and normalized and turned into a structure that can be effectively analyzed. The ensemble method combines the decision that is produced by three models into one decision. By using the experimental results, it can be concluded that the proposed approach highly effective with an accuracy rate of 89% that helps in differencing the real and fake news. It is possible to identify fake news using machine learning through this approach, which can work as a perfect solution to the problem.

Index Terms Ensemble Learning, Hard Voting, Logistic Regression, Random Forest, Decision Tree.

AUTOMATED CROP SECURITY: A DEEP LEARNING APPROACH TO DETECTING AND DETERRING BIRDS INTEGRATED WITH A LASER SYSTEM

Muhammad Sajjad — 2025-05-22

Most people depend on agriculture for their livelihood. Farmers' income is closely linked to crop yield, which has been decreasing due to natural factors and a lack of advanced technology. Birds play a vital role in the ecosystem, but they can also significantly affect crop yield. The birds often damage grain crops, so special attention is needed to address the harm they cause. Controlling the birds is important in farming to prevent the loss of food. In the article, we developed an automatic system to deter birds. The system uses deep learning to detect birds and deter them from crop fields. When a bird enters the farm, the system identifies its location through a picture taken by a camera, and the trained model gives instructions to the laser controller to deter the bird from the crops. This model is trained on the dataset (Bird-feeder). The results show that the trained model performed well, achieving a validation accuracy value of 96.02%, macro F-1 scores of 96.5 %, macro precision of 95.8% and macro recall of 95.4%. The trained model can detect even small birds with accuracy. Farmers can use this model to improve the production of their crops by deterring the birds.

Key Points: Deep Learning; YOLOv11; Automatic Birds Detection and Deterrence; Real Time Object Detection; Agriculture.

IDENTITY THEFT RISK ASSESSMENT TOOLS IN THE BANKING SECTOR OF PAKISTAN

Amna Abro — 2025-08-06

Identity theft has emerged as a critical threat to the financial security of customers and the integrity of banking operations in Pakistan. With increasing digitization of financial services, banks face mounting risks related to fraudulent access, impersonation, and unauthorized data usage. This research investigates the effectiveness of existing risk assessment tools implemented by banks in Pakistan for detecting and preventing identity theft. By analyzing security practices, technologies used, and challenges faced by banking institutions, this study highlights gaps in risk assessment mechanisms and proposes an improved model based on global best practices. The findings contribute to building a more resilient banking infrastructure capable of proactively managing identity-related fraud.

Keywords. Identity Theft, Risk Assessment Tools, Banking Sector, Cyber Security, Case Study, Qualitative Study

SIMPLIFYING CARDIOVASCULAR RISK PREDICTION: COST-AWARE MACHINE LEARNING AND INTERPRETABILITY FOR RESOURCE-CONSTRAINED HEALTHCARE

Abdul Waheed — 2025-08-06

Heart disease, a leading global cause of mortality, underscores the need for early and accurate detection. This study evaluates five machine learning algorithms, Logistic Regression, Decision Tree, Random Forest, Support Vector Machine (SVM), and Artificial Neural Network (ANN), on the UCI Heart Disease dataset. Preprocessing included normalization, missing value imputation, and cost-aware feature selection via Recursive Feature Elimination (RFE). Models were assessed using accuracy, precision, recall, F1-score, and ROC-AUC metrics. Logistic Regression achieved the highest accuracy (90%), followed closely by SVM and ANN. A novel lightweight hybrid model, combining Logistic Regression with pruned Random Forest feature importance, was developed for resource-constrained settings, ensuring computational efficiency and interpretability. These results highlight the potential of simplified machine learning models as non-invasive tools for clinical decision support in low-resource environments.

Keywords: Feature Selection, Clinical Decision Support, Predictive Modeling, Resource-Constrained Deployment

DEEP CONVOLUTIONAL NEURAL NETWORKS FOR CAVITY AND PIT DETECTION IN DENTAL IMAGES

Muhammad Maaz Ali Khan — 2025-07-29

If it is not addressed, several oral issues can arise from dental caries, one of the most prevalent oral disorders. However, access to professional dental care is often limited, particularly in underserved communities. Our proposed AI-based solution empowers individuals to monitor their oral health and detect early signs of cavities. Children are particularly susceptible to pits and caries in permanent molars, which mostly arise in the cavities on the occlusal, buccal, and palatal surfaces of molars. To address this challenge, we propose developing an AI-powered solution that utilizes smartphone cameras to capture and analyze dental images for cavity detection. This research enables image processing techniques and computer vision algorithms to identify and classify various cavities, including early-stage lesions and enamel defects. Using the Teeth Cave Convolutional Neural Network (CAVTee-CNN), which was particularly designed for dental cavity detection, we were able to minimize information loss during the preprocessing of images. This custom network was developed through extensive experimentation with convolutional layers, activation functions, and pooling mechanisms adapted to highlight dental features. The CAVTee-CNN Model enhances both efficiency and accuracy; it yields the best result of 85.2. By providing timely and accurate cavity detection, the application can empower users to make informed decisions about their oral health and seek appropriate dental care when necessary.

REVOLUTIONIZING DERMATOLOGY: ADVANCED DEEP LEARNING TECHNIQUES FOR AUTOMATED SKIN DISEASE DETECTION

Faisur Rehman — 2025-07-29

Skin is the largest tissue in the humanoid body, and it protects body from external damage, regulates body temperature, and provides the sense of touch. It is made up of three layers: the epidermis, dermis, and subcutaneous tissue. Skin cancer is a disorder in which tissue cells grow abnormally and would extent to additional portions of body. It is typically caused by solar system or tanning bed exposure to ultraviolet (UV) radiation. Early detection of skin cancer is crucial because it increases the likelihood of successful cure and treatment. Although advanced stages of the disease are more difficult to treat and have a worst diagnosis. Regular skin self-examinations, combined with professional skin checks, can aid in the early detection of skin cancer. Machine learning has been used for detection of skin cancer. But it works on features, for which we must first extract the features from the raw data. Feature engineering is a time consuming and challenging task. Deep learning is an emerging area and subfield of artificial intelligence, which extracts useful features from raw data on its own. But it is quite time-consuming task to train such complex deep learning methods and requires huge computational resources. In this work, we have proposed a deep learning model based on ensemble learning for detection of skin cancer. In this model, we used the ensemble learning method for combining four different popular CNN architectures namely InceptionV3, ResNet50, Efficient net and Xception. In this study, we use HAM10000 dataset for our experiments. First, when we train these classifiers, our proposed model was not performing better due to presence of unbalancing in the data. After balancing the dataset, our proposed model was performing better in terms of accuracy, precision, recall and F1-score. On balanced dataset, our proposed model gave an accurate score of 0.94, Precision score of 0.93, Recall and F1-score of 0.94.

Keywords—Skin Cancer, Machine Learning, Deep Learning, HAM10000 dataset, Medical Imaging, Convolutional Neural Networks (CNN), Cancer Diagnosisy

THE EVOLUTION OF SDMS: TRENDS, TRADE-OFFS, AND FUTURE DIRECTIONS

Abdulrehman Arif — 2025-07-22

Software development has undergone significant paradigm shifts, transitioning from rigid, linear models to adaptive, iterative methodologies. This paper critically examines Software Development Methodologies (SDMs) by evaluating key developments, their associated strengths and limitations, and future trajectories. Unlike prior literature, which often presents uncritical endorsements of Agile methods, this study focuses on real-world implementation, contextual fit, and long-term sustainability. Drawing from academic sources, industry reports, and hybrid case studies, the analysis challenges mainstream success narratives and highlights the methodological trade-offs involved in SDM selection. Particular attention is given to the limitations of Agile in distributed and highly regulated environments, where hybrid models such as DevOps and SAFe have emerged as more context-appropriate solutions. The study advocates empirical, future-oriented approaches to guide the development and application of SDMs in increasingly complex and dynamic software engineering settings.

Keywords: Software Development Methodologies (SDMs), Agile vs. Waterfall, Hybrid Approaches, Methodological Trade-offs, Future Trends in Software Engineering

A COMPARATIVE ANALYSIS OF FUNDAMENTAL CONCEPTS OF OPERATING SYSTEM

Asna Riaz — 2025-07-22

This paper provides a comparative review of various fundamental cores of operating system, including Time Sharing, Multitasking, Kernel mode, User mode, Threads, File systems, Virtual memory, Paging, Paging and Swapping, Page hit, Page miss. The review synthesizes findings from 40 research and review papers, examining these fundamental core’s performance across different applications, methodologies, and optimization tasks. The discussion highlights key findings, identifies research gaps, and suggests future research.

Keywords – Time Sharing, Multitasking, Kernel mode, User mode, Threads, File systems, Virtual memory, Paging, Paging and Swapping, Page hit, Page miss.

QUANTUM POWERED FORENSICS: A NEW AGE OF CYBER INVESTIGATION

Muhammad Ahsan Naeem — 2025-07-08

The evolution of quantum computing presents both critical risks and opportunities for digital forensics. This study addresses the urgent need to adapt forensic science in response to quantum threats, especially the vulnerabilities in classical encryption standards exposed by quantum algorithms like Shor’s and Grover’s. This conceptual paper develops an integrated framework that links Quantum Computing Capability (QCC), Quantum-Resistant Cryptography (QRC), Digital Evidence Integrity (DEI), and Forensic Analysis Accuracy (FAA), moderated by Investigator Readiness and Skills (IRS) and Legal and Regulatory Adaptability (LRA). The research builds on theories from quantum mechanics and forensic science to model how QCC can both challenge and enhance forensic processes. Key findings highlight that QRC plays a crucial mediating role in preserving evidence authenticity under quantum threats. The originality of the study lies in its holistic framework, combining legal, technological, and human readiness dimensions. This framework offers strategic value for policymakers, legal institutions, and forensic professionals aiming to secure digital evidence in a post-quantum era. Future work may test this model empirically through SEM/PLS-SEM to guide technological adoption and regulatory reforms. Ultimately, the study contributes a forward-looking roadmap for resilient, quantum-compatible forensic systems.

Keywords: Quantum Computing Capability, Digital Forensics, Quantum-Resistant Cryptography, Evidence Integrity, Legal Adaptability, Post-Quantum Security.