Browse Items (16481 total)
Sort by:
-
A Hybrid FinTech Fraud Detection Model Integrating Multiscale Entropy and Transformer-GAT Techniques
Credit card fraud has become a major concern in the FinTech industry due to the rapid growth of digital payment platforms and the increasing sophistication of fraudulent activities. Accurate and timely detection of fraud is essential to minimize financial losses and maintain trust in FinTech services. This study presents a hybrid deep learning framework for credit card fraud detection using the 2023 Credit Card Fraud Detection Dataset. The proposed approach with data preprocessing, which includes handling missing values, removing duplicate entries, and encoding categorical features to ensure clean and structured input for modeling. Normalization is applied to scale features uniformly, preventing bias from varying magnitudes and improving model convergence. Multiscale Entropic (MSE) analysis is employed for feature extraction, capturing both short- and long-term temporal patterns within transaction sequences, enhancing the representation of complex transactional behaviors. The extracted features are then processed using a Transformer-GAT classifier, which combines the attention mechanism of Transformers with Graph Attention Networks (GAT) to learn complex inter-transaction dependencies and graph-based relationships. This hybrid architecture enables the model to capture both local and global patterns, improving fraud detection performance. On the training dataset, the model achieved outstanding results with 98.65% accuracy, 98.70% precision, 98.50% recall, and an F1-score of 98.60 %, demonstrating a strong balance between correctly identifying fraudulent transactions and minimizing false alarms. The approach offers significant advantages for FinTech applications, including robust handling of imbalanced data, effective detection of subtle fraud patterns, and strong generalization to unseen transactions. 2025 IEEE. -
A Systematic Review and Meta-Analysis of Pneumonia Diagnosis Using Machine Learning Techniques
Pneumonia is an infection that results in inflammation of the lungs and, if not identified in time, can be life-threatening. The most frequent method of diagnosing pneumonia is chest X-rays; the pictures are scrutinized closely. Pneumonia is still a global health burden. The accurate and timely diagnosis is difficult, especially in low-resource settings. X-rays have served as a primary key for the identification of pneumonia for many years. However, with the recent advancement in artificial intelligence technologies, especially deep learning and machine learning, there's now a potential to automatically detect and classify pneumonia using chest x-ray images. This review examines the research from 2019 to 2024 to understand the current trends and future direction in various deep learning and machine learning models. These encompass the convolutional neural networks, transfer learning methods, combined network designs, explainable AI models, and the use of radiomics with conventional machine learning techniques. However, the three significant challenges remain differences in the data, an imbalance between the classes, and a limited ability to apply these methods in real clinical settings. Based on the review, this paper suggests more future research on machine learning techniques for detecting pneumonia. In this work, a new system is also introduced to improve both case identification and the clinical diagnosis process. The proposed model was evaluated using the Key Parameter Indicator (KPI) as a feature and was compared with an earlier model. Finally, recommendations are provided for future research on trustworthiness, clinical usefulness, and multi-modal AI systems. 2025 IEEE. -
Integrating Behavioural Science using the Psycho-Intelligence Framework in Connected Systems
The fast-growing convergence of neuroscience, behaviour computing, and adaptive artificial intelligence (AI) offers the possibility to transform human, machine interaction. This work presents Psycho-Intelligence, a new, closed-loop system that merges electroencephalography (EEG) and inertial motion unit (IMU) signals to adaptively recognise and react to users' cognitive and affective states. Levying low-cost wearable sensors (Muse EEG and MPU-6050), the system has real-time signal acquisition, sophisticated preprocessing, spectral and statistical feature extraction, as well as multimodal fusion features. Dimensionality reduction and feature selection techniques, including Principal Component Analysis and XGBoost gain metrics, enhance learning optimally. Multiple machine learning algorithms like Random Forest, SVM and XGBoost are trained to identify engagement states with high accuracy, warranted by extensive testing through cross-validation, ROC AUC, and F1-scores. The pipeline is incorporated into an adaptive feedback system that can regulate chatbot tone, learning material, or interactive graphics based on detected user states. Statistical validation with linear mixed models confirms the robustness of EEG-derived measurements in engagement prediction. The research establishes a new paradigm for emotionally intelligent AI systems and provides a technical foundation for ethical, real-time psycho-behavioural intelligence for communication networks, education systems, and cognitive health monitoring. 2025 IEEE. -
Machine Learning Algorithms for Optimizing Blockchain-Based Decentralized Autonomous Organizations
This research investigates the integration of machine learning algorithms within blockchain-based Decentralized Autonomous Organizations (DAOs) to enhance operational efficiency, resource allocation, decision-making, and governance. While DAOs provide a transparent and trustless mechanism for digital collaboration, they face challenges related to scalability, bias, data privacy, and coordination. We propose a novel framework that leverages supervises learning models for predictive analytics, reinforcement learning for autonomous decision-making, and unsupervised learning for anomaly detection in DAO voting and resource usage patterns. The study also addresses security and privacy risks by incorporating federated learning and homomorphic encryption. Our proposed model demonstrates improved throughput, decision accuracy, and fairness, as evidenced by performance benchmarks against traditional DAO implementations. The findings suggest that machine learning can significantly optimize DAO architecture and contribute to a more scalable, democratic, and intelligent decentralized ecosystem. 2025 IEEE. -
HALC: An AI-Driven Legal Decision-Making Framework - A Qualitative NVivo Case Study on Tribal Rights
This study proposes a structured human-AI collaboration framework for legal and ethical decision-making, integrating artificial intelligence with human expertise. Unlike fully automated AI systems, it prioritizes transparency, accountability, and ethical oversight. Through expert interviews with legal professionals and AI technologists, we identified key challenges, including bias, lack of explainability, and the need for human validation. Using thematic analysis in NVivo, we developed a stepwise framework that ethically collects data, applies AI-driven analysis, ensures human oversight, and informs policy decisions. This approach enhances human judgment rather than replacing it, with potential applications in law, governance, and public policy. Future research will test and refine this framework. 2025 IEEE. -
Exploring Social Cues and Engagement in Humanoid Robots: A Robosen K1 Case Study
With the increase adoption of humanoid robots in today's world, the need to understand the ways through which these robots communicate social cues has become indispensable for effective human-robot interaction (HRI) in everyday life. The focus of this study is on the examination of the influence of nonverbal behaviour of Robosen K1 (a humanoid robots) on human perceptions and emotional responses. K1 was programmed to perform expressive full-body movements, due its lack of facial expressions, such as dancing, push-ups, and standing on its head. The research design was a mixed-method approach, which combined behavioural observations from live interactions with data from an online survey. Findings from the study revealed positive emotional reactions from participants, most of which described the robot as 'impressive', 'curious', and 'amusing'. Also, results indicated that 89.8% of participants were favourably disposed to engaging with similar robots in the future. Finally, it was found that the robot's gestures, being highly expressive, contributed to perceived personality traits such as 'playful' and 'friendly'. The study, therefore, concluded that a well-designed non-verbal cues would play critical role in enhancing emotional connection, engagement, and trust in humanoid robots, hence, their importance for successful HRI design. 2025 IEEE. -
Enhancing Healthcare Ecosystems Through the Integration of IoT for Patient-Centric Solutions
The Internet of Things (IoT) is a newly implemented technology in the field of healthcare and can enhance patient-centrical care and efficiency in the healthcare field. IoT may be used to assist in delivering real-time health data and predictive diagnostics and custom care plans by interconnecting medical equipment, sensors, and information systems. This paper will discuss how the IoT technologies, particularly wearable sensors, cloud-based analytics and smart health architectures are changing the way healthcare is delivered. The article highlights how the data merge on the utilization of clinical and non-clinical sources to aid in remote patient monitoring, resources use optimization, and positive patient outcomes. It also identifies the concerns of the implementation of IoT such as the security risks, data privacy and the failure to connect the devices and systems. To address them, the paper discusses the new structures that integrate blockchain and artificial intelligence and ensure safe implementation of data management and heightened clinical decision-making. The results of various works of 2015-2020 have revealed that IoT applications and patent health care-related tendencies are growing, which implies that the shift towards interconnected and intelligent ecosystems is rapid. The consequences of this digital transformation are not confined to the hospital sphere only, as it is extended to homecare, telemedicine, and state population management. Lastly, IoT will allow healthcare stakeholders to shift their healthcare systems to patient-centered, rather than hospital-centered systems, in which a focus on accessibility, efficiency, and personalization would be placed on healthcare provision. 2025 IEEE. -
Artificial Intelligence Personalization: Opportunities, Risks, and the Need for Ethical Data Practices
The benefits of AI personalization are numerous; however, the question still remains, what are the side effects of this feature? Will it help make shopping and enjoying content more enjoyable and efficient or will it destroy the trust of the users by creating privacy concerns. These are the issues this article has attempted to discuss and investigate. Artificial Intelligence (AI) personalization is a very helpful and useful feature, but there must remain a proper balance between personalization and data collection in order to ensure client comfort. A transparent and honest collection of data is to be expected for almost all the companies and this data is to be used responsibly. Activities like profiling must be regulated and controlled and should not be left unregulated. Despite laws being updated to be more considerate towards the privacy of users, the development of better enforcement is imperative. Regardless of the presence of appropriate laws, it is important that each individual practice the respective ethics while present in the digital universe. This article is discussed about AI personalization issues and its research challenges. 2025 IEEE. -
Predicting Job Risk from Artificial Intelligence in London Using Supervised Machine Learning Models
This study investigates the risk of job automation in London due to artificial intelligence (AI), applying supervised machine learning techniques to identify occupations most at risk. Leveraging a dataset encompassing job-specific features such as primary tasks, industry domains, and associated AI models, the research develops two predictive models. A Random Forest Classifier is used to categorize jobs as low, medium, or high automation risk, while a Linear Regression model estimates the proportion of each occupation's workload likely to be automated. The Random Forest model achieved a high accuracy rate of 97% in classifying job risk, indicating strong predictive capability. Meanwhile, the regression model explained 85% of the variance in the AI workload ratio, highlighting a significant relationship between job attributes and automation potential. These results suggest that job characteristics are reliable indicators of AI impact, particularly in routine, repetitive, and low-skilled roles that are more easily codified and replicated by algorithms. The findings align with broader economic theories such as creative destruction and technological waves, suggesting that AI not only displaces certain roles but also drives structural transformation within the labor market. By focusing on London, this study provides a localized understanding of how AI is reshaping employment patterns. It underscores the growing urgency for strategic workforce re-skilling and adaptive policy frameworks to mitigate negative outcomes and maximize opportunities presented by AI. Ultimately, this research contributes valuable insights into the interaction between AI technologies and employment, helping policymakers, employers, and educators anticipate change and prepare for a more resilient, inclusive labor market. 2025 IEEE. -
Improving Voltage Regulation in High-Power Solar Applications
This paper presents an advanced solar-powered isolated DC-DC converter optimized for high-power applications, with a focus on precise voltage regulation at the output stage. To mitigate high-voltage stresses typically encountered in single-stage DC-DC converters, a Lossless Active Clamp Flyback circuit is integrated, offering soft switching capabilities and regenerative energy features. The proposed topology is designed using low-voltage devices, enhancing overall system efficiency. A hardware prototype rated at 2 kW has been developed to empirically validate the circuit's performance. Additionally, a novel control algorithm is introduced to further optimize the converter's operational characteristics. The proposed converter is benchmarked against existing solutions, highlighting significant improvements in terms of component count, voltage handling, and energy regeneration. The results demonstrate superior efficiency and robustness, making the system highly suitable for high-power renewable energy applications. Through this innovative approach, the converter offers substantial gains in performance and operational feasibility, especially in scenarios demanding high power density and stringent efficiency standards. 2025 IEEE. -
Innovative Power Conversion Solutions for Renewable Energy and Electric Mobility
The global transition to renewable energy sources and electrification demands efficient power conversion systems for applications like hybrid electric vehicles (HEVs) and energy storage systems. This paper introduces a novel Multi-Port Bidirectional DC-DC/DC-AC Converter (MBPC) with high efficiency, compact design, and versatile functionality. The MBPC supports two input and two output ports, enabling energy flow between renewable energy sources, storage systems, and loads. Its efficiency exceeds 95%, with a power density of over 10W/cm2. The innovative design minimizes component count, reducing manufacturing costs by 30% compared to conventional converters. Extensive experimentation validates its ability to handle varying current-voltage profiles in multiple operational modes, including DC-DC and DC-AC conversions. With applications in grid-tied systems and electric vehicles, the MBPC addresses efficiency, cost, and flexibility challenges in modern power systems. This work contributes to advancing renewable energy integration and efficient electrification solutions. 2025 IEEE. -
AI-Powered Solutions for Legal Compliance in Industrial Workspaces a Psychological and Labour Law Perspective
The complicated issue of the ways to ensure the workplaces in the industries meet the legal standards brings together the psychology of the workplaces, the labour law, and the convolution of the AI design. In this dissertation, the author investigates the use of AI-mediated options to deal with regulatory compliance and psychological and legal issues. The risk associated with the fair labour, discrimination, and safety issues can be addressed with the help of predictive analytics, automated compliance, and AI-compliance monitoring. The use of AI can also be expanded to support worker well-being and mental health through the identification of work stressors, burnout prevention, and creation of a physiologically safe workplace. But in the case of AI, there are ethical or legal considerations around the agency of workers, bias in algorithms, as well as privacy or confidentiality of data. Due to these reasons, it is necessary to adopt the strategy approach, where AI and human observation are used to determine the work decisions trade-offs in an observable and just manner to employers and employees alike. This dissertation also added to the contribution of how AI can benefit the responsible design of industrial workplaces that do not fail to achieve ethical standards, hold to psychological sustainability, and adhere to labour laws, evaluating the psychological effects of labour law actors as well as effects on the workplace. 2025 IEEE. -
Explainable Intrusion Detection System for Internet of Things-explainability with reliability
Explainable Artificial Intelligence (XAI) based Intrusion Detection System (IDS) (X-IDS) has transformed the traditional IDS into interpretable and transparent system with the goal of providing interpretable justification for IDS models. XAI is now being used to extract more appropriate features for specific cyber-attacks. The black-box model of ML based IDS is not capable of giving reason for false positive to the cyber defense personnel. XAI tools reduces this abstraction by locally interpreting the model's behaviour at some datapoints along with global interpretability. This article proposes an explainable IDS by using XAI tools. We used SHAP (SHapley Additive exPlanations) to identify the variations in feature importance of selected ML based IDSs and explain the variations of their detection accuracies. Also, we have shown that with same dataset, feature importance varies differently with different ML models. This leads us to the conclusion that specific set of features are required for specific ML models while other can be discarded. The explainability proposed in this study also help to select less set of features to overcome time of execution and cost. 2025 IEEE. -
Innovative Hybrid Models for Predicting Diabetes: CNN-LSTM Hybrid and Calibrated Soft Voting Model
This study assesses four ensemble techniques - stacking, soft voting, hard voting, and calibrated soft voting - for predicting diabetes onset using the Pima Indians Diabetes dataset. Traditional single-model methods are contrasted with these advanced ensemble approaches, which integrate multiple models to enhance predictive accuracy. The evaluation included metrics such as accuracy, precision, recall, F1 score, and AUC. The CNN-LSTM model was also examined, achieving an accuracy of 75%, precision of 70%, recall of 69%, and an F1 score of 72%. Among the suggested methods, the calibrated soft vote model was the most effective, with improved performance compared to the rest of the techniques. Upcoming studies will address the combination of these models with real-time monitoring systems and deploying their use across a broad range of datasets and medical conditions. 2025 IEEE. -
Indoor Localization and Tracking with IoT: A Critical Survey of Technologies, Challenges, and Future Trends
Indoor localization and tracking have been important areas of research throughout the past 10 years, driven by the expanding Internet of Things (IoT) technologies. The shortcomings of conventional GPS in indoor environments have called for the development of replacement localization methods. This paper presents a methodical review of IoT-enabled indoor localization techniques covering both well-known technologies such as Bluetooth Low Energy (BLE), Radio-Frequency Identification (RFID), Ultra-Wideband (UWB), and Wi-Fi fingerprinting, as well as newer approaches such as Visible Light Communication (VLC). We critically evaluate these technologies by way of a comprehensive analysis of modern research and case studies, emphasizing significant performance criteria such as accuracy, scalability, and energy efficiency as well as pragmatic concerns such as cost and security. Our work looks at field trends still in development, highlights significant gaps and problems, and integrates the current state of the art. We also stress potential application fields - such as smart homes, healthcare, and industrial automation - that stand to benefit significantly from advances in indoor localization. Finally, we outline future research intended to address current limitations, including the need of higher accuracy in complex environments and more robust security measures. 2025 IEEE. -
DermAI: A Deep Learning-Based Mobile Application for Multi-type Skin Cancer Detection
The significance of early skin cancer detection for effective prevention and treatment is underscored by the limitations of traditional manual diagnostic methods used by dermatologists. Leveraging Convolutional Neural Networks (CNNs) and the HAM10000 dataset, this research aims to automate skin cancer classification through dermatoscopic image analysis. The primary objective of the research is an accurate classification system identifying seven specific skin cancer types. The novelty is the deployment of the classification system using a Mobile Application - DermAI. The trained CNN model, spanning 10 epochs, achieved remarkable precision, peaking at a 97.90 percentage test accuracy during the 7th epoch. Evaluation metrics like the confusion matrix confirm its reliability in categorizing lesions, minimizing misclassifications, and validating its efficiency as a diagnostic tool. Transforming the model into TensorFlow Lite format enables seamless integration into mobile platforms, optimizing computational resources. This allows users to access prompt skin cancer classification via an Android application, fostering accessibility to preliminary assessments. Early identification facilitates timely medical intervention, a crucial factor in enhancing prognosis. Through CNNs, TensorFlow Lite, and mobile deployment, this research strives to bridge technology and healthcare accessibility, empowering individuals to proactively manage their skin health based on classification results and initiate timely discussions with healthcare professionals. 2025 IEEE. -
Depression Severity Prediction Among Higher Education Students Using Neural Network Model
Depression significantly affects students' mental health and academic performance, highlighting the need for effective early detection methods. This study investigates machine learning approaches for automated classification of depression severity using responses from the Patient Health Questionnaire-9 (PHQ-9). Deep Neural Network (DNN), Long Short-Term Memory (LSTM), and hybrid models combining structured PHQ-9 scores with descriptive text responses were evaluated. The experimental results show that the LSTM model achieved the highest classification precision (90%), demonstrating its ability to capture sequential relationships between items in PHQ-9. The findings indicate that sequence-based models are well suited to assess the severity of depression in student populations. Integrating such predictive models into digital mental health screening systems may support the early identification of at-risk students and enable timely, data-driven interventions in academic settings. 2026 IEEE. -
Alzheimer's Disease Detection using Deep Feature Extraction and Explainable Machine Learning
Alzheimer's disease (AD) is a progressive neurodegenerative disorder characterized by gradual cognitive decline, posing significant diagnostic challenges that necessitate automated detection systems to aid clinical decision-making. This study presents an explainable machine learning framework for binary dementia classification using deep feature extraction from magnetic resonance imaging. A pretrained ResNet50 convolutional neural network was employed to extract 2048-dimensional feature vectors from 86,437 MRI slices derived from the OASIS1 dataset, encompassing 347 subjects. The dataset was imbalanced, containing 67,222 Non-demented and 19,215 demented slices (combining very mild, mild, and moderate dementia). The aggregated features at the Subject-level were used to train three machine learning classifiers: Logistic Regression, Random Forest, and XGBoost. The XGBoost model achieved the highest accuracy of 77.14, with a precision of 0.84 and a recall of 0.87 for Nondemented cases, demonstrating strong discriminative capability. Gradient-weighted Class Activation Mapping (Grad-CAM) visualizations highlighted the hippocampus and temporal lobes as key regions influencing predictions, aligning with established Alzheimer's pathology. The study demonstrates the potential of combining deep feature extraction with interpretable machine learning for automated dementia screening. 2026 IEEE. -
Enhancing Sign Language Recognition Through LSTM Model
Sign language recognition is a remarkable task in this project completed through two state-of-the-art methods, Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU). This way, the system is able to quickly process each frame of the webcam with real-time information regarding face, body and posture in order to extract critical values. this research seeks to provide the necessary resources and opportunities for deaf people to be able to communicate effectively, obtain an education and enjoy their lives as much as other human beings This makes it a very important tool for education where the system can convert sign motions into text on-the-fly. The data was collected through a live camera, and key points from face, body, and pose were detected for training the model. Kindergarten used the four categories of vegetables, fruits, colors and animals. There were 40 video sequences of 40 frames with a sign in each. the model tries to fit too much to noisy points of data. However comprehensive the training, after 19 epochs the validation accuracy is an impressive 93%. The oscillations in the truth values of models are indicative of some uncertainty in learning where the accuracy is still to be settled. The graph in general shows that the LSTM based sign language movement classifier has a good capacity to learn and identify sign language movements with high precision. 2025 IEEE. -
Machine Learning Models for Apple Disease Detection With Texture Feature Fusion and Feature Selection
Computer vision has become an integral part of modern agriculture. One of the key applications of Computer vision is the automatic detection and classification of plant disease from digital images of plant leaves. In this study we evaluate the discriminatory capability of selected texture features and their fusion in identifying plant diseases from leaf images. Further, the performance of four feature selection algorithms is also evaluated. Texture features are extracted from resized raw images. Experiments are carried out with public data sets of Apple plants. Through extensive experimentation, two classifiers - Random forest and XGBoost are chosen for the evaluation. The feature fusion and feature selection resulted in 85% accuracy. The result is promising as the features are extracted from whole leaf images, without any segmentation. 2025 IEEE.