Browse Items (16481 total)
Sort by:
-
Unmasking the Masked: A Classical Machine Learning Pipeline for Detecting Forged Receipts
The abundance of digital and paper document forgery requires strong automated detection tools against financial fraud. This research provides a classical machine learning method for forged receipt detection using multimodal features from image and text modalities. The approach entailed designing a feature set to obtain textural and statistical attributes from receipt images via Local Binary Patterns (LBP) and Canny edge detection, along with structural features obtained from the associated text files. Another demanding issue in this area is the excessive class imbalance between genuine and forged documents. To overcome this issue, Synthetic Minority Over-sampling Technique (SMOTE) is used to create a balanced training dataset. The models are assessed using the macro F1-score, precision, recall, PR AUC and ROC AUC to address class imbalance. The enhanced detection of the minority class is achieved using SMOTE, while hyperparameter tuning leads to the improvements in performance. The final Tuned Support Vector Machine model achieves a macro F1-score of 0.5429, and it has the highest recall on forged receipts, demonstrating that it detects more histories of tampered documents effectively. This research sets a good baseline for receipt forgery detection and emphasizes that class imbalance solving is a key towards creating a working system. 2025 IEEE. -
Privacy Risk Prediction from Social Media Metadata using Feature Selection Approaches
Millions of new people sign up to online social networks (OSNs) every year, which contributes to the growing spread of Personally Identifiable Information. This often ends up occurring unconsciously, either due to the low stakes involved or because the user doesn't understand or underestimates what can go wrong. This trend indicates the need for a trustworthy means to quantify the privacy danger of sharing information online. The volume of OSN data can simply be too staggering for any degree of meaningful manual review, given both the time and man-hours this would entail. This research presents a two-step, unsupervised, and efficient method to estimate privacy risks at the post level. The first step involves using the most advanced reasoning-based Large Language Model, Gemini 2.5 Pro, to generate a comprehensive 'vulnerability score', which is used as a reference for model training. The next step involves comparing the two most used machine learning feature selection techniques, Recursive Feature Elimination (RFE) and Correlation-Based Selection, to select the best features for predicting this score from metadata alone. The results indicate that Correlation-Based Selection produces better results for both the regression and classification-based models, and the top-performing regression model achieves an R-squared of 0.86. Through this, a practical and scalable method to identify privacy-sensitive content effectively on large datasets has been presented in this study. 2025 IEEE. -
Early Sepsis Prediction using Hybrid LightGBM and LSTM Model
Sepsis is a critical organ malfunction that results from an abnormal response of the body to infection and might be lethal. The early detection of sepsis is essential for the patient's life. However, the traditional clinical diagnostic systems are not capable of analyzing the complicated changes in the patient's vitals over time. Therefore, a hybrid predictive framework that merges Light Gradient Boosting Machine (LightGBM) and Long Short-Term Memory (LSTM) networks for fast and accurate sepsis detection in real-time using freely accessible MIMIC-III data, has been proposed in this research. Using LightGBM, the nonlinear relationships among the features are learnt very fast and efficient, while the LSTM gives the temporal dependencies in the sequence of the patient vital signs. The combined output of the two models is said to be more sensitive and robust than that of the single models. A Streamlit-based clinical dashboard is being provided, allowing for real-time predictions and visualization for healthcare professionals. The proposed system has shown a considerable increase in the accuracy of early sepsis detection and offers a non-restricted method for AI-assisted ICU monitoring. 2025 IEEE. -
Premium Unlocked AI for Medical Document Decoding
As healthcare systems evolve to become more digital, an enormous volume of medical data is available in various formats, including unstructured data, scanned documents, handwritten prescriptions, diagnostic images, audio transcriptions, and clinical video recordings. The complexity and unorganised form of data continue to pose serious challenges with regard to automation, accuracy, and consistency in healthcare and insurance businesses. This study introduces an AI-based multimodal framework that incorporates the use of Optical Character Recognition (OCR), the MiniCPMV-4.5 model, and Retrieval-Augmented Generation (RAG) with Large Language Models (LLMs) to enhance the intelligent processing and contextual comprehension of intricate medical data, thus overcoming these limitations. It applies OCR to scanned images and handwritten documents to precisely recover the textual information from them and uses domain-specific named entity recognition (NER) to recognize significant medical information, e.g., patient information, diagnoses, procedures, and financial information. The extracted information is then converted to vector embeddings and stored in a powerful vector database, Milvus, that enables fast and efficient semantic search as well as context-sensitive reasoning. The proposed framework, along with the visual and auditory inputs, video understanding, multilingual capacity, and the S2S (speech-to-speech) and TTS (text-to-speech) translation, makes it more accessible and engaging to the user. This system reduces the level of human involvement and provides real-time insights quickly and more precisely so that more efficient decisions and operations can be made in the fields of healthcare and insurance. 2025 IEEE. -
Edge and Fog Computing in Cyber-Physical Systems
The benefits of cyber-physical system advances include low latency and high bandwidth data processing in areas such as automotive, healthcare, and business automation. Traditional environments are often located in centralized and remote locations and cannot meet the demand. Edge computing and cloud computing have become fundamental concepts that will bring computing closer to the center of the data. Edge computing can reduce latency and bandwidth consumption by processing data on or near IoT devices. Fog computing adds another layer to this by distributing work and storage across multiple nodes, thus providing a scalable and flexible infrastructure. This article discusses the principles, benefits, and challenges of integrating edge and cloud computing into a CPS environment. It leverages the power of proximity-based edge computing and the centralized capabilities of cloud computing to provide scalable, instantaneous responses to CPS applications or time to optimize services. The demonstration shows a variety of things from smart cities to the use of IoT in healthcare in CPS. The article also covers some specific security and privacy issues and future directions in distributed computing, including the role of AI and 5G, which are supposed to offer additional resources in various applications. 2025 IEEE. -
An Algorithmic Approach to Intrusion Detection in Ad Hoc Wireless Networks Based on Artificial Intelligence
The self-configured, autonomous, and framework-free modes of communication that mobile adhoc networks (MANETs) offer have revolutionized our culture. As a result, efforts have been made to explore ways to maximize the potential of MANETs through increased and improved utilization. Standards for AI have been developed thanks to the most recent release of new machine learning technologies. Different security-related issues from malware assaults affect mobile ad hoc networks (MANETs). Any node operates as a router to move data without centralized control, making nodes more vulnerable to threats from other nodes or attackers because of their brief existence. Because of this, MANET needs particular security policies to detect the incorrect entrance of misbehaving nodes. If all nodes are self-assured and correctly collaborate, the networks function better. The paper presents a practical artificial intelligence algorithm-based security system that uses AdaBoost and DT algorithms to recognize and identify packet falling nodes, classify information packets as normal or abnormal, and detect insider threats in real-time. The results showed that DT performed better than AdaBoost, with a 98% accurate prediction rate. Consequently, DT is better able to recognize damaging attacks in MANETs. 2025 IEEE. -
A Comprehensive Review of Advanced Analytics for Predicting HRQoL in Cancer Survivors Using a Synergistic Approach
This systematic review explores the role applied and emerging methods including AI, Explainable AI and Quantum machine learning techniques in the prediction of Health-Related Quality of Life (HRQoL) of cancer survivors. It also gives possible benefits and limitation of using the advanced analytics to predict the HRQoL. In all, 141 research papers implemented in the last fifteen years with focus between the years 2008 to 2023 are analyzed. For the convenience, this literature review is divided into four primary categories - (i) Artificial intelligence, (ii) Explainable artificial intelligence, (iii) Quantum machine learning, and (iv) Synergistic integration. The third way the present systematic review paper differs from other papers in the domain is that the paper offers a direction of future research. Furthermore, the hypothetical illustration is provided in order to compare outcomes of the synergistic approach with the existing data. Consequently, this analysis provides beneficial insights for further research and development of the synergistic approach in both research and clinical practice. The assessment shows that there is a continued need for research focusing on improving the quality of life of those that survived cancer. 2025 IEEE. -
AADS: An Automated Accident Detection and Nighttime Surveillance System Using Fine-Tuned YOLOv10 Deep Learning Techniques
Computer vision-based surveillance is very important today's security systems to detect, track and regulate the security much better than standard cameras. However, like any other performance measurement systems they have potential pitfalls and technical, ethical, and legal implications must be well understood. The continuous rise in connection and interaction implies that safety of the public especially when navigating roads or operating in public domains is paramount. The conventional approaches to accident identification include observation or reporting from witnesses and always record slow and imprecise outcomes. With the improvement of AI and computer visions, especially with deep learning models such as YOLO, accident detection is changing. YOLO v10 which is incorporated in the surveillance systems, performs real time video analysis to provide object and pattern recognition of accidents including car accidents and incidents involving the pedestrians. When applied to the initial set of annotated accident images, the fine-tuning of the YOLO v10 model enhances its detection capability. The system is in watching the video frames that contain aberrations and issues and alarms are issued when the accidents happen and relayed to the monitoring stations or emergency departments for proper response. The optimized YOLOv10 here delivers a meaningful testing score of 72.3% mAP to outperform the regular YOLOv10 efficiency in incident detection. 2025 IEEE. -
An Explainable AI Techniques for Advancing Diabetes Prediction Using Machine Learning
Researchers have developed an automated system to identify diabetes risk. This system combines data from two sources: a collection of female patients in Bangladesh and an expanded dataset from a local textile factory. The expanded dataset includes information from 203 additional patients. The system uses several techniques to improve its accuracy. It first identifies the most important factors for predicting diabetes, then employs a special model to estimate insulin levels. It also addresses challenges like imbalanced data (where one outcome is more common) and explains its predictions using artificial intelligence techniques. This system achieved the superlative results has an 81.0% accuracy rate, 0.812 F1 score, and 0.844 Area Under the Curve (AUC).. These metrics indicate strong performance in identifying diabetes risk. 2025 IEEE. -
A Comprehensive Study for Application of Blockchain Technologies for the Decentralized Grid Utilization Possibilities
This thorough research explores the world of blockchain technology and its significant effects on the use of decentralized grids. Decentralized networks have shown promise in addressing the rising need for renewable energy and effective resource management. Blockchain, a technology based on distributed ledgers, presents creative approaches to improve grid oversight, enable through peer-to- energy trade, and guarantee openness and protection in the management of the grid. This paper investigates the advantages and disadvantages of using blockchain in decentralized grids. From trading in renewable energy to grid optimization and demand response, we examine a variety of use cases and applications. We offer insight into the practical viability and scalability of blockchain-based solutions through a thorough analysis of real-world deployments and case reports. We also address the legislative and technological challenges to be solved before blockchain technology can realize its full potential in decentralized grid setups. Our research strongly emphasizes that regulatory architectures, seamless integration, and standardization all contribute to supporting the harmonious adaptation of blockchain technology worldwide. This study provides lawmakers, industry players, and investigators in the efforts to build a sustainable and effective energy future with an informative tool: it will shed light on the potential and constraints presented by blockchain technology in a decentralized environment of grid usage. 2025 IEEE. -
Enhancing Lung Cancer Detection Accuracy: Implementing Smote for Balanced Learning
This research goal is to forecast lung cancer using machine learning, and addressing the dataset's class imbalance is a top priority. The data that was initially gathered was extremely unbalanced, with 87.38% of instances being of the minority class of lung cancer and only 12.62% being non-cancer cases. To address this imbalance, minority over-sampling through self-generated SMOTE (Synthetic Minority Over-sampling Technique) was implemented wherein there were 64.85% cases of lung cancer and 35.15% of non-lung cancer cases after deduplication. Logistic regression (LR), Gaussian naive Bayes, Support Vector Machine (SVM), Bernoulli naive Bayes, K nearest neighbors (KNN), Random Forest (RF), multi-layer perceptron, and extreme gradient boosting are among the machine learning methods that were tested. The best test performance was shown by the Random Forest and Extreme Gradient Boosting methods that achieved an accuracy of 97.3% followed by K Nearest Neighbors at 95.95%, and Multi-Layer Perceptron at 93.24%. This highlights the necessity of data balance and the ways in which these methods can improve the efficacy of predictive models for lung cancer. As such, this addition contributes to the dearly needed critical knowledge which may be a stepping stone for innovation within the domains of diagnosis and treatment medicine through machine learning. 2025 IEEE. -
Advances in Type II Diabetes Prediction: A Comprehensive Review of Machine Learning Techniques
Type II diabetes mellitus, on the other hand has been regarded as one of the growing concerns globally and thus clearly raises the need for making accurate forecasts of diabetes. The risk for Type II diabetes can be predicted using Ma-chine Learning as well as any other form to make the predictions much more enhanced than the traditional methods. This paper aims to give a broad overview of literature that has so far been available on the ML algorithms used in the management of Type II diabetes including such supervised algorithms as logistic regression, alphabet regression, random forest, support vector regression along with other methods such as, ensemble learning, deep learning, and hybrid. Analysis of the main aspects for the performance model such as parameter selection, the way to face and cope with imbalance parameters, interpretability and generalizability across different populations, another aspect that was regarded is the possibility of using real-time data collected with wearable devices and applying tissue and other biomarkers for better prediction. Finally, the key obstacles and future directions towards developing ML algorithms and models explainable and clinically relevant have been introduced to help researchers and practitioners toward effective, personalized, and scalable interventions. 2025 IEEE. -
A Study on Gynecological Cancers Using Artificial Intelligence: A Revolutionary Approach
An examination of the role of artificial intelligence (AI) in ovarian, cervical, and breast cancer early detection and management is presented in this paper. Artificial intelligence (AI) can improve diagnostic accuracy, streamline treatment protocols, and facilitate personalized medicine approaches by leveraging advancements in machine learning (ML) and deep learning (DL). Various Artificial Intelligence models have demonstrated success in improving the outcomes of cancer diagnostics, including their ability to distinguish benign from malignant tumors. Technology challenges and ethical issues related to the integration of AI into clinical practice are also discussed in the review. Specifically, we want to illustrate how artificial intelligence can lead to better prognoses and reduced mortality rates for cancer patients by enhancing early detection capabilities. 2025 IEEE. -
Behavioral Analytics for Predictive Modeling of Mental Health Disorders: A Review of Machine Learning Techniques and Challenges
Mental health issues, including anxiety, stress, and depression, may remain untreated until they escalate to a severe level. The issues significantly impact an individual's overall well-being and productivity. Timely identification is crucial for the effectiveness of both intervention and therapy. The application of machine learning techniques makes behavioral analytics a powerful tool for mental health disease prediction modeling. By analyzing behavioral data, this technology facilitates the early detection of various illnesses. This work aims to provide a thorough overview of the use of machine learning techniques, including models that employ Deep structured learning as well as both unsupervised as well as supervised learning, to behavioral data, including activity levels, speech patterns, and facial movements, in order to identify signs of mental health. The benefits and drawbacks of a broad range of machine learning algorithms are examined, with a focus on how these computer algorithms may be applied to identify patterns linked to illnesses like stress, anxiety, and emotional depression. This study looks into the problems that this business encounters as well. These difficulties include combining behavioral data with extra environmental issues and physiological features from the immediate surroundings, the necessity for large and diverse datasets, the need for security of information, and the capacity to understand models. 2025 IEEE. -
Fraud Prevention in Banking: Innovative Techniques for Detecting Payment Fraud
Fraud detection in banking remains one of the most critical challenges, as fraudulent patterns continue changing to avoid detection. Classic rule-based methods provide a basis approach but often lead to high rates of false positives and negatives, which limiting their efficiency. Due to the rapid growth of fraud particularly in Banking Payments, tackling this challenge has become imperative. To this end, we employ the Banksim dataset-a synthetic tool that replicates the various payment behavior of customers- to assess a number of machine learning models, Support Vector Machines, Random Forest, Logistic Regression, AdaBoost and Decision trees. Our model evaluation, using confusion matrices and classification reports, demonstrates the ability of these approaches to provide precision and reliability in detecting fraudulent transactions. This research contributes to enhancing the reliability and integrity of banking services through fraud payment detection improvements. 2025 IEEE. -
Use of AI in Selected Financial Institutions: A Double-Edged Sword Effect
In the contemporary era, our nation is progressing towards increased integration of advanced technologies, and these innovations are influencing various facets of our daily activities. Notably, artificial intelligence (AI) has become pervasive across all sectors and functions within organizations. Its impact is particularly evident in financial institutions where AI plays a crucial role in activities such as accounting transactions and fraud detection. The relationship between AI and the concept of a 'Double Edged Sword' is explored in this paper. The primary objective is toexamine and assess the utilization of AI in diverse financial institutions and its connection to the 'Double Edged Sword' phenomenon. Additionally, the paper delves into the pros and cons following the implementation of AI in the accounting industry, incorporating insights from selected Chartered Accountants regarding their perspectives on AI application. 2025 IEEE. -
Psycho-Intelligent Dialogue Agents for Enhancing Emotional Self-Regulation in Autistic Teenagers
Autistic adolescents often experience the inability to identify their emotions and self-regulate them, thus creating the impulse for the construction of intelligent assistive technologies. Building on this premise, this work proposes a novel Psycho- Intelligent Dialogue Agent (PIDA) system, which attempts to incorporate advances in affective computing, contextualised rea-soning, and psychotherapeutic dialogues, in aiding emotional self- regulation with teenagers with autism spectrum disorder (ASD). This system integrates a visual emotion recognition model with an adaptive conversational bow. To train the emotion classifier for real time application trained using transfer learning techniques based on the VGG16 architecture of deep convolutional neural networks, it was trained on a specialised dataset comprising of autistic children's facial expressions and achieved an accuracy of 71% at a 5-emotion recognition task. The Effect recognition module serves the context-aware dialogue manager in real time adapting and personalising the emotional regulation frameworks to be employed. PIDA's dialogues are based on the principles of clinical psychotherapy, with psychotherapeutic techniques and intervention strategies which are individually tuned to the emotional state and contextual parameters of the situation. The system was designed and built salted with caregiver integration features to enable guardians to monitor progress and active participant in the personalization of the intervention. Primary experimental results reflect the feasibility of this dimension in emotional awareness and emotional regulation and coping strate- gies. To support we provide uninterrupted emotional assistance to autistic young people and offer flexible support resources during and in between emotional therapy appointments. 2026 IEEE. -
Development of a VR-Based Solid Waste Management Awareness Platform Utilizing YOLOv12 and MSCNN
Waste management is not an issue concerning an individual but a collective responsibility. It refers to our environment. Our project, 'Solid Waste Management,' verifies the efficacy of virtual reality, a novel learning modality for the user, acquiring knowledge of waste segregation and the right way of waste disposal simulation through virtual reality. The implication of virtual reality's addition into the educational system was analyzed through the acceptance of the model and the acquisition of knowledge through the task performed by the user. The simulation and the environment of virtual reality are implemented through the use of Spatial Awareness, Haptic Simulation, Hand Tracking using HCI, and Immersive Learning Environment for a genuine simulation experience for the user. The simulation environment and software were developed using the Unity environment creating a gamified world using a 3D environment and the Blender software used for the development of the 3D models. The simulation environment's implementation into the various HMD devices also used the OpenXR plugin. The simulation is further segmented into two parts: interior and exterior waste management. This novel simulation technique will not only enable the acquisition of the most requisite skills of waste segregation but also the acquisition of environment knowledge and the promotion of people towards sustainable practices. 2026 IEEE. -
Volatility-Based Stock Categorization and Risk-Informed Investment Support System
This paper presents an application that will benefit novice investors by categorizing stocks by levels of volatility so users can better understand risk by average parameters and increase factors assessed for long-term wealth generation. The offering model supports portfolio creation by enabling novice investors to choose appropriate options based on their risk tolerance, something that would be more challenging for investors with limited financial savvy under different circumstances. To develop an investment companion application that reduces emotional investing and increases strategic long-term financial decision making through effective data visualization, especially for novices. An application that uses statistics on volatility to compartmentalize stocks by low, medium and high-risk options for selection, with the ability to create and assess a portfolio based on this criteria through a virtual interface. Increased investment construction accessibility, the ability to create diverse portfolios, and a foundation for subsequent advanced investment features like notifications and trend predictive analysis. 2026 IEEE. -
Emotionally Adaptive AI Companions for Supporting Routine Management in Autistic Adolescents
Autistic Adolescents usually experience difficulties in the management of emotions, routine transitions and social cue interpretation. Many existing tools that aim to fill in the gap are often non-personalise, static or lack real-time responsiveness in handling these challenges. This study conceptualises and empirically validates a prototype of an emotionally adaptive AI companion that focuses on reducing stress due to routine transition, emotional regulation and social cue interpretation while increasing personalised management by providing contextual support. A quasi-experimental, mixed methods design is adopted. The core of this system conducts facial multimodal emotion recognition through facial expression and simulated voice tone using transfer learning across three CNN architectures (ResNet-18, MobileNetV2, and EfficientNet-B0) as comparison tests. The resulting emotion output is feeds into a contextual engine for real-time personalised interventions which can also be continuously improved through critical feedback-in-the-loop control architecture based on caregiver logs. The key model trade-offs are validated, the findings established that ResNet18 possesses the highest accuracy of 48%, EfficientNet-B0 with a superior F1 Score of 0.31 and MobileNetV2 proves to be efficient but slightly lower performance compared to other architectures. Simulated user feedback validation resulted in high preliminary acceptability, as high as 87.5% acceptability for an intervention like 'Reassurance'. This validated the utility of this responsive system. This transfer-learning based, multi-modal pipeline is robust. The results of the comparative analysis uncovered a very profound and instructive trade-off between the complexity of models, their efficiency, and performance metrics relating to accuracy versus the F1-score. 2026 IEEE.