This paper proposes the hybrid framework of privacy preserving that combines the concept of federated learning and homomorphic encryption with differential privacy, to address the privacy issue of collaborative machine learning for healthcare application. The proposed approach makes three contributions: (1) multi-layered architecture using federated learning in combination homomorphic encryption (based on CKKS scheme) and differential privacy that offers defense against inference attacks at different layers, (2) the implementation which alleviates the computational overhead compared to homomorphic encryption only with optimised cryptographic parameters, and (3) the application of the Grasshopper-Black Hole Optimization (G-BHO) for the optimisation of privacy parameters (e, deltas, gradient clipping thresholds) in order to balance the privacy-utility trade-off. Cryptographic keys are produced using the principles of cryptographically secure random number generation. Experimental evaluation on two healthcare data sets (MIMIC-III and chest X rays of the patients of Covid-19) to compare the hybrid approach to the single technique baselines in four metrics: classification accuracy (93.0±1.2% vs. 89.0±1.5% for federated learning only), differential privacy guarantee (ε=0.7, δ=10⁻⁵), computational overhead (2.5x baseline vs. 8x for homomorphic encryption only) and the resistance to membership inference attacks (92% vs. 68%) The observed improvement in the accuracy is unexpected, and potentially a consequence of side-effects due to the effects of the regularization in the differential privacy noise; this finding needs to be further explored in theory. The evaluation is restricted to the tasks of healthcare classification, while generalization to other domains needs more validation. The main contribution is an empirical proof that by using a combination of several privacy mechanisms, it will be possible to achieve a stronger attack resistance with a lower computational overhead than by using homomorphic encryption alone. 

The rapid growth of mobile wallet usage has led to a sharp increase in fraudulent transactions, making fraud detection in portable wallets a pressing concern. Accurately detecting fraud is difficult because transaction data is complicated and unbalanced. Conventional rule-based systems are less flexible and frequently provide large false positive rates along with poor accuracy. Effective feature selection is crucial to the performance of Machine Learning (ML) models, notwithstanding their increased detection rates. Redundancy and noise are introduced by high-dimensional data, which lowers model performance and raises computing costs. The advantages of hybrid feature selection are frequently overlooked in current research, particularly when it comes to portable wallet fraud detection. By combining Random Forest Importance, LASSO Regression, Recursive Feature Elimination (RFE), and Mutual Information (MI) with resampling to solve class imbalance, this study fills that gap. Our approach provides a more reliable and effective solution for safe portable wallet fraud detection by removing superfluous features, increasing accuracy, and reducing computing cost. The model becomes faster and more effective when superfluous characteristics are eliminated because this reduces the computational effort. By concentrating just on the most instructive data, it increases accuracy. By addressing class imbalance and combining several selection strategies, the hybrid approach guarantees robustness. All things considered, this leads to a scalable and safe fraud detection system for transactions using mobile wallets. Our results show that a successful feature selection approach improves fraud detection accuracy, which in turn improves operational effectiveness and financial security. 

The rapid rise of mobile technology paired with the steady growth of the internet, has led to a massive increase in the amount of user generated content, such as online consumer reviews, accessible through the browser. As the volume of user-generated content continues to rise, it becomes increasingly important to develop sophisticated methods for performing sentiment analysis on the texts collected from users, especially those that have been generated in relation to restaurants and similar types of service establishments. In this paper, we will present a new approach to sentiment analysis which incorporates Latent Dirichlet Allocation topic models, Term Frequency- Inverse Document Frequency vector representations and XGBoost Classifiers into a unified framework. Unlike conventional implementations, this study integrates probabilistic topic distributions from LDA with multi-level n-gram TF-IDF features and evaluates their combined impact using XGBoost for enhanced classification performance. Using three distinct n-gram levels (unigrams, bigrams, and trigrams), we will evaluate various aspects of text-based data including common linguistic patterns and sentiment trends. Higher-order n-grams were included to capture contextual dependencies beyond single-word features. Overall, our results demonstrate that the performance of our proposed framework is superior to traditional corpus-based models on multiple evaluation metrics, including: classification accuracy 96.07%, classification sensitivity 95.43%, classification specificity 97.12% and F1-Score 96.16%. 

Lung cancer is responsible for many deaths from cancer around the globe, primarily because it is difficult to find malignant lung nodules early enough to be treatable. We developed a hybrid deep learning approach to the automated classification of lung nodules from chest computed tomography (CT) images. Our model uses convolutional neural networks (CNNs) for hierarchical feature extraction, an attention mechanism for feature refinement in targeted regions of interest, and a support vector machine (SVM) classifier for robust margin-based decision making. Furthermore, we use a patch-based learning strategy within the model to improve sensitivity to small and ambiguous lung nodules. The model is tested on the publicly available LIDC-IDRI dataset and achieves 94.2% accuracy, 95.1% recall, and an area under the receiver operating characteristic curve (AUC-ROC) score of 0.971, which outperforms multiple baseline deep learning methods. The proposed method provides a synergistic integration of attention-weighted feature enhancements and traditional machine learning classifications as compared to traditional end-to-end architectures, resulting in improved model generalization and interpretability. Grad-CAM visualizations are also used to provide qualitative insights into the model decision-making process. The proposed hybrid approach provides a novel and interpretable solution for the classification of lung nodules from CT images that may assist in the development of computerized systems to assist physicians in making diagnoses using medical images.

This paper presents the implementation and evaluation of a Multi-key Multi-modalities Biometric Encryption System designed for business enterprises, leveraging cloud storage for secure and scalable data management. The system integrates multiple biometric modalities fingerprint, iris scan, and face recognition to enhance data security through advanced multi-key encryption techniques, utilizing algorithms such as Advanced Encryption Standard (AES) and Rivest-Shamir-Adleman (RSA). The encrypted biometric data is securely stored in the cloud, providing enterprises with efficient storage solutions. The system's performance was evaluated across several parameters including encryption/decryption time, biometric match accuracy, data transfer speeds, energy consumption, cost, and user satisfaction. The results demonstrate that multi-modal systems offer superior accuracy and security compared to single-modality systems, reducing error rates and enhancing reliability. However, multi-modal authentication incurs higher costs, energy consumption, and slightly longer processing times. Despite these trade-offs, the system achieved high user satisfaction, particularly in high-security environments where data protection is a priority. The findings indicate that the proposed system is a viable solution for businesses seeking a secure, scalable, and efficient method of protecting sensitive data.

Suicide remains a critical global public health issue, claiming vast number of lives each year. Traditional assessment methods, often reliant on subjective evaluations, have limited effectiveness. This study examines the potential of Bidirectional Encoder Representations from Transformers (BERT) in revolutionizing suicide risk prediction by extracting textual biomarkers from relevant data. The research focuses on the efficacy of BERT in classifying suicide-related text data and introduces a novel BERT-based approach that achieves state-of-the-art accuracy, surpassing 97%. These findings highlight BERT's exceptional capability in handling complex text classification tasks, suggesting broad applicability in mental healthcare. The application of Artificial Intelligence (AI) in mental health poses unique challenges, including the absence of established biological markers for suicide risk and the dependence on subjective data, which necessitates careful consideration of potential biases in training datasets. Additionally, ethical considerations surrounding data privacy and responsible AI development are paramount. This study emphasizes the substantial potential of BERT and similar Natural Language Processing (NLP) techniques to significantly improve the accuracy and effectiveness of suicide risk prediction, paving the way for enhanced early detection and intervention strategies. The research acknowledges the inherent limitations of AI-based approaches and stresses the importance of ongoing efforts to address these issues, ensuring ethical and responsible AI application in mental health.

This research paper explores Blockchain (BC) technology-based identity verification's role in streamlining and securing the employee onboarding process within Human Resource (HR) management. It addresses this technology's potential benefits, challenges, and limitations in enhancing HR practices. This study is grounded in the theoretical foundation of BC technology and its applications. It examines existing identity verification systems in HR management and delves into the potential implications of adopting BC-based solutions. This research employs a comprehensive design encompassing a discussion of the background, research problem, objectives, and significance. A detailed overview of BC technology and its applications and an analysis of existing identity verification systems are presented. The study employs a well-defined research design, including a sampling strategy, sample size determination, data collection methods, and data analysis techniques. The study's findings reveal that BC-based identity verification has the potential to streamline and secure the employee onboarding process in HR management. However, the investigation also identified scalability, interoperability, and data security challenges. These findings contribute to understanding the feasibility of adopting BC technology in HR practices. The study informs HR managers and BC developers on the potential benefits and hurdles of implementing BC-based identity verification, enabling them to make informed decisions.

One of the most challenging tasks in the education sector in India is to predict student's academic performance due to a huge volume of student data. In the Indian context, we don't have any existing system by which analyzing and monitoring can be done to check the progress and performance of the student mostly in Higher education system. Every institution has their own criteria for analyzing the performance of the students. The reason for this happing is due to the lack of study on existing prediction techniques and hence to find the best prediction methodology for predicting the student academics progress and performance. Another important reason is the lack in investigating the suitable factors which affect the academic performance and achievement of the student in particular course. So to deeply understand the problem, a detail literature survey on predicting student’s performance using data mining techniques is proposed. The main objective of this article is to provide a great knowledge and understanding of different data mining techniques which have been used to predict the student progress and performance and hence how these prediction techniques help to find the most important student attribute for prediction. Actually, we want to improve the performance of the student in academic by using best data mining techniques. At last, it could also provide some benefits for faculties, students, educators and management of the institution.

This paper highlights important issues of higher education system such as predicting student’s academic performance. This is trivial to study predominantly from the point of view of the institutional administration, management, different stakeholder, faculty, students as well as parents. For making analysis on the student data we selected algorithms like Decision Tree, Naive Bayes, Random Forest, PART and Bayes Network with three most important techniques such as 10-fold cross-validation, percentage split (74%) and training set. After performing analysis on different metrics (Time to build Classifier, Mean Absolute Error, Root Mean Squared Error, Relative Absolute Error, Root Relative Squared Error, Precision, Recall, F-Measure, ROC Area) by different data mining algorithm, we are able to find which algorithm is performing better than other on the student dataset in hand, so that we are able to make a guideline for future improvement in student performance in education. According to analysis of student dataset we found that Random Forest algorithm gave the best result as compared to another algorithm with Recall value approximately equal to one. The analysis of different data mini g algorithm gave an in-depth awareness about how these algorithms predict student the performance of different student and enhance their skill.

Educational Data Mining (EDM) is one of the crucial application areas of data mining which helps in predicting educational dropout and hence provides timely help to students. In Indian context, predicting educational dropouts is a major problem. By implementing EDM, we can predict the learning habits of the student. At present EDM has not been introduced at higher education level. Due to this we cannot recognize the genuine problems of students during their education. The objective of this analysis is to find the existing gaps in predicting educational dropout and find the missing attributes if any, which my further contribute for better prediction. After that we try to find the best attributes and DM techniques which are frequently used for dropout prediction. Based on the combination of missing attribute and best attribute of student data thus far, a new algorithm can be tested which may overcome the shortcomings of previous work done.