Sri Vasavi Engineering College

This study investigates how laser processing parameters specifically power, powder feeding rate, and scanning speed affect the performance of NiCrBSi/60 wt

With the increasing prevalence of machine learning applications in financial data analysis, safeguarding the privacy of customers’ personal financial information during credit risk assessment is paramount. Despite ongoing research efforts in this field, establishing robust privacy-preserving credit risk prediction systems capable of mitigating diverse privacy attacks remains a formidable challenge. To address this issue, we propose a Privacy-Preserving Credit Risk Analysis (PPCRA) framework that leverages homomorphic encryption-aware Machine Learning (ML) on encrypted data. Various ML and Privacy-Preserving Machine Learning (PPML) models were built using the TenSEAL and Concrete ML on datasets from Germany, Taiwan, Japan, and Australia. When comparing PPML with traditional ML models, it is evident that PPML achieves a considerable level of privacy preservation with only minimal loss in accuracy. The experimental results indicate that Privacy-Preserving Logistic Regression (PPLR) outperformed other PPML models. Furthermore, the security analysis demonstrates that the proposed system effectively withstands multiple privacy threats, including poisoning, membership inference, evasion, model extraction, and model inversion attacks, across various stages of the machine learning lifecycle.

Quantum Machine Learning (QML) has emerged as a transformative paradigm for predictive modeling in healthcare, particularly in scenarios where traditional methods struggle with noisy, high-dimensional, and complex data. This paper proposes a quantum-assisted healthcare framework that deploys Quantum Support Vector Machines (QSVM) and Quantum Neural Networks (QNN) within a secure cloud environment to enhance predictive accuracy in medical applications. The study critically evaluates QSVM performance against classical Support Vector Machines (SVM) and QNNs, employing a hybrid quantum-classical pipeline that integrates dimensionality reduction (PCA), feature scaling, and quantum feature mapping. Experiments were conducted on benchmark datasets including Diabetes, Prostate Cancer, Breast Cancer, Wine, and IRIS. Results show that QSVMs demonstrate effectiveness in select high-dimensional noisy scenarios but remain inconsistent and often underperform compared to classical SVMs. In contrast, QNNs exhibit superior adaptability to non-linear medical data, consistently outperform QSVMs and achieving competitive or better accuracy than classical baselines. Beyond benchmarking, the paper introduces a system model for healthcare applications involving hospitals, cloud servers, and patients. Performance evaluation highlights QNN's robustness, with results confirming up to 100% accuracy on IRIS, 99% recall on Breast Cancer, and balanced improvements across all metrics-averaging 88-90% in precision, recall, sensitivity, specificity, and F1-score-establishing QNN as a reliable and scalable approach for biomedical predictive analytics.

ABSTRACT The integration of renewable energy sources and smart technologies has enhanced the efficiency of electrical grids but introduced new cybersecurity vulnerabilities. This study presents a novel privacy‐preserving machine learning framework that combines Random Vector Functional Link (RVFL) networks with Fully Homomorphic Encryption (FHE) for smart grid stability prediction. The framework enables end‐to‐end encrypted training and inference, ensuring data confidentiality throughout the ML pipeline. Experimental results using an enhanced smart grid dataset demonstrate that the standard RVFL model achieves 96.49% accuracy and an AUC of 1.0, while the encrypted RVFL(FHE) model attains 92.20% accuracy and an AUC of 0.92. The encryption introduces approximately 4.5× higher training time and increased memory consumption but offers significant privacy protection. This trade‐off is essential in scenarios where data sensitivity outweighs minor performance drops. The proposed method is applicable to real‐world smart grid systems, especially where privacy is critical, such as consumer load forecasting or demand response programs. Future work includes optimizing computational efficiency for edge deployment and validating the model in large‐scale, distributed smart grid environments.

This manuscript introduced a novel five-level inverter which employs eleven switches and two capacitors, driven through a level-shifted pulse-width modulation strategy. Owing to the inherent characteristics of the control scheme, the voltages across the floating capacitors remain naturally balanced, eliminating the need for auxiliary balancing circuits. Furthermore, six of the eleven switches experience only 25 % of the supply voltage, significantly reducing their voltage rating requirements and contributing to overall cost reduction. Comparative evaluation with existing topologies highlights the advantages of the proposed structure in terms of reduced voltage stress, component count, and performance. The effectiveness of the topology has been validated through both MATLAB/Simulink and hardware prototype. These results confirm that the proposed inverter is a cost-effective, reliable, and practical solution for modern grid-connected power conversion systems.