Quantum Machine Learning Approaches For Large-Scale Credit Risk And Fraud Detection

Authors: Dr. Pankaj Malik, Adarsh Mishra, Diya Agrawal, Dushyant Pratap Singh, Garvita Gangwal, Aniruddh Sharma

Abstract: The increasing complexity and volume of financial transactions demand advanced analytical methods capable of ensuring accurate and real-time credit risk evaluation and fraud detection. Traditional machine learning models, though effective, struggle with scalability and computational efficiency when handling large, high-dimensional financial datasets. This paper presents a Quantum Machine Learning (QML) framework integrating Quantum Support Vector Machines (QSVM) and Quantum Neural Networks (QNN) to enhance predictive accuracy and reduce training time in large-scale credit risk and fraud detection tasks. The proposed hybrid quantum–classical model leverages quantum parallelism to efficiently process nonlinear patterns and entangled relationships within financial data. Experimental results on benchmark credit scoring and fraud detection datasets demonstrate that the proposed QML framework achieves a 9.6% improvement in detection accuracy and a 17.3% reduction in computation time compared to conventional deep learning models. Additionally, the quantum-enhanced approach exhibits higher precision in identifying minority fraud cases, reducing false negatives by 12.5%. The findings confirm that QML provides a promising pathway for scalable, high-performance financial analytics, enabling faster and more reliable decision-making in credit risk management and fraud prevention. Future research will focus on optimizing quantum circuit depth and extending the framework for real-time deployment on Noisy Intermediate-Scale Quantum (NISQ) devices.