Quantum Kernel-Enhanced Support Vector Machines for Financial Fraud Detection: A Benchmarking Study on NISQ Simulators Against Classical Baselines

Key Points

• This research aims to assess the effectiveness of quantum kernel-enhanced support vector machines for detecting financial fraud compared to classical methods.
• Utilized quantum kernel methods for support vector machines on NISQ simulators
• Applied classical baseline methods including PCA and LDA
• Evaluated kernel alignment scores using depolarizing noise models.
• Quantum kernel methods improved detection rates over classical methods
• Notably higher kernel alignment scores were achieved
• Demonstrated robustness against depolarizing noise.