Reproduce “Classification of Financial Data Using Quantum Support Vector Machine” on OpenResearch

About this paper

Seemanta Bhattacharjee, MD. Muhtasim Fuad, A.K.M. Fakhrul Hossain

Researchers from Shahjalal University Of Science and Technology applied quantum kernel methods to classify daily changes in the Dhaka Stock Exchange Broad Index, finding that quantum kernels, particularly one based on the Pauli YYY feature map, consistently outperformed a classical RBF-kernel Support Vector Machine baseline. The study also provided resource estimations for scaling these quantum financial classification models.

Key insights

• Quantum kernels, particularly those built with the Pauli YYY feature map, can empirically outperform classical RBF-kernel SVMs in classifying complex, real-world financial data, even with small dataset sizes.
• The Phase Space Terrain Ruggedness Index (PTRI) framework can help identify problem instances where quantum kernels are likely to offer an empirical advantage over classical methods.
• Resource requirements for quantum kernel-based classification, including qubit count, gate count, and circuit depth, can be estimated, providing practical guidance for scaling quantum machine learning experiments in finance.

Summary by alphaXiv.