May 2, 2025
Conference Paper

PQML: Enabling the Predictive Reproducibility on NISQ Machines for Quantum ML Applications

Abstract

Quantum computing provides a new paradigm for high-performance computing. In the last few years, we have seen the evolution of quantum computers from 1 qubit processor to over 400 qubits processor. With a high number of qubits in a processor comes advantages such as faster computational ability, which is impossible with classical computing. However, noise present in all quantum computer processors poses a challenge for researchers since it affects the outcome of experiments. Reproducibility in classical computing is when we can reproduce identical results when using an algorithm on identical or very similar hardware systems. However, reproducibility in quantum computing is highly affected by the noise stemming from quantum processors. This eventually creates challenges in testing quantum computing applications since it brings high variance in the test accuracies of an application. Our software tool helps us predict the test accuracy of two quantum computers running QML algorithms. Using our PQML tool, users can accurately predict the test accuracy of a quantum computer running a quantum machine learning algorithm. To the best of our knowledge, this is the first study that targets reproducibility in quantum computing, especially in quantum machine learning. To the best of our knowledge, PQML is a first-of-its-kind tool that helps predict QML applications' performance running on NISQ machines.

Published: May 2, 2025

Citation

Senapati P., S. Chen, B. Fang, T.M. Athawale, A. Li, W. Jiang, and Z. Wang, et al. 2024. PQML: Enabling the Predictive Reproducibility on NISQ Machines for Quantum ML Applications. In IEEE International Conference on Quantum Computing and Engineering (QCE 2024), September 15-20. 2024, Montreal, Canada, 1413-142. Piscataway, New Jersey:IEEE. PNNL-SA-180185. doi:10.1109/QCE60285.2024.00168