aiqbench analysis
This study experimentally validates quantum kernel learning (QKL) on liquid-state NMR hardware, demonstrating its ability to efficiently process quantum data without expensive tomography protocols. By benchmarking QKL against classical methods for regression and classification tasks, the research confirms that encoding feature maps into Hilbert spaces offers a practical advantage for capturing inherent structures in operator space. The work effectively bridges theoretical machine learning concepts with native quantum implementation using 3-qubit registers to handle both entangling and non-entangling operators.
Experimentally Extending Quantum Kernel Learning to Quantum Data by NMR
arXiv:2412.09557v3 Announce Type: replace-cross Abstract: Quantum kernel learning (QKL) promises efficient machine learning by encoding feature maps onto exponentially large Hilbert spaces inherent in quantum systems. Using the liquid-state nuclear magnetic resonance (NMR) platform, we implement and benchmark QKL for one-dimensional regression and two-dimensional classification tasks. We then classify entangling and non-entangling operators by extending QKL to handle parametrized or non-parameterized operator inputs. We first compute the kernel numerically for a double-layered star system and then experimentally validate it on a 3-qubit NMR register. QKL provides a practical route to compare operators on native quantum hardware without expensive tomography protocols. Our results confirm the superiority of QKL over other classical methods for processing quantum data, thereby highlighting its ability to capture the inherent structure of quantum space and to extend its domain of operation beyond the training domain by exploiting symmetries in the operator space.
Abstract:Quantum kernel learning (QKL) promises efficient machine learning by encoding feature maps onto exponentially large Hilbert spaces inherent in quantum systems. Using the liquid-state nuclear magnetic resonance (NMR) platform, we implement and benchmark QKL for one-dimensional regression and two-dimensional classification tasks. We then classify entangling and non-entangling operators by extending QKL to handle parametrized or non-parameterized operator inputs. We first compute the kernel numerically for a double-layered star system and then experimentally validate it on a 3-qubit NMR register. QKL provides a practical route to compare operators on native quantum hardware without expensive tomography protocols. Our results confirm the superiority of QKL over other classical methods for processing quantum data, thereby highlighting its ability to capture the inherent structure of quantum space and to extend its domain of operation beyond the training domain by exploiting symmetries in the operator space.
Quantum Physics (quant-ph); Machine Learning (cs.LG); Applied Physics (physics.app-ph)
arXiv:2412.09557 [quant-ph]
(or arXiv:2412.09557v3 [quant-ph] for this version)
https://doi.org/10.48550/arXiv.2412.09557 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 238 words
Related coverage
Hybrid quantum-neural network beats classical machine learning
Quantum Zeitgeist · Aug 17, 2026
Researchers Predict Quantum Evolution Using Gaussian Processes
Quantum Zeitgeist · Sep 7, 2026
Qutrit-Based Neural Quantum Kernels for Classification Tasks
arXiv · Jul 28, 2026
Synthesis Superintelligence: from Semiconductors to Superconductors — Periodic Labs’ Liam Fedus and Ekin Dogus Cubuk
Latent Space · Oct 8, 2026