Curator's Take
This research reveals a fascinating counterintuitive finding that operational noise can actually improve quantum kernel performance by increasing model expressivity and complexity, challenging the conventional wisdom that noise is always detrimental in quantum computing. The work extends quantum kernel methods beyond traditional gate-based circuits to analog quantum systems, potentially opening new pathways for near-term quantum machine learning applications on current noisy devices. Particularly intriguing is the practical application to estimating non-Markovianity from sparse data, suggesting these noise-enhanced kernels could tackle real-world problems that are challenging for classical methods. This represents a promising shift in perspective where researchers can potentially harness unavoidable noise as a feature rather than fighting it as a bug.
— Mark Eatherly
Summary
The quantum kernel method, a promising quantum machine learning algorithm, possesses substantial potential for demonstrating quantum advantage. Although the majority of the quantum kernel is constructed in the context of gate-based quantum circuits, inspired by the idea of analog quantum computing, here we construct an analog quantum kernel and a hybrid quantum kernel, and show their competitiveness against other kernel methods in a benchmarking task and the practical problem of estimating non-Markovianity from sparse data. Additionally, we also incorporate operational noise into the quantum kernels. Our results reveal that the presence of operational noise can be beneficial to the performance of the developed quantum kernels. We attribute this counterintuitive noise-enhanced performance to the improved expressivity and higher model complexity induced by noise. These results pave the way for practical implementations of quantum kernel methods and provide an efficient approach for estimating non-Markovianity with reduced experimental demands.