analiza aiqbench
The study finds that tuning the classical RBF kernel's bandwidth and the quantum feature map's data scaling matters far more than any quantum-classical difference. On four real datasets, tuned quantum kernels are statistically indistinguishable from tuned RBF kernels, and the synthetic-data advantage appears mainly on data generated by the feature map itself. For readers assessing quantum machine learning claims, the practical lesson is to check protocol controls, per-fold recomputation, and circuit-evaluation cost before trusting reported speedups or accuracy gains.
Empirical Evaluation of Quantum Kernel Methods for Classification: A Comparative Study with Classical Kernels
arXiv:2610.42029v1 Announce Type: cross Abstract: Quantum kernel methods are the most concrete near-term proposal for quantum machine learning on classical data, yet published benchmarks disagree on whether they outperform tuned classical kernels, and the disagreement cannot be resolved from the literature because the comparisons are not protocol-controlled. We argue that the obstacle is one of measurement, not theory: the usual setup fixes the quantum feature map's data scaling while grid-searching the classical kernel's hyperparameters, recomputes nothing per fold, and reports cost in simulator time. We therefore treat the comparison itself as the object of study and impose three controls. The bandwidth is tuned on the same grid budget as the classical $\gamma$, the kernel is recomputed inside every fold from fold-fitted preprocessing, and cost is counted in hardware circuit evaluations. On top of these we run a factorial design of 324 cells in which dimensionality, training size, class separability and data structure are explicit covariates, together with four standard datasets, a shots-versus-gate-noise ablation, and a cost sweep. The dominant effect we measure is not quantum: tuning the feature map's bandwidth is worth 0.120 accuracy on synthetic data and 0.339 on real data, exceeding every quantum-classical difference we observe. Tuned quantum kernels are then statistically indistinguishable from a tuned RBF on all four real datasets. The advantage on synthetic data is confined almost entirely to data generated by the feature map itself, and it shrinks with dimensionality (-0.0045 per qubit) while growing with sample size. Finite sampling costs an order of magnitude more accuracy than gate and readout noise, and a fidelity kernel needs $5\times10^{7}$ circuits at n = 10,000. Protocol variance is thus sufficient to explain the field's contradictory conclusions.
Abstract:Quantum kernel methods are the most concrete near-term proposal for quantum machine learning on classical data, yet published benchmarks disagree on whether they outperform tuned classical kernels, and the disagreement cannot be resolved from the literature because the comparisons are not protocol-controlled. We argue that the obstacle is one of measurement, not theory: the usual setup fixes the quantum feature map's data scaling while grid-searching the classical kernel's hyperparameters, recomputes nothing per fold, and reports cost in simulator time. We therefore treat the comparison itself as the object of study and impose three controls. The bandwidth is tuned on the same grid budget as the classical $\gamma$, the kernel is recomputed inside every fold from fold-fitted preprocessing, and cost is counted in hardware circuit evaluations. On top of these we run a factorial design of 324 cells in which dimensionality, training size, class separability and data structure are explicit covariates, together with four standard datasets, a shots-versus-gate-noise ablation, and a cost sweep. The dominant effect we measure is not quantum: tuning the feature map's bandwidth is worth 0.120 accuracy on synthetic data and 0.339 on real data, exceeding every quantum-classical difference we observe. Tuned quantum kernels are then statistically indistinguishable from a tuned RBF on all four real datasets. The advantage on synthetic data is confined almost entirely to data generated by the feature map itself, and it shrinks with dimensionality (-0.0045 per qubit) while growing with sample size. Finite sampling costs an order of magnitude more accuracy than gate and readout noise, and a fidelity kernel needs $5\times10^{7}$ circuits at n = 10,000. Protocol variance is thus sufficient to explain the field's contradictory conclusions.
Quantum Physics (quant-ph); Machine Learning (cs.LG)
arXiv:2610.42029 [quant-ph]
(or arXiv:2610.42029v1 [quant-ph] for this version)
https://doi.org/10.48550/arXiv.2610.42029 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
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.
Text integral sub licența deschisă a publicației · 372 cuvinte
Subiecte conexe
Large Classification-Risk-Optional Label Acquisition
arXiv · 9 sept. 2026
Leveraging Cell-level Spatial Transcriptomics as Weak Supervision for Nuclei Classification
arXiv · 19 oct. 2026
WiMi’s Quantum Kernel Convolution Runs on NISQ Devices Now
Quantum Zeitgeist · 15 iun. 2026
Dimensionality Reduction for Cyberattack Classification: A Comparative Evaluation of PCA and Linear Predictive Coding
arXiv · 6 iun. 2026