Researchers have investigated the survival of geometric information within a four-qubit quantum kernel on IBM Quantum Hardware. The study focused on a specific frozen ZZ feature-map kernel, analyzing its performance across three execution configurations: baseline, dynamical decoupling, and gate twirling. Results indicated that while the Gram matrix generally survived execution and preserved statevector geometry to a significant degree, gate twirling proved to be the most faithful method. However, the most faithful configuration exhibited the lowest alignment with the target labels, suggesting a disconnect between implementation fidelity and task relevance in current quantum machine learning studies. AI
RANK_REASON Academic paper detailing experimental results on quantum hardware. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
