AI and the World's Hard Problems · Quantum physics & quantum computing · arXiv · 2026-10-23
Can Entanglement-Induced Feature-Map Designs Improve Projected Quantum Kernels?
Projected quantum kernels (PQKs) often rely on extensive classical tuning of the RBF kernel and SVM regularization parameters, which can mask the native power of their quantum feature maps. The study compares untuned and tuned PQKs against an RBF baseline to isolate their inherent inductive bias. The proposed two-qubit entanglement feature maps, designed to capture multivariate correlations in IoT data, show a significantly stronger native inductive bias.