AI and the World's Hard Problems · Quantum physics & quantum computing · arXiv · 2026-10-20
Prototype-Guided Optimization of Quantum Data Embeddings with Linear Cost
In supervised quantum machine learning, optimizing quantum data embeddings is costly because existing methods evaluate fidelity between every pair of samples, scaling as O(B²). The authors propose a prototype-guided objective that needs only O(B) fidelity evaluations and integrate it into a pipeline that also optimizes the circuit architecture. Across five benchmarks with up to 20 qubits, it stays competitive with quantum kernel SVMs while cutting evaluation cost, with notable gains over pairwise optimization on T-MNIST.