AI and the World's Hard Problems · Quantum physics & quantum computing · arXiv · 2026-10-19
Task-aligned spectral inductive biases in parameterized quantum circuits
The study examines how parameterized quantum circuits can acquire task-aligned inductive biases from known spectral task structure. In data reuploading circuits, encoding a fixed, task-defined spectral query changes the hypothesis class by modulating how existing signal frequencies are combined. In small-scale spectral and radar-like tasks, including a classification task inspired by synthetic aperture radar, circuits with aligned queries learned better from limited data than parameter-matched circuits with mismatched or no queries.