Quantum sensing-computational advantage using hybrid quantum sensing-processors

We explain a kind of quantum advantage that can arise by combining quantum sensing with quantum computing: a quantum sensing-computational advantage (QSCA). Quantum sensors are typically optimized to estimate one or more unknown parameters, and those parameter estimates might then be used in downstream tasks requiring postprocessing, including complex practical tasks such as classification or forecasting. Measurements of quantum sensors are fundamentally noisy, and as a result any classically postprocessed quantities are also noisy. In this work, we show several examples illustrating how it is possible to reduce the error in downstream tasks by performing some of the computation on the sensed parameters within the quantum system prior to measurement, using algorithmic techniques such as quantum signal processing (QSP) and quantum neural networks (QNNs). For machine-learning classification tasks where processing is performed using QNNs, one can think of the combined quantum sensor and quantum processor as a quantum smart sensor or quantum computational sensor whose goal is not to directly output sensed parameters but decisions about those parameters. We show how a quantum sensing-computational advantage can be obtained both with entangled and with unentangled quantum systems, including ones comprising just a single qubit or qumode.