IonQ preprint reports radar change-detection gains on quantum hardware

IonQ researchers report that a quantum model detected changes in satellite radar images more accurately than two classical methods on a test dataset. In an arXiv preprint, they describe running training and inference on a 20-qubit trapped-ion processor. On unsmoothed images of Marine Corps Air Station Miramar, the hardware run reached a filtered F1 score of 0.37, compared with 0.24 and 0.16 for the classical baselines. The model generates reference samples from pairs of radar images to estimate a background for change detection without smoothing away fine detail. On a separate volcanic-lava dataset, it matched the classical peak score of approximately 0.66 rather than exceeding it.