IonQ says quantum generative model beats classical baselines in high-resolution SAR change detection tests
IONQ•IonQ said its quantum generative model outscored two classical baselines on non-Gaussian SAR tests, with filtered F1 scores of 0.41 versus 0.24 and 0.16. The advantage narrowed on roughly Gaussian data, while results on a volcanic lava flow InSAR test were comparable.
1. SAR test results
IonQ released research showing a Quantum Circuit Born Machine improved change detection on high-resolution SAR and InSAR satellite radar data. On non-Gaussian SAR tests, it scored a filtered F1 of 0.41, compared with 0.24 and 0.16 for two classical baselines.
2. Performance across data
The performance edge narrowed when preprocessing made pixel distributions roughly Gaussian, pointing to gains where statistics are sparse or complex. Results held when the simulation model ran on an IonQ Forte Enterprise system. In a volcanic lava flow InSAR test, quantum and classical methods delivered comparable peak results.
