Automated experimental masking of qubit states on arbitrary disks via machine learning
Phys. Rev. A 114, 032423 – Published 9 September, 2026
DOI: https://doi.org/10.1103/kmd4-t22s
Abstract
Quantum information masking (QIM) encodes single-qubit information into bipartite entanglement, but the absence of a universal masker restricts standard bipartite implementations to specific Bloch sphere disks. Here, we experimentally demonstrate an automated quantum masking machine that can be reconfigured for target disks of different orientations. Driven by a machine-learning-assisted feedback algorithm, a self-calibrating adapter determines the required rotation directly from photon-count measurements for a fixed base masker and maps a selected target disk to a maskable latitudinal configuration. Leveraging this framework, we experimentally demonstrate a proof-of-principle quantum secret sharing scheme. For the tested states, quantum state tomography shows that the encoded information is retained in the global entanglement, while the corresponding marginal states exhibit no resolvable input dependence within experimental uncertainty. Rather than altering the no-masking principle, this closed-loop architecture broadens the input compatibility of a fixed masker and provides a reconfigurable platform for masking-based quantum communication.
