There has been an effect, termed proximity effect, that can lead to spintronics without energy dissipation or qubits with error tolerance. However, identifying proximity effect has been nontrivial: the effect extends about ~1 nanometer space, comparable to instrumental resolution for neutron reflectometry. In our work, we built a comprehensive machine learning framework and showed it can improve the resolution by a factor of 2, which allows a more conclusive identification of the proximity effect. The work can also be used to identify other nuance spectroscopic signals.
