Pattern Recognition with Deep Learning in Quantum Materials*

The capabilities of surface probe experiments are rapidly expanding, providing views on quantum materials at unprecedented length and time scales. Many such materials display intricate pattern formation in the electronic properties on the observable surface. This rich spatial information contains information about interactions, dimensionality, and disorder.[1] A well-tuned machine learning framework can decipher this information with minimal effort from the user.[2] We show the effectiveness of our deep learning framework on simulations of statistical models. We then use our machine learning model to analyze experimental data from an optical microscope[3] on a vanadium dioxide film as it goes through the insulator-metal transition.
[1] B. Phillabaum, et al. Nat Commun 3, 915(2012).
[2] L. Burzawa, et al. Phys. Rev. Materials 3, 033805(2019).
[3] A. Zimmers, et al. Phys. Rev. Lett. 110, 056601(2013).