Deep learning approach to genome of two-dimensional materials with flat electronic bands

Electron-electron correlations play central role in condensed
matter physics, governing phenomena from superconductivity
to magnetism and numerous technological
applications. Two-dimensional (2D) materials with
flat electronic bands provide natural playground to explore
interaction-driven physics, thanks to their highly
localized electrons. The search for 2D flat band materials
has attracted intensive efforts, especially now with
open science databases encompassing thousands of materials
with computed electronic bands. Here we automate
the otherwise daunting task of materials search and
classification by combining supervised and unsupervised
machine learning algorithms. To this end, convolutional
neural network was employed to identify 2D flat band
materials, which were then subjected to symmetry-based
analysis using a bilayer unsupervised learning algorithm.
Such hybrid approach of exploring materials databases
allowed us to construct a genome of 2D materials hosting
flat bands and to reveal material classes outside the
known flat band paradigms.