In this article we apply the random forest machine learning model to classify 1D topological phases when strong disorder is present. We show that using the entanglement spectrum as training features the model gives high classification accuracy. The trained model can be extended to other regions in phase space, and even to other symmetry classes on which it was not trained and still provides accurate results. After performing a detailed analysis of the trained model we find that its dominant classification criteria captures degeneracy in the entanglement spectrum.
@article{arxiv.2001.10565,
title = {Classification of Strongly Disordered Topological Wires Using Machine Learning},
author = {Ye Zhuang and Luiz H. Santos and Taylor L. Hughes},
journal= {arXiv preprint arXiv:2001.10565},
year = {2020}
}
Comments
7 pages, 7 figures, submitted for a memorial article in a special issue of SPIN for Shou-Cheng Zhang