Deep learning in the heterotic orbifold landscape

We use deep autoencoder neural networks to draw a chart of the heterotic ${\mathbb{Z}}_6$-II orbifold landscape. Even though the autoencoder is trained without knowing the phenomenological properties of the ${\mathbb{Z}}_6$-II orbifold models, we are able to identify fertile islands in this chart where phenomenologically promising models cluster. Then, we apply a decision tree to our chart in order to extract the defining properties of the fertile islands. Based on this information we propose a new search strategy for phenomenologically promising string models.