Exploring string theory solutions with machine learning
Fabian Ruehle (Oxford University)

We will start with a pedagogical introduction to machine learning with particular emphasis on the subfield of reinforcement learning. Subsequently, we discuss how to apply reinforcement learning to the study of a particular class of string theory solutions, so-called Type II intersecting brane worlds. We explain the physical consistency conditions that arise in these constructions, without assuming any prior string theory knowledge. We end by mapping this set of string theory consistency constraints onto an environment that can be analyzed with reinforcement learning and present some preliminary results of the AI exploration of this corner of the string theory landscape.