"Virtuously Safe Reinforcement Learning."

Henrik Aslund et al. (2018)
a service of Schloss Dagstuhl - Leibniz Center for Informatics