BRUNSKILL, E. When Policies Can Be Trusted: Analyzing a Criteria to Identify Optimal Policies in MDPs with Unknown Model Parameters. Proceedings of the International Conference on Automated Planning and Scheduling, [S. l.], v. 20, n. 1, p. 218-221, 2010. DOI: 10.1609/icaps.v20i1.13438. Disponível em: https://ojs.aaai.org/index.php/ICAPS/article/view/13438. Acesso em: 24 nov. 2024.