@inproceedings{whyisposteriorsamplingbetterthan,
  title = {Why is Posterior Sampling Better than Optimism for Reinforcement Learning?},
  author = {Ian Osband and Benjamin Van Roy},
  year = {2016},
  booktitle = {ICML 2017 8},
  eprint = {1607.00215},
  archivePrefix = {arXiv},
  url = {http://arxiv.org/abs/1607.00215v3},
}