Benchmarking Uncertainty Estimates with Deep Reinforcement Learning for Dialogue Policy Optimisation | IEEE Conference Publication | IEEE Xplore