1 code implementation • NeurIPS Workshop LMCA 2020 • Konrad Czechowski, Tomasz Odrzygóźdź, Michał Izworski, Marek Zbysiński, Łukasz Kuciński, Piotr Miłoś
We propose $\textit{trust-but-verify}$ (TBV) mechanism, a new method which uses model uncertainty estimates to guide exploration.
Model-based Reinforcement Learning reinforcement-learning +1