… paradigm which facilitates our empirical analysis of the difficulties in offline reinforcement …
detailed empirical analysis of the failure modes of passive (ie non-interactive, offline) learning, …

… Reproducibility of benchmarked deep reinforcement learning tasks for continuous control.
In: Proceedings of the ICML 2017 workshop on Reproducibility in Machine Learning (RML). …

… As we do not have the ability to generate more data through an exploration of novel treatment
strategies, we develop a policy using offline, batch reinforcement learning. In this setting, it …

… I propose using reinforcement learning to analyze the regret … I have developed a new library
of reinforcement learning tools, … Finally, I use our new reinforcement learning tools to provide …

… We empirically analyze the properties of these new emphatic algorithms to observe how
qualitative properties such as convergence, learning speed and variance manifest in practice. …

… More specifically, reinforcement learning methods, including multiple regressions, random …
crucial aspect of machine learning models. In addition, reinforcement learning is analogous to …

… evaluation for the analysis, but we will compare both approaches empirically. Now, we …
the broader impact of our contribution to be the same as the one of reinforcement learning. …

… To identify the effect of overfitting in independent reinforcement learners, we introduce joint
policy correlation (JPC) matrices. To simplify the presentation, we describe here the special …

… Reinforcement learning (RL) formalises the problems of evaluation and optimisation of
an … We refer to these approaches as categorical distributional reinforcement learning (CDRL). …

… This work mostly looked at a theoretical foundation of reinforcement learning, which, in
the author’s view, contributes to understand RL as a general and natural phenomenon of the …