Abstract: Offline reinforcement learning can enable policy learning from pre-collected, sub-optimal datasets without online interactions. This makes it ideal for real-world robots and safety-critical ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results