Offline reinforcement learning: how conservative algorithms can enable new applications

Open in new window