Leveraging Offline Data from Similar Tasks to Improve Batch Reinforcement Learning
Transferring knowledge from similar source tasks can significantly improve the learning performance of the target reinforcement learning task, even with limited target data.
