Similarity-based Knowledge Transfer for Cross-Domain Reinforcement Learning
This work addresses the challenge of knowledge transfer in cross-domain RL, which is incremental as it builds on prior methods by removing the need for data alignment or expert supervision.
The paper tackles the problem of selecting a source of knowledge for cross-domain reinforcement learning by measuring task similarity, resulting in a method that effectively transfers policies without requiring aligned data or expert policies, as demonstrated on Mujoco control tasks.
Transferring knowledge in cross-domain reinforcement learning is a challenging setting in which learning is accelerated by reusing knowledge from a task with different observation and/or action space. However, it is often necessary to carefully select the source of knowledge for the receiving end to benefit from the transfer process. In this article, we study how to measure the similarity between cross-domain reinforcement learning tasks to select a source of knowledge that will improve the performance of the learning agent. We developed a semi-supervised alignment loss to match different spaces with a set of encoder-decoders, and use them to measure similarity and transfer policies across tasks. In comparison to prior works, our method does not require data to be aligned, paired or collected by expert policies. Experimental results, on a set of varied Mujoco control tasks, show the robustness of our method in effectively selecting and transferring knowledge, without the supervision of a tailored set of source tasks.