Ma, Tao ORCID: 0000-0002-8062-9217, Yang, Xuzhi and Szabo, Zoltan (2024) To switch or not to switch? Balanced policy switching in offline reinforcement learning. . arXiv. (Submitted)