离线强化学习推荐系统因果决策变换器（英文版）

英文标题：Causal Decision Transformer for Recommender Systems via Offline Reinforcement Learning中文摘要：本文提出了一种基于因果决策转换器的推荐系统，即 CDT4Rec，用于处理离线数据集的强化学习模型。该模型采用了变压器架构，能够处理大规模的离线数据集，并捕捉数据中的长短期依赖关系，以估计动作、状态和奖励之间的因果关系。我们通过对六个真实世界离线数据集和一个在线模拟器的实验，证明了该模型的可行性和优越性。英文摘要：Reinforcement learning-based recommender systems have recently gainedpopularity. However, the design of the reward function, on which the agentrelies to optimize its recommendation policy, is often not straightforward.Exploring the causality unde