arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2210.04723cs.AIcs.HC
收藏

强化学习的经验解释

Experiential Explanations for Reinforcement Learning

Georgia Institute of Technology(佐治亚理工学院);Harvard University(哈佛大学)

Amal Alabdulkarim, Madhuri Singh, Gennie Mansi, Kaely Hall, Upol Ehsan, Mark O. Riedl

AI总结:

针对强化学习系统难以解释的问题,提出“经验解释”技术,通过训练影响预测器生成反事实解释,恢复策略反映环境的信息,两项人类评估显示其在行为预测准确性及解释质量上均优于其他方法。

摘要:

强化学习(RL)系统可能复杂且难以解释,这使得非AI专家难以理解或干预其决策。部分原因在于RL的序列特性:行动的选择基于其获得未来奖励的可能性。然而,RL智能体丢弃了训练中的定性特征,导致难以恢复用户可理解的“为何选择某行动”的信息。我们提出一种名为“经验解释”的技术,通过在训练RL策略的同时训练影响预测器来生成反事实解释。影响预测器是学习不同奖励源在不同状态下如何影响智能体的模型,从而恢复策略如何反映环境的信息。两项人类评估研究显示,与其他标准解释类型相比,获得经验解释的参与者能更准确地猜测智能体的行为。参与者还认为经验解释更易理解、更令人满意、更完整、更有用且更准确。定性分析提供了经验解释中最有用的因素以及参与者期望的解释特征的信息。

英文摘要:

Reinforcement learning (RL) systems can be complex and non-interpretable, making it challenging for non-AI experts to understand or intervene in their decisions. This is due in part to the sequential nature of RL in which actions are chosen because of their likelihood of obtaining future rewards. However, RL agents discard the qualitative features of their training, making it difficult to recover user-understandable information for "why" an action is chosen. We propose a technique Experiential Explanations to generate counterfactual explanations by training influence predictors along with the RL policy. Influence predictors are models that learn how different sources of reward affect the agent in different states, thus restoring information about how the policy reflects the environment. Two human evaluation studies revealed that participants presented with Experiential Explanations were better able to correctly guess what an agent would do than those presented with other standard types of explanation. Participants also found that Experiential Explanations are more understandable, satisfying, complete, useful, and accurate. Qualitative analysis provides information on the factors of Experiential Explanations that are most useful and the desired characteristics that participants seek from the explanations.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑