arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2106.03498cs.LGmath.OC
收藏

逆强化学习中的可辨识性

Identifiability in inverse reinforcement learning

Alan Turing Institute(艾伦·图灵研究所);University of Oxford(牛津大学);University of Edinburgh(爱丁堡大学)

Haoyang Cao, Samuel N. Cohen, Lukasz Szpruch

摘要:

逆强化学习试图利用对代理行为的观察来重构马尔可夫决策问题中的奖励函数。正如Russell [1998]中已经观察到的那样,这个问题是不适定的,即使在存在关于最优行为的完美信息的情况下,奖励函数也是不可识别的。我们提供了一个解决方案,这种不可识别的熵正则化的问题。对于一个给定的环境中,我们充分刻画奖励函数导致一个给定的政策,并证明,在两个不同的折扣因子下,或在足够不同的环境下,相同的奖励的行动示范,未观察到的奖励可以恢复到一个常数。我们还给出了在有限视野上重建时间齐次奖励和行动独立奖励的一般必要和充分条件,推广了Kim et al. [2021]和Fu et al. [2018]的最新结果。

英文摘要:

Inverse reinforcement learning attempts to reconstruct the reward function in a Markov decision problem, using observations of agent actions. As already observed in Russell [1998] the problem is ill-posed, and the reward function is not identifiable, even under the presence of perfect information about optimal behavior. We provide a resolution to this non-identifiability for problems with entropy regularization. For a given environment, we fully characterize the reward functions leading to a given policy and demonstrate that, given demonstrations of actions for the same reward under two distinct discount factors, or under sufficiently different environments, the unobserved reward can be recovered up to a constant. We also give general necessary and sufficient conditions for reconstruction of time-homogeneous rewards on finite horizons, and for action-independent rewards, generalizing recent results of Kim et al. [2021] and Fu et al. [2018].

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。

↑