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 1809.03060cs.LGcs.AIstat.ML
收藏

主动逆向奖励设计

Active Inverse Reward Design

UC Berkeley(加州大学伯克利分校)

Sören Mindermann, Rohin Shah, Adam Gleave, Dylan Hadfield-Menell

AI总结:

提出主动逆向奖励设计,通过查询设计者比较奖励函数来高效学习真实奖励,优于现有IRD方法,并能从线性查询中推断非线性奖励。

摘要:

AI 智能体的设计者通常通过试错过程迭代调整奖励函数,直到获得期望的行为,但这仅能保证在训练环境中的良好表现。我们提出将这一过程构建为一系列查询,要求用户比较不同的奖励函数。由此,我们可以主动选择对真实奖励信息量最大的查询。与要求设计者提供最优行为的方法不同,我们的方法通过引出对次优行为的偏好来收集额外信息。每次查询后,我们需要根据设计者选择的代理奖励函数更新真实奖励函数的后验分布。最近提出的逆向奖励设计(IRD)方法支持这一更新过程。我们的方法在测试环境中显著优于 IRD。特别是,它可以查询设计者关于可解释的线性奖励函数,并仍能推断出非线性奖励函数。

英文摘要:

Designers of AI agents often iterate on the reward function in a trial-and-error process until they get the desired behavior, but this only guarantees good behavior in the training environment. We propose structuring this process as a series of queries asking the user to compare between different reward functions. Thus we can actively select queries for maximum informativeness about the true reward. In contrast to approaches asking the designer for optimal behavior, this allows us to gather additional information by eliciting preferences between suboptimal behaviors. After each query, we need to update the posterior over the true reward function from observing the proxy reward function chosen by the designer. The recently proposed Inverse Reward Design (IRD) enables this. Our approach substantially outperforms IRD in test environments. In particular, it can query the designer about interpretable, linear reward functions and still infer non-linear ones.

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。

↑