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 2402.13419cs.AI
收藏

基于模型的规划智能体行为保证的奖励下界

Reward Bound for Behavioral Guarantee of Model-based Planning Agents

University of California, Merced(加州大学默塞德分校)

Zhiyu An, Xianzhong Ding, Wan Du

AI总结:

本文研究基于模型的规划智能体在指定时间步内到达目标状态的行为保证问题,证明目标奖励存在下界,并扩展至多目标偏好约束。

摘要:

近年来,人们对野外基于机器学习智能体的可信度日益关注,尤其是在机器人领域,以便为工业界提供安全保障。为这些智能体获取行为保证仍然是一个重要问题。在这项工作中,我们专注于保证基于模型的规划智能体在特定未来时间步内到达目标状态。我们证明,目标状态处的奖励存在一个下界,如果所述奖励低于该下界,则不可能获得这样的保证。进一步地,我们展示了如何在多个目标上强制执行偏好。

英文摘要:

Recent years have seen an emerging interest in the trustworthiness of machine learning-based agents in the wild, especially in robotics, to provide safety assurance for the industry. Obtaining behavioral guarantees for these agents remains an important problem. In this work, we focus on guaranteeing a model-based planning agent reaches a goal state within a specific future time step. We show that there exists a lower bound for the reward at the goal state, such that if the said reward is below that bound, it is impossible to obtain such a guarantee. By extension, we show how to enforce preferences over multiple goals.

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。

↑