Cooperative Inverse Reinforcement Learning
Dylan Hadfield-Menell Anca Dragan Pieter Abbeel Stuart Russell Electrical Engineering and Computer Science University of California at Berkeley Berkeley, CA 94709 {dhm, anca, pabbeel,
Abstract
For an autonomous system to be helpful to humans and to pose no unwarranted risks, it needs to align its values with those of the humans in its environment in such a way that its actions contribute to the maximization of value for the humans. We propose a formal definition of the value alignment problem as cooperative inverse reinforcement learning (CIRL). A CIRL problem is a cooperative, partial-information game with two agents, human and robot; both are rewarded according to the human’s reward function, but the robot does not initially know what this is. In contrast to classical IRL, where the human is assumed to act optimally in isolation, optimal CIRL solutions produce behaviors such as active teaching, active learning, and communicative actions that are more effective in achieving value alignment. We show that computing optimal joint policies in CIRL games can be reduced to solving a POMDP, prove that optimality in isolation is suboptimal in CIRL, and derive an approximate CIRL algorithm.
中文速览
让AI真正"为人类服务"而不惹麻烦，关键在于让它准确理解人类的目标——这就是所谓的价值对齐（value alignment）问题。传统的逆强化学习（inverse reinforcement learning, IRL）试图通过观察人类行为来推断其偏好，但它假设人类总是"最优地"独自行动，这导致人和机器人之间缺乏真正的互动与沟通。这篇论文把价值对齐重新定义为一个合作博弈，称为合作逆强化学习（cooperative inverse reinforcement learning, CIRL）：人类知道真正的奖励函数而机器人不知道，两者却共享同一个目标——最大化人类的收益，从而激励人类主动"教"、机器人主动"学"。作者证明了传统IRL中"人类只管自己最优演示"的做法在CIRL框架下并非最优，并将求解CIRL最优策略转化为一个可部分观测马尔可夫决策过程（POMDP）来高效求解，同时给出了近似算法并通过实验验证。这项工作为设计真正懂得与人协作、主动理解人类意图的AI系统提供了严格的理论基础。
原文 arXiv:1606.03137;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1606.03137v4