Awesome RL with Human Feedback / 2022 / arXiv

SURF: Semi-supervised Reward Learning with Data Augmentation for Feedback-efficient Preference-based Reinforcement Learning

作者：Jongjin Park, Younggyo Seo, Jinwoo Shin, Honglak Lee, Pieter Abbeel, Kimin Lee · 单位：University of Michigan · 来源：arXiv · 发布日期：2022

奖励建模偏好学习样本效率数据筛选
新标签打开 AlphaXiv
SURF: Semi-supervised Reward Learning with Data Augmentation for Feedback-efficient Preference-based Reinforcement Learning figure
AlphaXiv 中文概览（可滚动查看）