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 2008.01913cs.ROcs.LGcs.SYeess.SYmath.OC
收藏

模仿学习的泛化保证

Generalization Guarantees for Imitation Learning

Princeton University(普林斯顿大学)

Allen Z. Ren, Sushant Veer, Anirudha Majumdar

摘要:

由于不完善的演示或模仿学习算法无法准确推断专家的策略,来自模仿学习的控制策略通常无法推广到新的环境。在本文中,我们提出了严格的泛化保证模仿学习,利用可能近似正确(PAC)贝叶斯框架,以提供在新的环境中的政策的预期成本的上限。我们提出了一个两阶段的训练方法,其中一个潜在的政策分布是第一次嵌入多模态的专家行为,使用条件变分自编码器,然后“微调”在新的训练环境中显式优化的泛化范围。我们展示了强大的泛化边界和他们的紧密性相对于经验性能的模拟(一)抓住不同的杯子,(二)平面推动视觉反馈,(三)基于视觉的室内导航,以及通过硬件实验的两个操作任务。

英文摘要:

Control policies from imitation learning can often fail to generalize to novel environments due to imperfect demonstrations or the inability of imitation learning algorithms to accurately infer the expert's policies. In this paper, we present rigorous generalization guarantees for imitation learning by leveraging the Probably Approximately Correct (PAC)-Bayes framework to provide upper bounds on the expected cost of policies in novel environments. We propose a two-stage training method where a latent policy distribution is first embedded with multi-modal expert behavior using a conditional variational autoencoder, and then "fine-tuned" in new training environments to explicitly optimize the generalization bound. We demonstrate strong generalization bounds and their tightness relative to empirical performance in simulation for (i) grasping diverse mugs, (ii) planar pushing with visual feedback, and (iii) vision-based indoor navigation, as well as through hardware experiments for the two manipulation tasks.

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。

↑