Papers of the day   All papers

Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems

Comments

Sergey Levine: Want to learn about offline RL? Check out our new tutorial on offline RL, w/ Aviral Kumar, @georgejtucker, Justin Fu: https://arxiv.org/abs/2005.01643 Offline RL may enable RL algorithms to use large offline datasets, and thus make it applicable to a wide range of real-world problems. https://t.co/tj0vfKlffZ

5 replies, 499 likes


Sergey Levine: We've updated our offline RL tutorial and survey article to include discussion of some recent work that has come up since the original release of the survey! http://arxiv.org/abs/2005.01643

1 replies, 251 likes


mat kelcey: "Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems" https://arxiv.org/abs/2005.01643 https://t.co/8w3mTGbDQQ

0 replies, 140 likes


Sergey Levine: To learn more about offline RL, check out our tutorial: https://arxiv.org/abs/2005.01643 And if you want to try it out, check out our D4RL offline RL benchmarks: https://sites.google.com/view/d4rl/home

0 replies, 36 likes


Sergey Levine: And of course if you want to learn more about offline RL, check out our (now slightly out of date...) tutorial on offline RL algorithms: https://arxiv.org/abs/2005.01643

1 replies, 14 likes


Isaac Kargar: In this thread, I've collected some useful resources for offline RL: @svlevine's talks, review paper, and blog post: - https://www.youtube.com/watch?v=cjM5ArKvFMw - https://www.youtube.com/watch?v=IUAePhU0E7Y&t=1s - https://medium.com/@sergey.levine/decisions-from-data-how-offline-reinforcement-learning-will-change-how-we-use-ml-24d98cb069b0 - https://arxiv.org/abs/2005.01643

1 replies, 12 likes


Daisuke Okanohara: This is an excellent review article on offline RL (aka batch RL). Offline RL is promising for mission-critical (e.g.,. autonomous driving, healthcare) or human involving tasks (e.g., dialog). Need to consider distribution shift and epistemic uncertainty. https://arxiv.org/abs/2005.01643

0 replies, 4 likes


Reinforcement Learning Turkiye: You should read this paper, it's about offline RL! "Offline reinforcement learning algorithms hold tremendous promise for making it possible to turn large datasets into powerful decision making engines. " https://arxiv.org/pdf/2005.01643.pdf

1 replies, 3 likes


志水克大: Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems. (arXiv:2005.01643v2 [cs.LG] UPDATED) http://arxiv.org/abs/2005.01643

0 replies, 2 likes


HotComputerScience: Most popular computer science paper of the day: "Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems" https://hotcomputerscience.com/paper/offline-reinforcement-learning-tutorial-review-and-perspectives-on-open-problems https://twitter.com/svlevine/status/1257477600655466496

0 replies, 1 likes


mooopan: http://arxiv.org/abs/2005.01643 says: > To avoid this confusion, we will instead use the term “offline reinforcement learning” in this tutorial. but I'm not sure who introduced the term. Is it https://arxiv.org/abs/1911.11361?

0 replies, 1 likes


午後のarXiv: "Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems", Sergey Levine, Aviral Kumar,… https://arxiv.org/abs/2005.01643

0 replies, 1 likes


Content

Found on May 05 2020 at https://arxiv.org/pdf/2005.01643.pdf

PDF content of a computer science paper: Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems