Author : Tony Z. Zhao*, Anusha Nagabandi*, Kate Rakelly*, Chelsea Finn, Sergey Levine
Paper Link : arxiv.org/abs/2010.13957

Website : sites.google.com/view/meld-lsm/home

Code: https://github.com/tonyzhaozh/meld

Talk : https://youtu.be/P7Z68dRvne4

 

  • PEARL의 이미지 input 버전
  • PEARL에다 SLAC에서 사용한 latent dynamic를 활용
  • 실제 로봇으로 real world에서 meta-RL을 검증한 첫(아마도?) 사례이기 때문에 의의가 있음

'AI & RL > Meta & Multi-Task RL' 카테고리의 다른 글

[요약] A Consciousness-Inspired Planning Agent for Model-Based Reinforcement Learning (Mingde Zhao, NeurIPS 2021)  (0) 2021.10.02
[북마크] Bootstrapped Meta-Learning (Sebastian Flennerhag, ICLR 2022 Oral)  (0) 2021.09.22
[요약] Improving Context-Based Meta-Reinforcement Learning with Self-Supervised Trajectory Contrastive Learning (Bernie Wang, ArXiv 2021)  (0) 2021.05.28
[정리] VariBAD: A Very Good Method for Bayes-Adaptive Deep RL via Meta-Learning (Luisa Zintgraf, ICLR 2020)  (0) 2021.01.31
[정리] PEARL: Efficient Off-policy Meta-learning via Probabilistic Context Variables (Kate Rakelly, ICML 2019)  (2) 2020.05.27

+ Recent posts

티스토리툴바