Corpus ID: 182952687

Watch, Try, Learn: Meta-Learning from Demonstrations and Reward

@article{Zhou2020WatchTL,
  title={Watch, Try, Learn: Meta-Learning from Demonstrations and Reward},
  author={Allan Zhou and Eric Jang and Daniel Kappler and Alex Herzog and Mohi Khansari and Paul Wohlhart and Yunfei Bai and Mrinal Kalakrishnan and Sergey Levine and Chelsea Finn},
  journal={ArXiv},
  year={2020},
  volume={abs/1906.03352}
}
  • Allan Zhou, Eric Jang, +7 authors Chelsea Finn
  • Published 2020
  • Computer Science, Mathematics
  • ArXiv
  • Imitation learning allows agents to learn complex behaviors from demonstrations. However, learning a complex vision-based task may require an impractical number of demonstrations. Meta-imitation learning is a promising approach towards enabling agents to learn a new task from one or a few demonstrations by leveraging experience from learning similar tasks. In the presence of task ambiguity or unobserved dynamics, demonstrations alone may not provide enough information; an agent must also try… CONTINUE READING

    Figures, Tables, and Topics from this paper.

    Citations

    Publications citing this paper.
    SHOWING 1-5 OF 5 CITATIONS

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 44 REFERENCES

    Overcoming Exploration in Reinforcement Learning with Demonstrations

    VIEW 2 EXCERPTS

    Deep Q-learning From Demonstrations

    VIEW 1 EXCERPT

    One-Shot Imitation from Observing Humans via Domain-Adaptive Meta-Learning

    Hierarchical Imitation and Reinforcement Learning

    VIEW 2 EXCERPTS

    Learning to reinforcement learn

    VIEW 4 EXCERPTS
    HIGHLY INFLUENTIAL