Corpus ID: 73728904

Skew-Fit: State-Covering Self-Supervised Reinforcement Learning

@article{Pong2019SkewFitSS,
  title={Skew-Fit: State-Covering Self-Supervised Reinforcement Learning},
  author={Vitchyr H. Pong and Murtaza Dalal and Steven Lin and Ashvin Nair and Shikhar Bahl and Sergey Levine},
  journal={ArXiv},
  year={2019},
  volume={abs/1903.03698}
}
  • Vitchyr H. Pong, Murtaza Dalal, +3 authors Sergey Levine
  • Published 2019
  • Computer Science, Mathematics
  • ArXiv
  • In standard reinforcement learning, each new skill requires a manually-designed reward function, which takes considerable manual effort and engineering. Self-supervised goal setting has the potential to automate this process, enabling an agent to propose its own goals and acquire skills that achieve these goals. However, such methods typically rely on manually-designed goal distributions, or heuristics to force the agent to explore a wide range of states. We propose a formal exploration… CONTINUE READING

    Citations

    Publications citing this paper.
    SHOWING 1-10 OF 45 CITATIONS

    An Intrinsically-Motivated Approach for Learning Highly Exploring and Fast Mixing Policies

    VIEW 2 EXCERPTS
    CITES METHODS & BACKGROUND

    Contextual Imagined Goals for Self-Supervised Robotic Learning

    VIEW 3 EXCERPTS
    CITES BACKGROUND & METHODS

    Emergent Real-World Robotic Skills via Unsupervised Off-Policy Reinforcement Learning

    VIEW 1 EXCERPT
    CITES BACKGROUND

    FILTER CITATIONS BY YEAR

    2019
    2020

    CITATION STATISTICS

    • 5 Highly Influenced Citations

    • Averaged 23 Citations per year from 2019 through 2020

    • 65% Increase in citations per year in 2020 over 2019

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 49 REFERENCES

    Visual Reinforcement Learning with Imagined Goals

    VIEW 9 EXCERPTS

    Many-Goals Reinforcement Learning

    VIEW 2 EXCERPTS

    Unsupervised Control Through Non-Parametric Discriminative Rewards

    VIEW 6 EXCERPTS
    HIGHLY INFLUENTIAL

    Automatic Goal Generation for Reinforcement Learning Agents

    VIEW 4 EXCERPTS
    HIGHLY INFLUENTIAL

    Episodic Curiosity through Reachability

    VIEW 2 EXCERPTS