Policy Distillation

Policy Distillation is a policy learning method that transfers knowledge from a high-capacity teacher policy to a more compact student policy by minimizing the divergence between their action distributions across states, enabling efficient deployment in data-constrained robotic environments.

22

Publications

1.2K

Citations

140

Authors

32

Institutions

Publications per year

2017–2026

20

Authors

140

  1. RP
    Razvan Pascanu

    Google (United States), United States

    2 publications · 126 citations

    2

    h-index

  2. CT
    Chen Tessler

    Technion – Israel Institute of Technology, Israel

    2 publications · 363 citations

    2

    h-index

  3. RH
    Raia Hadsell

    Google DeepMind (United Kingdom), United Kingdom

    2 publications · 103 citations

    2

    h-index

  4. MV
    Michiel van de Panne

    University of British Columbia, Canada

    2 publications · 85 citations

    2

    h-index

  5. HC
    Hugo Caselles-Dupré

    SoftBank Robotics (France), France

    2 publications · 52 citations

    2

    h-index

1–5 of 140

Venues

  1. arXiv (Cornell University)

    Repository · 12 publications · 548 citations

    12

    h-index

  2. Proceedings of the AAAI Conference on Artificial Intelligence

    Conference · 3 publications · 371 citations

    3

    h-index

  3. IEEE Intelligent Systems

    1 publications · 30 citations

    1

    h-index

  4. 1

    h-index

  5. IEEE/ACM Transactions on Networking

    TON · 1 publications · 30 citations

    1

    h-index

1–5