Multi-task Reinforcement Learning

Multi-Task Reinforcement Learning is a reinforcement learning paradigm that enables a single agent to simultaneously learn multiple distinct tasks by sharing knowledge across them through a common policy or value function, thereby improving generalization and sample efficiency compared to learning each task independently.

130

Publications

8.3K

Citations

451

Authors

147

Institutions

Publications per year

2017–2026

106

Variants

Mtrl

Authors

451

  1. ME
    Matthew E. Taylor

    Washington State University, United States

    5 publications · 363 citations

    5

    h-index

  2. WM
    Wojciech Marian Czarnecki

    Google (United States), United States

    4 publications · 454 citations

    4

    h-index

  3. RM
    Rémi Munos

    Google (United States), United States

    3 publications · 685 citations

    3

    h-index

  4. HS
    Haifeng Sun

    Beijing University of Posts and Telecommunications, China

    3 publications · 188 citations

    3

    h-index

  5. LE
    Lasse Espeholt

    Google (United States), United States

    3 publications · 839 citations

    3

    h-index

1–5 of 451

Venues

  1. arXiv (Cornell University)

    Repository · 35 publications · 2.8K citations

    23

    h-index

  2. Proceedings of the AAAI Conference on Artificial Intelligence

    Conference · 5 publications · 412 citations

    6

    h-index

  3. IEEE Transactions on Vehicular Technology

    2 publications · 106 citations

    3

    h-index

  4. IEEE Access

    3 publications · 48 citations

    3

    h-index

  5. IEEE Robotics and Automation Letters

    1 publications · 26 citations

    3

    h-index

1–5