Home > Research > Publications & Outputs > Algorithms or Actions?

Research

Associated organisational units

Electronic data

  • ijcai-2018

    Accepted author manuscript, 830 KB, PDF document

    Available under license: CC BY-NC: Creative Commons Attribution-NonCommercial 4.0 International License

Links

Text available via DOI:

Keywords

View graph of relations

Algorithms or Actions?: A Study in Large-Scale Reinforcement Learning

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSN › Conference contribution/Paper › peer-review

Published

Standard

Algorithms or Actions? A Study in Large-Scale Reinforcement Learning. / Rocha Tavares, Anderson; Anbalagan, Sivasubramanian; Soriano Marcolino, Leandro et al.
Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence (IJCAI-18). International Joint Conferences on Artificial Intelligence, 2018. p. 2717-2723.

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSN › Conference contribution/Paper › peer-review

Harvard

Rocha Tavares, A, Anbalagan, S, Soriano Marcolino, L & Chaimowicz, L 2018, Algorithms or Actions? A Study in Large-Scale Reinforcement Learning. in Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence (IJCAI-18). International Joint Conferences on Artificial Intelligence, pp. 2717-2723, 27th International Joint Conference on Artificial Intelligence, IJCAI 2018, Stockholm, Sweden, 13/07/18. https://doi.org/10.24963/ijcai.2018/377

APA

Rocha Tavares, A., Anbalagan, S., Soriano Marcolino, L., & Chaimowicz, L. (2018). Algorithms or Actions? A Study in Large-Scale Reinforcement Learning. In Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence (IJCAI-18) (pp. 2717-2723). International Joint Conferences on Artificial Intelligence. https://doi.org/10.24963/ijcai.2018/377

Vancouver

Rocha Tavares A, Anbalagan S, Soriano Marcolino L, Chaimowicz L. Algorithms or Actions? A Study in Large-Scale Reinforcement Learning. In Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence (IJCAI-18). International Joint Conferences on Artificial Intelligence. 2018. p. 2717-2723 doi: 10.24963/ijcai.2018/377

Author

Rocha Tavares, Anderson ; Anbalagan, Sivasubramanian ; Soriano Marcolino, Leandro et al. / Algorithms or Actions? A Study in Large-Scale Reinforcement Learning. Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence (IJCAI-18). International Joint Conferences on Artificial Intelligence, 2018. pp. 2717-2723

Bibtex

@inproceedings{db871d12c839483c81991c0440b528c3,
title = "Algorithms or Actions?: A Study in Large-Scale Reinforcement Learning",
abstract = "Large state and action spaces are very challenging to reinforcement learning. However, in many domains there is a set of algorithms available, which estimate the best action given a state. Hence, agents can either directly learn a performance-maximizing mapping from states to actions, or from states to algorithms. We investigate several aspects of this dilemma, showing sufficient conditions for learning over algorithms to outperform over actions for a finite number of training iterations. We present synthetic experiments to further study such systems. Finally, we propose a function approximation approach, demonstrating the effectiveness of learning over algorithms in real-time strategy games.",
keywords = "Machine Learning: Reinforcement Learning, Multidisciplinary Topics and Applications: Computer Games, Uncertainty in AI: Markov Decision Processes, Machine Learning Applications: Game Playing",
author = "{Rocha Tavares}, Anderson and Sivasubramanian Anbalagan and {Soriano Marcolino}, Leandro and Luiz Chaimowicz",
year = "2018",
month = jul,
day = "1",
doi = "10.24963/ijcai.2018/377",
language = "English",
isbn = "9780999241127",
pages = "2717--2723",
booktitle = "Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence (IJCAI-18)",
publisher = "International Joint Conferences on Artificial Intelligence",
note = "27th International Joint Conference on Artificial Intelligence, IJCAI 2018 ; Conference date: 13-07-2018 Through 19-07-2018",

}

RIS

TY - GEN

T1 - Algorithms or Actions?

T2 - 27th International Joint Conference on Artificial Intelligence, IJCAI 2018

AU - Rocha Tavares, Anderson

AU - Anbalagan, Sivasubramanian

AU - Soriano Marcolino, Leandro

AU - Chaimowicz, Luiz

PY - 2018/7/1

Y1 - 2018/7/1

N2 - Large state and action spaces are very challenging to reinforcement learning. However, in many domains there is a set of algorithms available, which estimate the best action given a state. Hence, agents can either directly learn a performance-maximizing mapping from states to actions, or from states to algorithms. We investigate several aspects of this dilemma, showing sufficient conditions for learning over algorithms to outperform over actions for a finite number of training iterations. We present synthetic experiments to further study such systems. Finally, we propose a function approximation approach, demonstrating the effectiveness of learning over algorithms in real-time strategy games.

AB - Large state and action spaces are very challenging to reinforcement learning. However, in many domains there is a set of algorithms available, which estimate the best action given a state. Hence, agents can either directly learn a performance-maximizing mapping from states to actions, or from states to algorithms. We investigate several aspects of this dilemma, showing sufficient conditions for learning over algorithms to outperform over actions for a finite number of training iterations. We present synthetic experiments to further study such systems. Finally, we propose a function approximation approach, demonstrating the effectiveness of learning over algorithms in real-time strategy games.

KW - Machine Learning: Reinforcement Learning

KW - Multidisciplinary Topics and Applications: Computer Games

KW - Uncertainty in AI: Markov Decision Processes

KW - Machine Learning Applications: Game Playing

U2 - 10.24963/ijcai.2018/377

DO - 10.24963/ijcai.2018/377

M3 - Conference contribution/Paper

SN - 9780999241127

SP - 2717

EP - 2723

BT - Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence (IJCAI-18)

PB - International Joint Conferences on Artificial Intelligence

Y2 - 13 July 2018 through 19 July 2018

ER -