Dr. Strategy: Model-Based Generalist Agents with Strategic Dreaming

Hany Hamed, Subin Kim, Dongyeong Kim, Jaesik Yoon, Sungjin Ahn
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:17333-17353, 2024.

Abstract

Model-based reinforcement learning (MBRL) has been a primary approach to ameliorating the sample efficiency issue as well as to make a generalist agent. However, there has not been much effort toward enhancing the strategy of dreaming itself. Therefore, it is a question whether and how an agent can “dream better” in a more structured and strategic way. In this paper, inspired by the observation from cognitive science suggesting that humans use a spatial divide-and-conquer strategy in planning, we propose a new MBRL agent, called Dr. Strategy, which is equipped with a novel Dreaming Strategy. The proposed agent realizes a version of divide-and-conquer-like strategy in dreaming. This is achieved by learning a set of latent landmarks and then utilizing these to learn a landmark-conditioned highway policy. With the highway policy, the agent can first learn in the dream to move to a landmark, and from there it tackles the exploration and achievement task in a more focused way. In experiments, we show that the proposed model outperforms prior pixel-based MBRL methods in various visually complex and partially observable navigation tasks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-hamed24a, title = {Dr. Strategy: Model-Based Generalist Agents with Strategic Dreaming}, author = {Hamed, Hany and Kim, Subin and Kim, Dongyeong and Yoon, Jaesik and Ahn, Sungjin}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {17333--17353}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/hamed24a/hamed24a.pdf}, url = {https://proceedings.mlr.press/v235/hamed24a.html}, abstract = {Model-based reinforcement learning (MBRL) has been a primary approach to ameliorating the sample efficiency issue as well as to make a generalist agent. However, there has not been much effort toward enhancing the strategy of dreaming itself. Therefore, it is a question whether and how an agent can “dream better” in a more structured and strategic way. In this paper, inspired by the observation from cognitive science suggesting that humans use a spatial divide-and-conquer strategy in planning, we propose a new MBRL agent, called Dr. Strategy, which is equipped with a novel Dreaming Strategy. The proposed agent realizes a version of divide-and-conquer-like strategy in dreaming. This is achieved by learning a set of latent landmarks and then utilizing these to learn a landmark-conditioned highway policy. With the highway policy, the agent can first learn in the dream to move to a landmark, and from there it tackles the exploration and achievement task in a more focused way. In experiments, we show that the proposed model outperforms prior pixel-based MBRL methods in various visually complex and partially observable navigation tasks.} }
Endnote
%0 Conference Paper %T Dr. Strategy: Model-Based Generalist Agents with Strategic Dreaming %A Hany Hamed %A Subin Kim %A Dongyeong Kim %A Jaesik Yoon %A Sungjin Ahn %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-hamed24a %I PMLR %P 17333--17353 %U https://proceedings.mlr.press/v235/hamed24a.html %V 235 %X Model-based reinforcement learning (MBRL) has been a primary approach to ameliorating the sample efficiency issue as well as to make a generalist agent. However, there has not been much effort toward enhancing the strategy of dreaming itself. Therefore, it is a question whether and how an agent can “dream better” in a more structured and strategic way. In this paper, inspired by the observation from cognitive science suggesting that humans use a spatial divide-and-conquer strategy in planning, we propose a new MBRL agent, called Dr. Strategy, which is equipped with a novel Dreaming Strategy. The proposed agent realizes a version of divide-and-conquer-like strategy in dreaming. This is achieved by learning a set of latent landmarks and then utilizing these to learn a landmark-conditioned highway policy. With the highway policy, the agent can first learn in the dream to move to a landmark, and from there it tackles the exploration and achievement task in a more focused way. In experiments, we show that the proposed model outperforms prior pixel-based MBRL methods in various visually complex and partially observable navigation tasks.
APA
Hamed, H., Kim, S., Kim, D., Yoon, J. & Ahn, S.. (2024). Dr. Strategy: Model-Based Generalist Agents with Strategic Dreaming. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:17333-17353 Available from https://proceedings.mlr.press/v235/hamed24a.html.

Related Material