Hieros: Hierarchical Imagination on Structured State Space Sequence World Models

Paul Mattes, Rainer Schlosser, Ralf Herbrich
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:35079-35103, 2024.

Abstract

One of the biggest challenges to modern deep reinforcement learning (DRL) algorithms is sample efficiency. Many approaches learn a world model in order to train an agent entirely in imagination, eliminating the need for direct environment interaction during training. However, these methods often suffer from either a lack of imagination accuracy, exploration capabilities, or runtime efficiency. We propose HIEROS, a hierarchical policy that learns time abstracted world representations and imagines trajectories at multiple time scales in latent space. HIEROS uses an S5 layer-based world model, which predicts next world states in parallel during training and iteratively during environment interaction. Due to the special properties of S5 layers, our method can train in parallel and predict next world states iteratively during imagination. This allows for more efficient training than RNN-based world models and more efficient imagination than Transformer-based world models. We show that our approach outperforms the state of the art in terms of mean and median normalized human score on the Atari 100k benchmark, and that our proposed world model is able to predict complex dynamics very accurately. We also show that HIEROS displays superior exploration capabilities compared to existing approaches.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-mattes24a, title = {Hieros: Hierarchical Imagination on Structured State Space Sequence World Models}, author = {Mattes, Paul and Schlosser, Rainer and Herbrich, Ralf}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {35079--35103}, 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/mattes24a/mattes24a.pdf}, url = {https://proceedings.mlr.press/v235/mattes24a.html}, abstract = {One of the biggest challenges to modern deep reinforcement learning (DRL) algorithms is sample efficiency. Many approaches learn a world model in order to train an agent entirely in imagination, eliminating the need for direct environment interaction during training. However, these methods often suffer from either a lack of imagination accuracy, exploration capabilities, or runtime efficiency. We propose HIEROS, a hierarchical policy that learns time abstracted world representations and imagines trajectories at multiple time scales in latent space. HIEROS uses an S5 layer-based world model, which predicts next world states in parallel during training and iteratively during environment interaction. Due to the special properties of S5 layers, our method can train in parallel and predict next world states iteratively during imagination. This allows for more efficient training than RNN-based world models and more efficient imagination than Transformer-based world models. We show that our approach outperforms the state of the art in terms of mean and median normalized human score on the Atari 100k benchmark, and that our proposed world model is able to predict complex dynamics very accurately. We also show that HIEROS displays superior exploration capabilities compared to existing approaches.} }
Endnote
%0 Conference Paper %T Hieros: Hierarchical Imagination on Structured State Space Sequence World Models %A Paul Mattes %A Rainer Schlosser %A Ralf Herbrich %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-mattes24a %I PMLR %P 35079--35103 %U https://proceedings.mlr.press/v235/mattes24a.html %V 235 %X One of the biggest challenges to modern deep reinforcement learning (DRL) algorithms is sample efficiency. Many approaches learn a world model in order to train an agent entirely in imagination, eliminating the need for direct environment interaction during training. However, these methods often suffer from either a lack of imagination accuracy, exploration capabilities, or runtime efficiency. We propose HIEROS, a hierarchical policy that learns time abstracted world representations and imagines trajectories at multiple time scales in latent space. HIEROS uses an S5 layer-based world model, which predicts next world states in parallel during training and iteratively during environment interaction. Due to the special properties of S5 layers, our method can train in parallel and predict next world states iteratively during imagination. This allows for more efficient training than RNN-based world models and more efficient imagination than Transformer-based world models. We show that our approach outperforms the state of the art in terms of mean and median normalized human score on the Atari 100k benchmark, and that our proposed world model is able to predict complex dynamics very accurately. We also show that HIEROS displays superior exploration capabilities compared to existing approaches.
APA
Mattes, P., Schlosser, R. & Herbrich, R.. (2024). Hieros: Hierarchical Imagination on Structured State Space Sequence World Models. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:35079-35103 Available from https://proceedings.mlr.press/v235/mattes24a.html.

Related Material