Bounding-Box Inference for Error-Aware Model-Based Reinforcement Learning

Erin J Talvitie, Zilei Shao, Huiying Li, Jinghan Hu, Jacob Boerma, Rory Zhao, Xintong Wang; 5:2440−2460, 2024.

Presented at the Reinforcement Learning Conference (RLC), Amherst Massachusetts, August 9–12, 2024.

Abstract

In model-based reinforcement learning, simulated experiences from the learned model are often treated as equivalent to experience from the real environment. However, when the model is inaccurate, it can catastrophically interfere with policy learning. Alternatively, the agent might learn about the model's accuracy and selectively use it only when it can provide reliable predictions. We empirically explore model uncertainty measures for selective planning and show that best results require distribution insensitive inference to estimate the uncertainty over model-based updates. To that end, we propose and evaluate bounding-box inference, which operates on bounding-boxes around sets of possible states and other quantities. We find that bounding-box inference can reliably support effective selective planning.

[abs][pdf]

BibTeX

@article{talvitie2024bounding,
    title={Bounding-Box Inference for Error-Aware Model-Based Reinforcement Learning},
    author={Talvitie, Erin J and Shao, Zilei and Li, Huiying and Hu, Jinghan and Boerma, Jacob and Zhao, Rory and Wang, Xintong},
    journal={Reinforcement Learning Journal},
    volume={5},
    pages={2440--2460},
    year={2024}
}