Code accompanying the paper "Universal Value-Function Uncertainties"

DOI:10.4121/2393f9ad-939d-4338-a44f-7d8f9251f244.v1

The DOI displayed above is for this specific version of this dataset, which is currently the latest. Newer versions may be published in the future. For a link that will always point to the latest version, please use
DOI: 10.4121/2393f9ad-939d-4338-a44f-7d8f9251f244

Datacite citation style

Zanger, Moritz A.; Weltevrede, Max; Oren, Yaniv; van der Vaart, Pascal; Horsch, Caroline et. al. (2025): Code accompanying the paper "Universal Value-Function Uncertainties". Version 1. 4TU.ResearchData. software. https://doi.org/10.4121/2393f9ad-939d-4338-a44f-7d8f9251f244.v1

Other citation styles (APA, Harvard, MLA, Vancouver, Chicago, IEEE) available at Datacite

Software

Keywords

deep ensembles deep learning exploration machine learning offline reinforcement learning reinforcement learning uncertainty quantification

Licence

MIT

Interoperability

RO-Crate Metadata

Export as...

RefWorks BibTeX Reference Manager Endnote DataCite NLM DC CFF

by Moritz A. Zanger

, Max Weltevrede, Yaniv Oren, Pascal van der Vaart

, Caroline Horsch, Wendelin Boehmer, M.T.J. (Matthijs) Spaan

This is the official code repository accompanying the paper "Universal Value-Function Uncertainties ".

History

2025-08-18 first online, published, posted

Publisher

4TU.ResearchData

Format

.py; .pkl

Associated peer-reviewed publication

Universal Value-Function Uncertainties

References

https://arxiv.org/abs/2505.21119

Funding

Epistemic AI (grant code 964505) [more info...] EXCELLENT SCIENCE - Future and Emerging Technologies (FET) (EU Horizons 2020)

Organizations

Delft University of Technology, Faculty of Electrical Engineering, Mathematics and Computer Science,
Department of Intelligent Systems, Sequential Decision Making Group

DATA - under embargo

The files in this dataset are under embargo until 2027-08-14.

Reason

This dataset contains source code developed to reproduce experiments for an ongoing research project that has not yet been formally published. While a preprint version of the associated manuscript is publicly available, the code itself remains confidential to protect the integrity of the research and the publication process. Direct access to the code will be granted only in exceptional circumstances and will remain restricted until the official publication of the associated work.