Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..

Scalable and adaptive prediction bands with kernel sum-of-squares

Authors: Louis Allain, Sébastien Da Veiga, Brian Staber

NeurIPS 2025 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, extensive experiments are conducted to show how our method compares to related work. All figures can be reproduced with the accompanying code at gitlab.com/drti/ksos-bands.
Researcher Affiliation Collaboration 1Safran Tech, Digital Sciences & Technologies,78114 Magny-Les-Hameaux, France 2Univ Rennes, Ensai, CNRS, CREST UMR 9194, F-35000 Rennes, France EMAIL EMAIL
Pseudocode No The paper only describes algorithmic steps in general terms without presenting them in a structured pseudocode or algorithm block.
Open Source Code Yes All figures can be reproduced with the accompanying code at gitlab.com/drti/ksos-bands.
Open Datasets Yes At last, we consider six real-world datasets commonly used for regression, which are detailed in Appendix B.5.
Dataset Splits No This dataset is split in two parts: a pre-training dataset Dn = {(Xi, Yi)}n i=1 and a calibration one Dm = {(Xi, Yi)}m i=1 with N = n + m. Table 1 reports the mean width of prediction intervals on a test set for all methods. The main text describes the conceptual split but not specific proportions or sizes for the experiments, deferring details to the Appendix.
Hardware Specification No The main text of the paper does not provide specific hardware details used for running experiments. The NeurIPS checklist indicates these details are in the Appendix.
Software Dependencies No The main text of the paper does not specify particular software dependencies with version numbers.
Experiment Setup Yes We consider a Matérn 5/2 kernel for km and kf. In the following, we thus fix it at λ2 = 1. As a result, we choose to set λ1 = 1. In the following we now fix b = 10 and compare kernel So S with HSIC-optimized θf to CQR and rescaled GPs on test case 1 with 20 replications.