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..
Wasserstein Transfer Learning
Authors: Kaicheng Zhang, Sinian Zhang, Doudou Zhou, Yidong Zhou
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The proposed methods are supported by rigorous theoretical analysis and are validated through extensive simulations and real-world applications. |
| Researcher Affiliation | Academia | 1School of Mathematical Sciences, Zhejiang University, China 2Division of Biostatistics and Health Data Science, University of Minnesota, USA 3Department of Statistics and Data Science, National University of Singapore, Singapore 4Department of Statistics, University of California, Davis, USA EMAIL, EMAIL, EMAIL, EMAIL |
| Pseudocode | Yes | Algorithm 1 Wasserstein Transfer Learning (Wa TL) Input: Target and source data {(x(0) i , Ī½(0) i )}n0 i=1 1 k K {(x(k) i , Ī½(k) i )}nk i=1 , regularization parameter Ī», and query point x Rp. Output: Target estimator bm(0) G (x). |
| Open Source Code | Yes | The code is available at https://github.com/h7nian/Wa TL. |
| Open Datasets | Yes | We evaluate the Wa TL algorithm using data from the National Health and Nutrition Examination Survey (NHANES) 2005 20063, focusing on modeling the distribution of physical activity intensity. ... 3https://wwwn.cdc.gov/nchs/nhanes/Continuous Nhanes/Default.aspx?Begin Year=2005 |
| Dataset Splits | Yes | The regularization parameter Ī» in Algorithm 1 is selected via five-fold cross-validation, ranging from 0 to 3 in increments of 0.1. |
| Hardware Specification | No | This paper introduces a new transfer learning method for regression with distributional outputs, emphasizing methodological and theoretical contributions rather than computational efficiency. The approach is not compute-intensive and was implemented using standard CPU resources, so details on compute infrastructure and execution time were not included. |
| Software Dependencies | No | The paper does not specify particular software dependencies or their versions. It describes algorithms and theoretical frameworks, and mentions implementation using standard CPU resources, but does not detail the software stack. |
| Experiment Setup | Yes | We vary the target sample size n0 from 200 to 800, while the source sample size is set as nk = kĻ, where Ļ {100, 200} and k = 1, . . . , K. The regularization parameter Ī» in Algorithm 1 is selected via five-fold cross-validation, ranging from 0 to 3 in increments of 0.1. |