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..
Understanding Fairness and Prediction Error through Subspace Decomposition and Influence Analysis
Authors: Enze Shi, Pankaj Bhagwat, Zhixian Yang, Linglong Kong, Bei Jiang
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments on both synthetic and real-world datasets validate our theoretical insights and show that the proposed method effectively improves fairness while preserving predictive performance. |
| Researcher Affiliation | Academia | Enze Shi, Pankaj Bhagwat, Zhixian Yang, Linglong Kong, Bei Jiang Department of Mathematical and Statistical Science University of Alberta EMAIL |
| Pseudocode | Yes | Algorithm 1 Estimation of Sufficient Projections Algorithm 2 Sequential Fair Projection: Post-SDR Fair Model Training |
| Open Source Code | Yes | Furthermore, we provide the code necessary to reproduce our results in the supplementary materials. All datasets used are publicly available, and we have uploaded code in the supplementary materials. |
| Open Datasets | Yes | We evaluate our proposed Sequential Fair Projection (SFP) method on two tabular datasets: Adult Kohavi [1996] and Bank Moro et al. [2014]. |
| Dataset Splits | Yes | The dataset is randomly split into 4000 training samples and 1000 testing samples. The data is standardized during preprocessing and split into training, validation, and test sets with a ratio of 70%:10%:20%. |
| Hardware Specification | Yes | We conduct all our experiments on an Ubuntu Server with CPU AMD Ryzen Threadripper PRO 3995WX 64-Cores Processor and 256G RAM. |
| Software Dependencies | No | The paper mentions using the MSAVE method and applying multivariate linear regression or logistic regression. It also mentions using official code from other authors for baselines. However, it does not specify version numbers for any key software components or libraries (e.g., Python, PyTorch, scikit-learn versions) used in their own implementation. |
| Experiment Setup | No | The paper states using multivariate linear regression for synthetic simulations and logistic regression for real-world datasets, and that the MCDP metric is used for model selection. It also provides some setup for baselines (INLP iterations, RLACE hyperparameters, SUP cross-validation). However, it does not explicitly state concrete hyperparameters like learning rate, batch size, or number of epochs for the logistic regression or linear regression models used in their SFP method. |