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 [1].
Fair Sequential Selection Using Supervised Learning Models
Authors: Mohammad Mahdi Khalili, Xueru Zhang, Mahed Abroshan
NeurIPS 2021 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The experiments on real-world datasets validate the theoretical results. |
| Researcher Affiliation | Academia | Mohammad Mahdi Khalili CIS Department University of Delaware Newark, DE, USA EMAIL Xueru Zhang CSE Department Ohio State University Columbus, OH, USA EMAIL Mahed Abroshan Alan Turing Institute London, UK EMAIL |
| Pseudocode | No | The paper describes algorithms using mathematical formulations and text, but does not provide structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | The codes are available here. |
| Open Datasets | Yes | FICO credit score dataset [41]. Adult income dataset [42]. |
| Dataset Splits | No | The paper mentions sample sizes and uses pre-trained models but does not explicitly specify train/validation/test splits, percentages, or absolute counts for dataset partitioning. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running the experiments. |
| Software Dependencies | No | We first train a logistic regression classifier (using sklearn package and default parameters) as the pretrained model. The paper mentions 'sklearn package' but does not specify a version number for it or any other software dependency. |
| Experiment Setup | Yes | We consider a common method where the decisions are made based on a threshold rule, i.e., selecting an applicant if its qualification score R = r(X, A) is above a threshold τ. ... the optimal thresholds τ0, τ1 in Table 1 are close to the maximum score 100, especially under EO and SP fairness notions. ... we add the following time constraint to optimization (13): the probability that no applicant is selected after 100 time steps should be less than 1. |