The Secretary Problem with Predicted Additive Gap

Authors: Alexander Braun, Sherry Sarkar

NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Complementing theoretical results, we run simulations in Section 6 which strengthen our theoretical findings.
Researcher Affiliation Academia Institute of Computer Science, University of Bonn. Email: alexander.braun@uni-bonn.de Carnegie Mellon University. Email: sherrys@andrew.cmu.edu
Pseudocode Yes Algorithm 1 Secretary with Exact Additive Gap
Open Source Code No We will make the code and data publicly available after the review process.
Open Datasets No The paper describes generating data by drawing weights i.i.d. from distributions (Pareto, Exponential, Chi-Squared) but does not provide concrete access information (link, DOI, citation) to a pre-existing, publicly available dataset.
Dataset Splits No The paper describes generating data for simulations (drawing weights i.i.d.) and does not mention explicit validation, training, or test splits of a pre-existing dataset. The NeurIPS checklist also states 'NA' for this question.
Hardware Specification Yes All experiments were implemented in Python 3.9 and executed on a machine with Apple M1 and 8 GB Memory.
Software Dependencies Yes All experiments were implemented in Python 3.9 and executed on a machine with Apple M1 and 8 GB Memory.
Experiment Setup Yes For each class of instances, we average over 5000 iterations. In each iteration, we draw n = 200 weights i.i.d. from the respective distribution together with 200 arrival times which are drawn i.i.d. from Unif[0, 1].