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..
Non-Clairvoyant Scheduling with Progress Bars
Authors: Ziyad Benomar, Romain Cosson, Alexander Lindermayr, Jens Schlöter
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We now present empirical experiments that validate our theoretical results. In the experiments, we consider instances with n = 500 jobs, where processing times are sampled independently from a Pareto distribution with parameter 1.1. Each point in the figures represents an average over at least 20 independent trials, with standard deviation indicated. In this section, we give a short overview of our findings. An extensive description of our setup and results is given in Appendix D. (Section 5 Experiments) |
| Researcher Affiliation | Academia | Ziyad Benomar CREST, ENSAE, Ecole Polytechnique, Fairplay joint team, Palaiseau, France EMAIL; Romain Cosson Courant Institute of Mathematical Sciences New York University EMAIL; Alexander Lindermayr Simons Institute for the Theory of Computing UC Berkeley EMAIL; Jens Schlöter Centrum Wiskunde & Informatica (CWI) Amsterdam, The Netherlands EMAIL |
| Pseudocode | Yes | Algorithm 1: (1 + α)-consistent, (1 + 1/αρ)-robust algorithm; Algorithm 2: Combining multiple algorithms A(1), . . . , A(g) with computable delays; Algorithm 3: Repeated Explore-Then-Commit with threshold k |
| Open Source Code | Yes | Question: Does the paper provide open access to the data and code, with sufficient instructions to faithfully reproduce the main experimental results, as described in supplemental material? Answer: [Yes] Justification: The code is included in the supplementary material (NeurIPS Paper Checklist, Question 5) |
| Open Datasets | No | In the experiments, we consider instances with n = 500 jobs, where processing times are sampled independently from a Pareto distribution with parameter 1.1. (Section 5 Experiments) |
| Dataset Splits | No | In the experiments, we consider instances with n = 500 jobs, where processing times are sampled independently from a Pareto distribution with parameter 1.1. Each point in the figures represents an average over at least 20 independent trials, with standard deviation indicated. (Section 5 Experiments) |
| Hardware Specification | No | Question: For each experiment, does the paper provide sufficient information on the computer resources (type of compute workers, memory, time of execution) needed to reproduce the experiments? Answer: [NA] Justification: The experiments presented in the paper are simulations of scheduling algorithms with advice and do not involve training or testing any machine learning models. Reporting of computing resources is not necessary. (NeurIPS Paper Checklist, Question 8) |
| Software Dependencies | No | The paper does not list specific software libraries or version numbers used for the simulations. The experiments are described as simulations of scheduling algorithms. |
| Experiment Setup | Yes | In the experiments, we consider instances with n = 500 jobs, where processing times are sampled independently from a Pareto distribution with parameter 1.1. Each point in the figures represents an average over at least 20 independent trials, with standard deviation indicated. In Figures 2a and 2b, the predictions (πi)i are noisy estimates of the true job sizes pi, with independent Gaussian noise: πi N(pi, σ). Figure 2a shows the behavior of Algorithm 1, with α = 0.5, for various values of ρ, where signal emission times are computed as βi = max(0, min(1, πi/pi)). The first two use tuned hyperparameters to ensure the same robustness level. Finally, in Figure 2c, we compare the performance of Algorithm 3 (with k = 1 and k = Θ(g2/3)). (Section 5 Experiments) |