Fair Submodular Cover
Authors: Wenjing Chen, Shuo Xing, Samson Zhou, Victoria Crawford
ICLR 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Finally, we complement our theoretical results with a number of empirical evaluations that demonstrate the efficiency of our algorithms on instances of maximum coverage. ... In this section, we evaluate several of our algorithms for FSC on instances of fair maximum coverage in a graph and fair image summarization. ... Figures 1a, 1b and 1c showcase the distribution of users speaking different languages in the solutions produced by these algorithms with τ = 2400. Figures 1d, 1e and 1f present the performance of these algorithms (f value, cost, and fairness difference) for varying values of τ. |
| Researcher Affiliation | Academia | Texas A&M University, EMAIL, EMAIL, EMAIL, EMAIL |
| Pseudocode | Yes | Algorithm 1 convert-fair ... Algorithm 2 cont-thresh-greedy-bi (cont-bi) ... Algorithm 3 decreasing-threshold-procedure (DTP) ... Algorithm 4 convert-continuous ... Algorithm 5 greedy-fairness-bi ... Algorithm 6 threshold-fairness-bi ... Algorithm 7 greedy-bi |
| Open Source Code | No | The paper does not contain any explicit statement about releasing source code or provide links to a code repository. |
| Open Datasets | Yes | The dataset utilized is a subset of the Twitch Gamers dataset (Rozemberczki & Sarkar, 2021)... The dataset employed in our set covering experiments is a subset of Corel5k Duygulu et al. (2002). |
| Dataset Splits | No | The paper mentions using specific datasets (Twitch Gamers, Corel5k) but does not provide details on how these datasets were split into training, validation, or test sets, nor does it refer to standard splits with citations. |
| Hardware Specification | Yes | All the experiments are conducted on a single machine equipped with a 13th Gen Intel(R) Core(TM) i7-13700 CPU, 32GB of RAM, and Ubuntu 22.04.3 LTS. |
| Software Dependencies | Yes | All the experiments are conducted on a single machine equipped with a 13th Gen Intel(R) Core(TM) i7-13700 CPU, 32GB of RAM, and Ubuntu 22.04.3 LTS. |
| Experiment Setup | Yes | To ensure a fair comparison based on the quality of the solutions, we use different default values for the parameter ε in each algorithm. ... Specifically, we set ε = 0.1, α = 0.2, uc = 1.1/C, lc = 0.9/C for greedy-bi and greedy-fair-bi (where C is the number of groups). For threshold-fairness-bi, we use ε = 0.05 while keeping the other parameters the same. |