An Efficient Evolutionary Algorithm for Minimum Cost Submodular Cover

Authors: Victoria G. Crawford

IJCAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In a practical application, EASC is demonstrated to outperform the greedy algorithm and converge faster than competing evolutionary algorithms for this problem. EASC is experimentally evaluated on instances of the Influence Threshold Problem (IT) [Goyal et al., 2013; Kuhnle et al., 2017] on real social network datasets.
Researcher Affiliation Academia Victoria G. Crawford Department of Computer and Information Science and Engineering, University of Florida, United States vcrawford01@ufl.edu
Pseudocode Yes Pseudocode for EASC can be found in Algorithm 1.
Open Source Code Yes Code to run the experiments is publicly available at https://gitlab.com/vcrawford/easc.git.
Open Datasets Yes The experiments are run on four real social networks from SNAP [Leskovec and Krevl, 2015]: ca-Gr Qc (n = 5242), ca-Hep Ph (n = 12008), wiki-Vote (n = 7115), and ego Facebook (n = 4039).
Dataset Splits No No specific training, validation, or test dataset splits (e.g., percentages, sample counts, or cross-validation setup) are explicitly mentioned in the paper.
Hardware Specification No No specific hardware details such as GPU models, CPU types, or memory configurations used for running experiments are provided.
Software Dependencies No The paper does not provide specific software dependency versions (e.g., library or solver names with version numbers) used in the experiments.
Experiment Setup Yes EASC is run with τ, ϵ = 0.05, and δ = 1 cmin/B where B is the cost of the output of the greedy algorithm when run with τ and ϵ = 0. POM is run with threshold (1 ϵ)τ for fair comparison with EASC. The independent cascade model is used to model activation from a seed set for the above four social networks with constant edge probabilities p = 0.07, p = 0.02, p = 0.04, and p = 0.013, respectively. The reverse influence sampling approach with 100,000 samples is used.