Inferring Dynamic Networks from Marginals with Iterative Proportional Fitting

Authors: Serina Chang, Frederic Koehler, Zhaonan Qu, Jure Leskovec, Johan Ugander

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

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, we conduct experiments with synthetic and realworld data, which demonstrate the practical value of our theoretical and algorithmic contributions.
Researcher Affiliation Academia 1Department of Computer Science, Stanford University 2Department of Statistics and Data Science Institute, University of Chicago 3Department of Economics, Stanford University 4Department of Management Science & Engineering, Stanford University.
Pseudocode Yes Algorithm 1 Our implementation of the iterative proportional fitting procedure.
Open Source Code Yes Our code is available at https://github.com/ snap-stanford/ipf-network-inference.
Open Datasets Yes Citibike data (Citi Bike, 2023) and Safe Graph mobility data (Dewey, 2023) are available online.
Dataset Splits No The paper does not provide explicit training, validation, and test dataset splits with proportions or sample counts. It describes experiments on entire datasets (synthetic or real-world over specific time periods) rather than predefined splits.
Hardware Specification No The paper does not mention any specific hardware (e.g., GPU, CPU models, cloud computing resources) used for running the experiments.
Software Dependencies No The paper mentions 'Python s statsmodels package' but does not provide a specific version number for this or any other software dependency.
Experiment Setup Yes We sample the row scaling factors eu Rm and column scaling factors e v Rn from Uniform(0, 4). We sample X Rm n from Uniform(0, 1). For a given sparsity level r [0, 1), we randomly select r mn entries from X (without replacement) and set them to 0. For each sparsity rate in r {0, 0.05, , 0.9}, we run 1000 random trials.