A Consistent Histogram Estimator for Exchangeable Graph Models

Authors: Stanley Chan, Edoardo Airoldi

ICML 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We test the SAS algorithm on both simulation data and real data (Section 5). The experiment of using the simulation data indicates that the SAS algorithm is superior to, both in terms of estimation quality and speed, several existing methods. Applying the SAS algorithm to real data, we estimate graphons of two large-scale social networks and reveal some structures. These results provide an alternative way of analyzing large-scale network data.
Researcher Affiliation Academia Stanley H. Chan SCHAN@SEAS.HARVARD.EDU School of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138, USA Edoardo M. Airoldi AIROLDI@FAS.HARVARD.EDU Department of Statistics, Harvard University, Cambridge, MA 02138, USA
Pseudocode No The paper describes the steps of the SAS algorithm textually and with a flowchart figure (Figure 2), but it does not provide a formal pseudocode or algorithm block.
Open Source Code Yes Code. Available at: https://github.com/airoldilab/SAS
Open Datasets Yes For this purpose, we consider the collaboration network of ar Xiv astro physics (ca-Astro Ph) and the who-trusts-whom network of Epinions.com (soc Epinions1) from Stanford Large Network Dataset Collection. 1http://www.cise.ufl.edu/research/sparse/matrices/SNAP/
Dataset Splits No The paper does not explicitly provide details about training, validation, or test dataset splits (e.g., percentages, sample counts, or specific split files/methods). It mentions '50 independent trials' for simulations, but not formal data splits.
Hardware Specification Yes Both algorithms are implemented on an Intel 3.5GHz machine with 16GB RAM, Windows 7 / MATLAB R7.12.0 platform.
Software Dependencies Yes Both algorithms are implemented on an Intel 3.5GHz machine with 16GB RAM, Windows 7 / MATLAB R7.12.0 platform.
Experiment Setup Yes For the choice of binwidth h, we set h = log n for the SAS algorithm, and an oracle h that minimizes the MSE for the SBA algorithm (i.e., using the ground truth).