One-Shot Compression of Large Edge-Exchangeable Graphs using Bits-Back Coding

Authors: Daniel Severo, James Townsend, Ashish J Khisti, Alireza Makhzani

ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we showcase the optimality of REC on large graphs representing real-world networks. We entropy code with REC using P olya s Urn (PU) model and compare the performance to state-of-the-art compression algorithms tailored to network compression. We report the average number of bits required to represent an edge in the graph (i.e., bits-per-edge) as is common in the literature.
Researcher Affiliation Academia 1Department of Electrical and Computer Engineering, University of Toronto 2Vector Institute for Artificial Inteligence 3Amsterdam Machine Learning Lab (AMLab), University of Amsterdam.
Pseudocode Yes Algorithm 1 Naive Random Edge Encoder and Algorithm 2 Random Edge Encoder
Open Source Code Yes Code implementing Random Edge Coding, P olya s Urn model, and experiments are available at https:// github.com/dsevero/Random-Edge-Coding.
Open Datasets Yes We used datasets containing simple network graphs with small-world statistics (see Section 4.1) such as You Tube, Four Square, Gowalla, and Digg (Rossi & Ahmed, 2015)... Skitter and DBLP networks (Leskovec & Krevl, 2014)
Dataset Splits No The paper describes compressing the entire graph in a one-shot manner and does not specify traditional train/validation/test dataset splits.
Hardware Specification No The paper does not explicitly describe the hardware used for running experiments, such as specific GPU or CPU models.
Software Dependencies No We use the ANS implementation available in Craystack (Townsend et al., 2020; 2021). However, specific version numbers for Craystack or other software dependencies are not provided.
Experiment Setup No The paper describes the general process of Random Edge Coding and Polya's Urn model, which is parameter-free, but does not provide specific experimental setup details like hyperparameters or training configurations for a learning model.