AUTHOR=Zhou Dawei , Zheng Lecheng , Xu Jiejun , He Jingrui 

TITLE=Misc-GAN: A Multi-scale Generative Model for Graphs

JOURNAL=Frontiers in Big Data

VOLUME=Volume 2 - 2019

YEAR=2019

URL=https://www.frontiersin.org/journals/big-data/articles/10.3389/fdata.2019.00003

DOI=10.3389/fdata.2019.00003

ISSN=2624-909X

ABSTRACT=Characterizing and modeling the distribution of a particular family of graphs are essential for3studying real-world networks in a broad spectrum of disciplines, ranging from market-basket4analysis to biology, from social science to neuroscience. However, it is unclear how to model5the complex graph organizations and learn generative models from an observed graph. The6key challenges come from the non-unique, high-dimensional nature of graphs, as well as the7graph community structures at different granularity levels. In this paper, we propose a multi-scale8graph generative model namedMisc-GAN, which models the underlying distribution of the graph9structures at different levels of granularity, and then ‘transfers’ such hierarchical distribution from10the graphs in the domain of interest to a unique graph representation. The empirical results on11both synthetic and real data sets demonstrate the effectiveness of the proposed framework.