Original title: A Network Science Perspective on Evaluating Deep Graph Generative Models
Tianrui Mao, Abele Malan, Megha Khosla, Lydia Chen +1 · September 2026
Scientists need realistic social networks to study how diseases or misinformation spread, but privacy rules keep real human contact data private. Older mathematical tools create artificial networks, but they miss many real-world patterns. To solve this, researchers tested deep graph generative modelsAI systems that use neural networks to learn and generate new networks based on data from real ones.. These computer systems use deep neural networks to learn directly from real network structures and create realistic synthetic networks. The researchers tested these deep models alongside a traditional configuration modelA traditional network model that generates random graphs while preserving specific properties like node degrees. from a network scienceThe study of complex networks, like social connections, focusing on patterns and properties. perspective. They checked how closely the fake networks matched real structures and tested if they could help find which key individuals to immunize. They found that two deep graph generative models generated synthetic networks that closely matched real-world structures, successfully identifying effective immunization strategies.
Imagine planning a large school event where you need to predict how students will interact to prevent the spread of a cold. But you cannot ask students about their private friendships due to school rules. You create a fake student list that copies the real friendship patterns without revealing actual names. In this paper, the real social networks are the actual friendships, and the deep learning models create synthetic versions that preserve the key interaction patterns. This allows researchers to test immunization strategies, like deciding which students to give extra hand sanitizer to, without breaking privacy rules.
Real social contact networks cannot be shared due to privacy risks, but researchers need them to develop and test epidemic mitigation strategies. This work shows that AI-generated synthetic networks are realistic enough to be used as alternatives, enabling public health planning without compromising individual privacy.
The abstract only evaluates two deep graph generative models and the configuration model, and does not specify which models or how many were tested. It focuses on topological similarity and utility for immunization strategies, but does not address other potential applications or validate findings against real-world epidemic data.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
A Network Science Perspective on Evaluating Deep Graph Generative Models — Tianrui Mao, Abele Malan, Megha Khosla et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.