Abstract:Graph diffusion models achieve state-of-the-art performance in graph generation but suffer from quadratic complexity in the number of nodes -- and much of their capacity is wasted modeling the absence of edges in sparse graphs. Inspired by latent diffusion in other modalities, a natural idea is to compress graphs into a low-dimensional latent space and perform diffusion in that space. However, unlike images or text, graph generation requires nearly lossless reconstruction, as even a single error in decoding an adjacency matrix can render the entire sample invalid. This challenge has remained largely unaddressed. We propose LG-Flow, a latent graph diffusion framework that directly overcomes these obstacles. A permutation-equivariant autoencoder maps nodes to fixed-dimensional embeddings that enable near-lossless reconstruction of both undirected graphs and DAGs. The dimensionality of this latent representation scales linearly with the number of nodes, thereby removing the quadratic adjacency-space bottleneck in the diffusion process and enabling the training of substantially larger generative backbones. In this latent space, we train a Diffusion Transformer with flow matching, enabling efficient and expressive graph generation. Our approach achieves competitive results against state-of-the-art graph diffusion models while delivering up to a $1000\times$ speed-up. Our code is available at this https URL .
| Comments: | Preprint, under review |
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2601.13780 [cs.LG] |
| (or arXiv:2601.13780v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2601.13780 arXiv-issued DOI via DataCite |
From: Antoine Siraudin [view email]
[v1]
Tue, 20 Jan 2026 09:37:53 UTC (359 KB)
[v2]
Wed, 25 Feb 2026 14:32:51 UTC (354 KB)
[v3]
Tue, 12 May 2026 08:38:31 UTC (349 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。