arXiv cs.AI· Michal Balcerak, Suprosanna Shit, Chinmay Prabhakar, Sebastian Kaltenbach, Michael S. Albergo, Yilun Du, Bjoern Menze·· 10 小时前AI 评分29
Graph Energy Matching：面向图生成的传输对齐能量建模框架
Graph Energy Matching: Transport-Aligned Energy-Based Modeling for Graph Generation
AI 导读
研究者提出 Graph Energy Matching（GEM），一种受 JKO 传输映射优化启发的离散生成框架，通过学习置换不变的能量函数引导噪声向高似然图区域传输并细化样本。在分子图基准上，GEM 在多数指标上匹配或超越强离散扩散基线，并支持组合生成、性质约束采样与图间插值。该工作为 NeurIPS 2026 定稿版本。
正文 · 原文
Abstract:Generative modeling of discrete data, such as graphs, underpins many scientific and industrial applications, including molecular discovery and materials design. In these domains, probabilistic inference is particularly valuable, as it enables composable generation and principled incorporation of desired constraints, such as structural or functional properties. Energy-based models naturally support this goal by capturing relative likelihoods and enabling composable inference by directly enforcing constraints during inference. However, discrete energy-based models typically struggle with efficient and high-quality sampling, as off-support regions often contain spurious local minima, trapping samplers and causing training instabilities, resulting in a fidelity gap compared to discrete diffusion models. To address this gap, we introduce Graph Energy Matching (GEM), a discrete generative framework inspired by the Jordan-Kinderlehrer-Otto (JKO) transport-map optimization perspective. GEM learns a permutation-invariant potential energy that simultaneously guides discrete transport from noise toward high-likelihood graph regions and refines samples within these regions. We further introduce a sampling protocol leveraging an energy-based switching strategy, seamlessly bridging rapid, gradient-guided transport and a local mixing regime for effective exploration. On molecular graph benchmarks, GEM matches or surpasses strong discrete diffusion baselines on most reported metrics. Beyond improving generation quality, GEM's relative likelihood modeling enables targeted exploration, facilitating compositional generation, property-constrained sampling, and interpolation between graphs. Project page: this https URL.
| Comments: | NeurIPS 2026 camera-ready version. 29 pages |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML) |
| Cite as: | arXiv:2603.23398 [cs.LG] |
| (or arXiv:2603.23398v4 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2603.23398 arXiv-issued DOI via DataCite |
Submission history
From: Michal Balcerak [view email]
[v1]
Tue, 24 Mar 2026 16:35:25 UTC (763 KB)
[v2]
Thu, 28 May 2026 22:03:51 UTC (371 KB)
[v3]
Mon, 1 Jun 2026 12:59:54 UTC (371 KB)
[v4]
Mon, 9 Nov 2026 23:34:32 UTC (384 KB)
来源：arXiv cs.AI · arxiv.org