arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2406.09357cs.LGstat.ML
收藏

通过Beta扩散推进图生成

Advancing Graph Generation through Beta Diffusion

The University of Texas at Austin(德克萨斯大学奥斯汀分校);Xidian University(西安电子科技大学)

Xinyang Liu, Yilin He, Bo Chen, Mingyuan Zhou

AI总结:

针对传统扩散模型难适配图数据离散连续混合特性的问题,提出基于beta扩散的图生成模型GBD及调制技术,在多基准测试中表现优异,可更好平衡图的离散与连续特征。

摘要:

扩散模型在自然图像生成领域表现出色,目前正被适配应用于包括图在内的多种数据类型。然而,传统模型通常依赖高斯或分类扩散过程,难以适配图数据所特有的离散与连续混合成分。图通常具备离散结构和连续节点属性,这些属性往往呈现出丰富的统计模式,包括稀疏性、有界范围、偏态分布和长尾特性。为应对这些挑战,我们提出了Graph Beta Diffusion(GBD,图Beta扩散),这是一款专门为处理图数据的多样特性而设计的生成模型。GBD利用beta扩散过程,可有效对连续和离散元素进行建模。此外,我们提出了一种调制技术,通过稳定关键的图拓扑结构,同时保持其他成分的灵活性,来提升生成图的真实感。GBD在多个通用和生化图基准测试中与现有模型表现相当,展现出其捕捉真实世界图数据中固有离散与连续特征间复杂平衡的能力。其PyTorch代码已在GitHub上公开。

英文摘要:

Diffusion models have excelled in generating natural images and are now being adapted to a variety of data types, including graphs. However, conventional models often rely on Gaussian or categorical diffusion processes, which can struggle to accommodate the mixed discrete and continuous components characteristic of graph data. Graphs typically feature discrete structures and continuous node attributes that often exhibit rich statistical patterns, including sparsity, bounded ranges, skewed distributions, and long-tailed behavior. To address these challenges, we introduce Graph Beta Diffusion (GBD), a generative model specifically designed to handle the diverse nature of graph data. GBD leverages a beta diffusion process, effectively modeling both continuous and discrete elements. Additionally, we propose a modulation technique that enhances the realism of generated graphs by stabilizing critical graph topology while maintaining flexibility for other components. GBD competes strongly with existing models across multiple general and biochemical graph benchmarks, showcasing its ability to capture the intricate balance between discrete and continuous features inherent in real-world graph data. The PyTorch code is available on GitHub.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑