arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2410.04263cs.LG
收藏

DeFoG:面向图生成的离散流匹配

DeFoG: Discrete Flow Matching for Graph Generation

EPFL, Lausanne, Switzerland(苏黎世联邦理工学院)

Yiming Qin, Manuel Madeira, Dorina Thanou, Pascal Frossard

AI总结:

针对图扩散模型训练采样耦合导致的低效问题,提出解耦式离散流匹配图生成框架DeFoG,理论验证其分布还原能力,设计新型采样方法,在多数据集多生成设置下实现SOTA且采样步数远少于扩散模型。

摘要:

图生成模型通过捕捉关系数据上的复杂分布,在众多科学领域中发挥着关键作用。其中,图扩散模型虽性能优异,但因训练与采样阶段紧密耦合,存在采样效率低、灵活性受限的问题。我们提出DeFoG,一种新型图生成框架,它将采样与训练解耦,为更高效、更有效的模型优化开辟了更广阔的设计空间。DeFoG采用尊重图固有对称性的离散流匹配公式。我们通过将训练损失与采样算法明确关联,从理论上论证了这种解耦公式的合理性,并证明DeFoG能够忠实还原真实图分布。在此基础上,我们全面探究了DeFoG的设计空间,提出了新型采样方法,显著提升了性能并减少了所需的细化步骤数。大量实验表明,DeFoG在合成数据集、分子数据集和数字病理数据集上均取得了当前最优性能,覆盖无条件生成和条件生成两种设置。仅用扩散模型5%-10%的采样步数,它的表现就优于大多数基于扩散的模型。

英文摘要:

Graph generative models are essential across diverse scientific domains by capturing complex distributions over relational data. Among them, graph diffusion models achieve superior performance but face inefficient sampling and limited flexibility due to the tight coupling between training and sampling stages. We introduce DeFoG, a novel graph generative framework that disentangles sampling from training, enabling a broader design space for more effective and efficient model optimization. DeFoG employs a discrete flow-matching formulation that respects the inherent symmetries of graphs. We theoretically ground this disentangled formulation by explicitly relating the training loss to the sampling algorithm and showing that DeFoG faithfully replicates the ground truth graph distribution. Building on these foundations, we thoroughly investigate DeFoG's design space and propose novel sampling methods that significantly enhance performance and reduce the required number of refinement steps. Extensive experiments demonstrate state-of-the-art performance across synthetic, molecular, and digital pathology datasets, covering both unconditional and conditional generation settings. It also outperforms most diffusion-based models with just 5-10% of their sampling steps.

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。

↑