Graph generation model predicts clean graphs with faster sampling
Efficient Graph Generation via Direct Prediction and Flow Matching
Machine Learning
Summary
Generating graphs that reflect real-world connections is important in many areas, but existing methods are slow and often don’t capture important graph details well. The authors introduce DiGFM, a new approach that directly predicts clean graphs instead of just noise. This method produces graphs faster—up to 257 times faster—and still matches or beats previous models in quality. This means DiGFM can quickly make realistic graphs, useful for things like modeling molecules or social networks.
What this means in practice
- •For pharmaceutical researchers: Generate molecular graphs quickly while maintaining realistic chemical structure patterns for drug discovery simulations.
- •For data scientists: Create synthetic social network graphs that accurately reflect real-world connectivity and degree distribution much faster than before.
Authors
Susie Lu
Abstract
Generative modeling of graph-structured data is crucial for tasks ranging from drug discovery to social network simulation. Among these models, denoising diffusion models have achieved great success in graph generation by learning to progressively reverse a process that adds noise to the original graph. However, the standard noise-prediction approach of diffusion models is suboptimal for graph data. The goal for a graph generative model is to learn the clean graphs' topological properties, such as connectivity and degree distribution. Because a diffusion model that predicts noise does not explicitly learn these topological properties, it is challenging for the model to output graphs with the desired structural statistics. To address this challenge, we introduce Direct Graph Flow Matching (DiGFM), a novel graph transformer model guided by two goals: predict clean graphs and improve sampling efficiency. Distinct from the prevailing diffusion approach, DiGFM employs a continuous flow-matching paradigm and integrates direct graph prediction. Specifically, DiGFM maps the prior noise distribution to the clean graph distribution via a multi-step process: the model repeatedly predicts the underlying clean graph, and a transformation is employed to convert the model output to the velocity vector that points in the direction toward the clean graph distribution. This design enables DiGFM to generate high-quality samples using only 2.5% to 15.6% of the steps required by diffusion-based models, which leads to a 5.3x to 257x speedup in wall-clock inference time. Experiments demonstrate that DiGFM outperforms or matches prior state-of-the-art models across general graph benchmarks and molecular datasets, generating graphs with strong adherence to ground-truth structural statistics at significantly faster inference speeds.