Flowette: Flow Matching with Graphette Priors for Graph Generation

2026-02-27

AI Summary

This work addresses the challenge of generating recurrent subgraph motifs—such as cycles, stars, and trees—in complex graph structures by proposing Flowette, a novel framework that effectively models long-range topological dependencies while preserving structural consistency. Flowette integrates a graph neural network Transformer with continuous flow matching, leveraging optimal transport couplings to maintain global topology. It further introduces graphette, a new probabilistic graph model serving as a structural prior, enabling controlled editing and generalization of subgraph motifs. Experimental results demonstrate that Flowette significantly outperforms existing baselines on both synthetic graph generation and small-molecule synthesis tasks, thereby validating the efficacy and superiority of combining structural priors with flow matching for modeling complex graph distributions.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Reasoning under Uncertainty: Graphical Models
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

We study generative modeling of graphs with recurring subgraph motifs. We propose Flowette, a continuous flow matching framework, that employs a graph neural network based transformer to learn a velocity field defined over graph representations with node and edge attributes. Our model preserves topology through optimal transport based coupling, and long-range structural dependencies through regularisation. To incorporate domain driven structural priors, we introduce graphettes, a new probabilistic family of graph structure models that generalize graphons via controlled structural edits for motifs like rings, stars and trees. We theoretically analyze the coupling, invariance, and structural properties of the proposed framework, and empirically evaluate it on synthetic and small-molecule graph generation tasks. Flowette demonstrates consistent improvements, highlighting the effectiveness of combining structural priors with flow-based training for modeling complex graph distributions.

Problem

• graph generation
• subgraph motifs
• structural priors
• graph representation
• complex graph distributions

Innovation

• flow matching
• graphettes
• graph generation
• structural priors
• optimal transport