Fine-Grained Graph Generation through Latent Mixture Scheduling

2026-05-04

AI Summary

Existing graph generation methods struggle to achieve fine-grained control over topological properties. To address this limitation, this work proposes a novel framework based on conditional variational autoencoders that dynamically aligns graph structures with attribute-driven latent representations. During decoding, a hybrid scheduling mechanism is introduced to progressively integrate graph priors with control priors, thereby enhancing both fidelity and controllability of the generated graphs. Experimental results across five real-world datasets demonstrate that the proposed method significantly outperforms recent baselines, achieving state-of-the-art performance while simultaneously preserving high-quality generation and enabling precise control over structural attributes.

Abstract

Structure aware graph generation aims to generate graphs that satisfy given topological properties. It has applications in domains such as drug discovery, social network modeling, and knowledge graph construction. Unlike existing methods that only provide coarse control over graph properties, we introduce a novel conditional variational autoencoder for fine-grained structural control in graph generation. The approach refines the decoder's latent space by dynamically aligning graph- and property-driven representations to improve both graph fidelity and control satisfaction. Specifically, the approach implements a mixture scheduler that progressively integrates graph and control priors. Experiments on five real-world datasets show the efficacy of the proposed model compared to recent baselines, achieving high generation quality while maintaining high controllability.

Problem

• graph generation
• fine-grained control
• topological properties
• structure-aware generation
• conditional graph modeling

Innovation

• fine-grained graph generation
• conditional variational autoencoder
• latent mixture scheduling
• structure-aware generation
• graph controllability