Diffusion Twigs with Loop Guidance for Conditional Graph Generation

TL;DR

The paper presents Twigs, a diffusion-based framework with multiple co-evolving flows and loop guidance for improved conditional graph generation, enabling better modeling of complex dependencies in tasks like molecular design.

Contribution

Introduction of Twigs, a novel diffusion framework with loop guidance that models interactions between primary and dependent variables in conditional graph generation.

Findings

Outperforms existing baselines in conditional graph generation tasks.

Effective in inverse molecular design and molecular optimization.

Demonstrates strong performance gains across experiments.

Abstract

We introduce a novel score-based diffusion framework named Twigs that incorporates multiple co-evolving flows for enriching conditional generation tasks. Specifically, a central or trunk diffusion process is associated with a primary variable (e.g., graph structure), and additional offshoot or stem processes are dedicated to dependent variables (e.g., graph properties or labels). A new strategy, which we call loop guidance, effectively orchestrates the flow of information between the trunk and the stem processes during sampling. This approach allows us to uncover intricate interactions and dependencies, and unlock new generative capabilities. We provide extensive experiments to demonstrate strong performance gains of the proposed method over contemporary baselines in the context of conditional graph generation, underscoring the potential of Twigs in challenging generative tasks such as inverse molecular design and molecular optimization.