Graph Diffusion Policy Optimization

TL;DR

This paper introduces GDPO, a reinforcement learning-based method to optimize graph diffusion models for complex, non-differentiable objectives, achieving state-of-the-art results in graph generation tasks.

Contribution

GDPO is a novel policy gradient approach specifically designed for graph diffusion models, enabling optimization for arbitrary objectives.

Findings

Achieves state-of-the-art performance in graph generation tasks

Effectively handles non-differentiable objectives

Improves over existing diffusion model optimization methods

Abstract

Recent research has made significant progress in optimizing diffusion models for downstream objectives, which is an important pursuit in fields such as graph generation for drug design. However, directly applying these models to graph presents challenges, resulting in suboptimal performance. This paper introduces graph diffusion policy optimization (GDPO), a novel approach to optimize graph diffusion models for arbitrary (e.g., non-differentiable) objectives using reinforcement learning. GDPO is based on an eager policy gradient tailored for graph diffusion models, developed through meticulous analysis and promising improved performance. Experimental results show that GDPO achieves state-of-the-art performance in various graph generation tasks with complex and diverse objectives. Code is available at https://github.com/sail-sg/GDPO.