Feedback Efficient Online Fine-Tuning of Diffusion Models

TL;DR

This paper introduces a reinforcement learning method for efficiently fine-tuning diffusion models to generate high-quality samples with desired properties, with theoretical guarantees and empirical validation across multiple domains.

Contribution

It proposes a novel RL approach that explores feasible sample manifolds efficiently, with theoretical regret bounds and validation on images, sequences, and molecules.

Findings

Achieves efficient exploration of high-reward samples

Provides theoretical regret guarantees

Validates effectiveness across diverse domains

Abstract

Diffusion models excel at modeling complex data distributions, including those of images, proteins, and small molecules. However, in many cases, our goal is to model parts of the distribution that maximize certain properties: for example, we may want to generate images with high aesthetic quality, or molecules with high bioactivity. It is natural to frame this as a reinforcement learning (RL) problem, in which the objective is to fine-tune a diffusion model to maximize a reward function that corresponds to some property. Even with access to online queries of the ground-truth reward function, efficiently discovering high-reward samples can be challenging: they might have a low probability in the initial distribution, and there might be many infeasible samples that do not even have a well-defined reward (e.g., unnatural images or physically impossible molecules). In this work, we propose a novel reinforcement learning procedure that efficiently explores on the manifold of feasible samples. We present a theoretical analysis providing a regret guarantee, as well as empirical validation across three domains: images, biological sequences, and molecules.