Efficient Controllable Diffusion via Optimal Classifier Guidance
This work addresses the need for efficient and resource-light controllable generation in applications like image and molecule generation, offering a method that avoids the overfitting and high costs of RL-based approaches.
The paper tackles the problem of controllable generation in diffusion models by framing it as a KL-regularized optimization and proposes SLCD, a supervised learning method that trains a classifier to guide generation, which provably converges to the optimal solution and maintains high-quality samples with nearly the same inference time as the base model in image and biological sequence generation.
The controllable generation of diffusion models aims to steer the model to generate samples that optimize some given objective functions. It is desirable for a variety of applications including image generation, molecule generation, and DNA/sequence generation. Reinforcement Learning (RL) based fine-tuning of the base model is a popular approach but it can overfit the reward function while requiring significant resources. We frame controllable generation as a problem of finding a distribution that optimizes a KL-regularized objective function. We present SLCD -- Supervised Learning based Controllable Diffusion, which iteratively generates online data and trains a small classifier to guide the generation of the diffusion model. Similar to the standard classifier-guided diffusion, SLCD's key computation primitive is classification and does not involve any complex concepts from RL or control. Via a reduction to no-regret online learning analysis, we show that under KL divergence, the output from SLCD provably converges to the optimal solution of the KL-regularized objective. Further, we empirically demonstrate that SLCD can generate high quality samples with nearly the same inference time as the base model in both image generation with continuous diffusion and biological sequence generation with discrete diffusion. Our code is available at https://github.com/Owen-Oertell/slcd