Inference-Time Alignment for Diffusion Models via Doob's Matching
By: Jinyuan Chang , Chenguang Duan , Yuling Jiao and more
Potential Business Impact:
Makes AI art follow instructions better.
Inference-time alignment for diffusion models aims to adapt a pre-trained diffusion model toward a target distribution without retraining the base score network, thereby preserving the generative capacity of the base model while enforcing desired properties at the inference time. A central mechanism for achieving such alignment is guidance, which modifies the sampling dynamics through an additional drift term. In this work, we introduce Doob's matching, a novel framework for guidance estimation grounded in Doob's $h$-transform. Our approach formulates guidance as the gradient of logarithm of an underlying Doob's $h$-function and employs gradient-penalized regression to simultaneously estimate both the $h$-function and its gradient, resulting in a consistent estimator of the guidance. Theoretically, we establish non-asymptotic convergence rates for the estimated guidance. Moreover, we analyze the resulting controllable diffusion processes and prove non-asymptotic convergence guarantees for the generated distributions in the 2-Wasserstein distance.
Similar Papers
Direct Distributional Optimization for Provable Alignment of Diffusion Models
Machine Learning (CS)
Makes AI create better pictures from your words.
Likelihood Matching for Diffusion Models
Machine Learning (Stat)
Teaches computers to draw pictures from descriptions.
Implicit score matching meets denoising score matching: improved rates of convergence and log-density Hessian estimation
Statistics Theory
Helps computers learn to create realistic images.