How much explicit guidance is necessary for conditional diffusion? We consider the problem of conditional sampling using an unconditional diffusion model and limited explicit guidance (e.g., a noised classifier, or a conditional diffusion model) that is restricted to a small number of time steps. We explore a model predictive control (MPC)-like approach to approximate guidance by simulating unconditional diffusion forward, and backpropagating explicit guidance feedback. MPC-approximated guides have high cosine similarity to real guides, even over large simulation distances. Adding MPC steps improves generative quality when explicit guidance is limited to five time steps.
@article{arxiv.2210.12192,
title = {Conditional Diffusion with Less Explicit Guidance via Model Predictive Control},
author = {Max W. Shen and Ehsan Hajiramezanali and Gabriele Scalia and Alex Tseng and Nathaniel Diamant and Tommaso Biancalani and Andreas Loukas},
journal= {arXiv preprint arXiv:2210.12192},
year = {2022}
}