A Plug-in Interpretation of Conditioning in Score-Based Diffusion Models
Decoupling diffusion dynamics for more transparent and controllable generative AI
We propose a conditioning mechanism for diffusion models based on multi-speed joint diffusion of the target and the condition. The mechanism learns an unconditional joint score network and enforces conditioning at inference via a plug-in correction term. The plug-in term separate...
ArXiv CS.CV
Source
Generated insight by OnAirToday AI
This research introduces a novel plug-in mechanism that decouples conditional information from unconditional generative dynamics, providing a clearer mathematical framework for steering AI outputs. By mitigating the discrepancy between ODE and SDE samplers, it offers a more efficient and stable approach to high-fidelity image generation.

Comments
Keep comments relevant and respectful. No links, advertising, self-promotion, or spam.