Wasserstein Convergence of ODE-Based Samplers in Decentralized Diffusion Model via Velocity Field Decomposition
Researchers have established a theoretical convergence guarantee for decentralized diffusion models using ODE-based sampling. This work provides the first Wasserstein-2 distance convergence result for such architectures, demonstrating that the distribution of the N-step discretization converges to the analytical solution at a rate of O(N^{-1/2} + \varepsilon). The findings are significant for understanding the privacy and scalability benefits of decentralized diffusion models. AI
IMPACT Establishes theoretical convergence for decentralized diffusion models, potentially enabling more private and scalable generative AI.