Training Latent Diffusion Models with Interacting Particle Algorithms
By: Tim Y. J. Wang, Juan Kuntz, O. Deniz Akyildiz
Potential Business Impact:
Makes AI create pictures from just words.
We introduce a novel particle-based algorithm for end-to-end training of latent diffusion models. We reformulate the training task as minimizing a free energy functional and obtain a gradient flow that does so. By approximating the latter with a system of interacting particles, we obtain the algorithm, which we underpin theoretically by providing error guarantees. The novel algorithm compares favorably in experiments with previous particle-based methods and variational inference analogues.
Similar Papers
Learning Latent Energy-Based Models via Interacting Particle Langevin Dynamics
Machine Learning (Stat)
Teaches computers to learn from data better.
Variational Formulation of the Particle Flow Particle Filter
Machine Learning (Stat)
Makes computer guesses better by learning from mistakes.
Evolvable Conditional Diffusion
Machine Learning (CS)
Helps computers discover new science without math.