Exploring Compositional Visual Generation with Latent Classifier Guidance

Understand

Diffusion probabilistic models have achieved enormous success in the field of image generation and manipulation.

• In this paper, we explore a novel paradigm of using the diffusion model and classifier guidance in the latent semantic space for compositional visual tasks.
• Specifically, we train latent diffusion models and auxiliary latent classifiers to facilitate non-linear navigation of latent representation generation for any pre-trained generative model with a semantic latent space.
• We demonstrate that such conditional generation achieved by latent classifier guidance provably maximizes a lower bound of the conditional log probability during training.