Diffusion Explainer
Diffusion explainers are generative models that leverage the principles of diffusion processes to create new data samples, primarily images and other high-dimensional data, by reversing a noise-addition process. Current research focuses on improving efficiency (e.g., one-step diffusion), enhancing controllability (e.g., through classifier-free guidance and conditioning on various modalities like text and 3D priors), and addressing challenges like data replication and mode collapse. These advancements are impacting diverse fields, from image super-resolution and medical imaging to robotics, recommendation systems, and even scientific simulations, by providing powerful tools for data generation, manipulation, and analysis.
Papers
Synthesizing realistic sand assemblies with denoising diffusion in latent space
Nikolaos N. Vlassis, WaiChing Sun, Khalid A. Alshibli, Richard A. Regueiro
Professional Basketball Player Behavior Synthesis via Planning with Diffusion
Xiusi Chen, Wei-Yao Wang, Ziniu Hu, Curtis Chou, Lam Hoang, Kun Jin, Mingyan Liu, P. Jeffrey Brantingham, Wei Wang
ZET-Speech: Zero-shot adaptive Emotion-controllable Text-to-Speech Synthesis with Diffusion and Style-based Models
Minki Kang, Wooseok Han, Sung Ju Hwang, Eunho Yang
LLM-grounded Diffusion: Enhancing Prompt Understanding of Text-to-Image Diffusion Models with Large Language Models
Long Lian, Boyi Li, Adam Yala, Trevor Darrell