Brownian Bridge
Brownian Bridge (BB) processes are increasingly used in generative modeling, particularly within diffusion models, to improve image and signal processing tasks. Current research focuses on leveraging BB's properties for efficient and high-quality image-to-image translation, speech enhancement, and other applications by incorporating it into various architectures, including conditional diffusion models and generative adversarial networks. This approach offers advantages in terms of speed, stability, and the generation of more deterministic and consistent outputs compared to traditional diffusion methods, impacting fields ranging from medical imaging to natural language processing.
Papers
November 4, 2024
October 13, 2024
August 15, 2024
July 7, 2024
July 6, 2024
June 10, 2024
May 9, 2024
January 18, 2024
December 28, 2023
October 11, 2023
August 30, 2023
May 28, 2023
May 23, 2023
May 9, 2023