Speaker:Professor Chieh-Hsin Lai (Sony AI's Music Foundation Model Team)
Topic:Diffusion Model and Flow
Speaker:Professor Chieh-Hsin Lai (Sony AI's Music Foundation Model Team)
Date Time:April 18 (Friday) , 2025, 11:10-12:00
Place: 4F-427, Assembly Building I
Online Seminars- Google Meet
Abstract
Diffusion models and flow matching techniques have emerged as powerful frameworks in generative modeling, driving significant advances in the synthesis of complex data structures. Despite their growing prominence, a unified and comprehensive understanding of their underlying principles, mathematical formulations, and practical applications remains scattered across disparate sources. This talk aims to bridge this gap by starting with the historical development of diffusion models, providing an in-depth exploration of diffusion models, flow matching, and their interplay, and offering both theoretical insights and practical methodologies.