Instructor: Stanley Chan, Purdue University
Level: Intermediate
Prerequisites: None

Benefits:
This course enables the attendee to:

• Understand intuitively what diffusion models are.
• Describe the core mathematical principles of diffusion models.
• Understand the basic implementation of diffusion models.
• Appreciate how diffusion models are used to perform text-to-image content creation.

Course Description:
The astonishing growth of generative tools in recent years has empowered many exciting applications in text-to-image generation and text-to-video generation. The underlying principle behind these generative tools is the concept of diffusion, a particular sampling mechanism that has overcome some shortcomings that were deemed difficult in the previous approaches. This course discusses the essential ideas underlying the diffusion models. The target audience of this tutorial includes graduate students and researchers who are interested in doing research on diffusion models or applying these models to solve other problems.

Topics include:

• Variational auto-encoder (VAE)
• Denoising diffusion probabilistic models (DDPM)
• Applications in text-to-image generation, and text-to-video generation
• Open challenges and trends

Course material is based on https://arxiv.org/abs/2403.18103

Intended Audience: Engineers and researchers in the camera industry; graduate students

Stanley Chan is the Elmore Professor of electrical and computer engineering at Purdue University. He received his PhD in electrical engineering and MA in mathematics from UC San Diego, and his BEng in electrical engineering from the University of Hong Kong. In 2012-2014, he was a postdoctoral research fellow at Harvard University. He is a senior member of IEEE and is currently a deputy editor in chief of the IEEE Transactions on Computational Imaging, where he is recognized by the IEEE Signal Processing Society as an outstanding editorial board member. In IEEE, he is the vice chair of the technical committee on computational imaging.