Under the umbrella of “Advanced Topics in CS”, we bring this new course in which we cover diffusion based algorithms for data synthesis. This is a hot topic in machine learning and image processing, offering iterative methods that start form random noise and end with a high quality synthesis of visual (or other!) content. This course will be given in a seminar format, in which each participant studies a specific topic/paper and lectures about it to the class.
This is a graduate version of the course Numerical Algorithms (234125), covering the same material exactly, but ending with a large project in which a recent paper on the topics of this course is used for a final project. This project will require (i) reading the paper, implementing it, and extending its ideas; (ii) creation of slides to describe the paper’s content and results; (iii) lectruring on this paper to the course participants; and (iv) writing a final report to summarize this project.