In this project, we aim for implementing the DDPM (Denoising Diffusion Probabilisitc Model) to perfom unconditional generation and the DDIM (Denoising Diffusion Implicit Model) to perform conditional generation. The original data contains five US satets, each constructed with 1000 points.
Unconditional prediction:
Conditional prediction:
Classification of states for the conditional predictions:
This project was developed as part of CSE 849 (Deep Learning - Spring 25 Semester) at the Computer Science Department of Michigan State University, taught by Dr. Zijun Cui (@zijunjkl), with TA support from Gautam Sreekumar (@gautamsreekumar). Special thanks to them for their guidance and materials.