Homework 4
ECE285
In this homework, we learn how to use GAN to generate medical images.
Take the chest X-ray dataset here
https://www.kaggle.com/paultimothymooney/chest-xray-pneumonia
Take the code of GAN here
https://pytorch.org/tutorials/beginner/dcgan_faces_tutorial.html
Then train a GAN model on the chest X-ray dataset, use the trained GAN
to generate chest X-ray images, and write a short report (1-2 pages, NeurIPS
format).
In the report, you can include the following: 1) hyperparameter settings; 2)
training loss curves; 3) Inception score and FID score of generated images, 4)
examples of generated images.
Submit the report to Gradescope. No need to submit code.
1
