Beginner ML project: adversarially robust MNIST classifier (part 1)

I work on a project to improve an MNIST classifier to make it robust against adversarial attacks. In this video, I show a baseline MNIST classifier that I'll be making adjustments to.

This video is the first part of my work, which I plan to continue to show in future videos.

I used these resources.

ML Safety Scholars curriculum: https://docs.google.com/document/d/1jKAeq6Sm9HTuA8N3545fZrhTxIw-gFYa4tBmNZPJAXE/edit?usp=sharing
Project guidelines: https://docs.google.com/document/d/1TXY8vjo8I4UO4FQ3Ls9c6VAqkkiqrQZzQCbXMYCNGss/edit?usp=sharing
MNIST classifier Google Colab notebook: https://colab.research.google.com/github/rpi-techfundamentals/fall2018-materials/blob/master/10-deep-learning/04-pytorch-mnist.ipynb
"TOWARDS THE FIRST ADVERSARIALLY ROBUST
NEURAL NETWORK MODEL ON MNIST": https://openreview.net/pdf?id=S1EHOsC9tX

Видео Beginner ML project: adversarially robust MNIST classifier (part 1) канала Naoya Okamoto