A Fourier Perspective on Model Robustness

Project assignment done for the following graduate course at the University of Toronto

CSC2541 - Topics in Machine Learning: Neural Network Training Dynamics by Roger Grosse

This course was taken as an elective for the degree of Master of Science, Department of Medical Biophysics, University of Toronto. The goal of the assignment was to create a colab notebook that illustrates some of the key ideas from your chosen paper, presenting at least one visualization that goes beyond what is presented in the paper. The following paper was chosen due to my interest and background in signal processing.

Yin, D., Gontijo Lopes, R., Shlens, J., Cubuk, E. D. & Gilmer, J. A Fourier Perspective on Model Robustness in Computer Vision. in Advances in Neural Information Processing Systems vol. 32 (Curran Associates, Inc., 2019).

Please note that part of the code might exceed the free resources provided by colab, so this version was run using kaggle. Some code outputs may not be rendered.