Noise robustness of persistent homology on greyscale images, across filtrations and signatures
Supplementary material

Author: Renata Turkes



The folder turkevs2021noise contains the code used for computational experiments on the MNIST dataset of handwritten digits in the aforementioned paper: turkevs2021noise.ipynb.

The code allows to replicate our study, or to easily investigate the noise robustness of persistent homology with different filtrations and persistence signatures, and their parameters and corresponding metric, for other datasets of greyscale images, types of noise and/or classifiers.