Counting 2,285 Big Data & Machine Learning Frameworks, Toolsets, and Examples...
Suggestion? Feedback? Tweet @stkim1

Simple Adversarial Examples

alt text

This repo contains an ipython notebook (along with corresponding helper python classes) that implements adversarial example generation for a neural network trained on MNIST, and also implements some (very) simple security measures against adversarial attacks. The majority of the base neural network impelmentation was shamelessly lifted from To run this notebook you will have to have jupyter notebook or ipython notebook installed. In addition, the code will need numpy and matplotlib as dependencies.

Once everything is installed go to the root directory of this repo in the terminal and run either

$ jupyter notebook


$ ipython notebook

A window should open up automatically in your browser. If not the output to the terminal should have a link you can put into a browser of your choice. It looks something like this


Finally, in the tree view of the ipython notebook click on adversarial_example.ipynb to start the notebook.

For more info about how to use Jupyter Notebooks check out the docs