Adversarial machine learning session: hands-on
Welcome to this hands-on session on adversarial machine learning!
First, download the data required for the hands-on (42MB), given as a python pickle file, unzip it and place it in the root of the git project. This data was extracted from the CIFAR100 public dataset with super classes.
The main part of the hands-on is the notebook: 2025-07-09_FRomeu_hands-on-AML.ipynb.
Since it could be a little long on the first part, I also give a version where this first part (preprocessing data, setting up classification model) is done. You can choose where to start from, since setting up the machine learning model can be good experience for some of you.
Here are instructions to install the Python libraries you will need for the session.
apt update
apt install python3-pip
apt install python3.10-venv
python3 -m venv myenv
source myenv/bin/activate
git clone [...]
cd [...]
wget https://insatoulousefr-my.sharepoint.com/:u:/g/personal/leleux_insa-toulouse_fr/IQAitTNJGlpRSIJvzDqS9wbxAXqb3TN5MObWULhCFeRNUfQ?e=asRPvG
tar -zxvf [...]
pip install -r requirements_ENSEEIHT.txt
snap install code --classic
code &Installer les extensions Python et Jupyter sur Visual Studio.
Ouvrir le dossier du TP.
Sélectionner le Kernel Python "myenv"
Check if the first cell is executed (with the imports). If it is the case, you win !
Enjoy the Summer School !