Hands-On Adversarial AI

Designed to be used either as a stand-alone and self-contained course, or as a practical component of a more detailed university-level course on adversarial AI / machine learning and related topics. This project began from student lecture notes from Dr. Allison Bishop’s graduate Adversarial AI course at the City College of New York.

📚 Instructor Manual • 🤝 Contributing Guide • 👥 Acknowledgements

📋 Prerequisites

• Python 3.7+ experience
• Basic understanding of data structures and algorithms
• Calculus 1 and Introductory Probability

Note: no prior experience with machine learning or deep learning is required, we want to make these materials as self contained as possible.

🎓 Components

1. Part 1: Classical Adversarial Thinking - Hash function vulnerabilities and collision attacks
2. Part 2: Neural Network Fundamentals - Building CNNs with PyTorch
3. Part 3: Adversarial Examples and Attacks - FGSM, PGD, and fooling deep learning models

🚀 Getting Started

1. Clone the repository:
   git clone https://github.com/jadidbourbaki/adversarial-ai.git cd adversarial-ai
2. Choose your starting point:
   • New to adversarial AI? Start with Part 1
   • Have ML background? Jump to Part 3
   • Instructor? Check the Instructor Manual
3. Set up your environment:
   cd part1 # or part2, part3 pip install -r requirements.txt jupyter notebook

📝 Citation

If you use these materials in your research or teaching, please cite:

⚠️ Disclaimer

These materials are for educational purposes only. The goal is for students to understand and defend against security vulnerabilities in machine learning.