Practical Adversarial AI: Comprehensive Resources for Mastering Adversarial Machine Learning and Security Vulnerabilities

This project provides a flexible, stand-alone educational course or a practical component for university-level studies in adversarial AI and machine learning. Originating from Dr. Allison Bishop’s lectures at the City College of New York, it requires familiarity with Python 3.7+, basic data structures and algorithms, and introductory calculus and probability. The materials are designed to be self-contained, without the need for prior machine learning experience. The course covers three key parts: classical adversarial thinking, neural network fundamentals focusing on building CNNs with PyTorch, and an exploration of adversarial examples and attacks like FGSM and PGD. To get started, users can clone the repository, set up their environment, and access the content via Jupyter Notebook. The materials are intended solely for educational purposes, helping students understand and mitigate the security vulnerabilities in machine learning. Proper citation is requested for any academic use.

Source link