Biases, ML performance and adversarial ML threats

This lecture covers the basics of Machine Learning, its application under adversarial conditions, privacy implications, and challenges in deploying ML systems. It delves into traditional programming versus ML, supervised learning examples, adversarial ML threats, model stealing, defending against adversarial examples, and privacy issues in ML. The lecture also discusses the Base Rate Fallacy, distributional shift, and the impact of biases in ML models, emphasizing the difficulty of deploying ML systems in real-world scenarios.