Attacking Machine Learning - On the Security and Privacy of Neural Networks
Explore the security and privacy concerns surrounding machine learning in this 48-minute RSA Conference talk by Google Research Scientist Nicholas Carlini. Delve into two critical issues: the use of adversarial examples to deceive state-of-the-art vision classifiers, potentially impacting technologies like self-driving cars, and methods for extracting private training data from trained neural networks. Learn to recognize the potential impact of adversarial examples, understand how sensitive data can be leaked through exposed APIs, and identify when to deploy defenses against these emerging threats in the machine learning era. Gain insights into evasion attacks, privacy concerns, and the fundamentals of machine learning to better grasp the presented concepts.