Model Evasion
Florian discusses the vulnerability of machine learning models to evasion attacks, where slight changes in input can drastically alter model predictions. He delves into the implications of this issue for security-critical applications and the ongoing research to develop more robust models.In this clip
From this podcast
Data Skeptic
Stealing Models from the Cloud
Related Questions
What are adversarial attacks on machine learning models as discussed in the episode #040 - Adversarial Examples (Dr. Nicholas Carlini, Dr. Wieland Brendel, Florian Tramèr) and the clip Machine Learning Defense Strategies?
What are adversarial attacks on machine learning models as discussed in the episode Fooling Computer Vision and the clip Generalized Attack Challenges?