Decoupling for Robustness
Yaron discusses the necessity of separating model building from model security to achieve robustness. By implementing an AI firewall, the focus shifts from improving the model to safeguarding it against errors caused by bad data, highlighting the importance of decoupling for effective protection.In this clip
From this podcast
Practical AI
Eliminate AI failures
Related Questions
How can we defend against adversarial attacks on machine learning models?
What are adversarial attacks on machine learning models in the context of the episode Robustness, Detectability, and Data Privacy in AI // Vinu Sankar Sadasivan // #289 and the clip AI Defense Mechanisms?
What are adversarial attacks on machine learning models as discussed in the episode Dawn Song: Adversarial Machine Learning and Computer Security | Lex Fridman Podcast #95 and the clip Real World Vulnerabilities?