Secure Machine Learning Ideas

Source: O'Reilly Media, Responsible Machine Learning, Patrick Hall, Navdeep Gill, and Benjamin Cox.

See secure_ml_ideas.pdf for a drafty write-up of these attacks.

Follow-ups to the original draft include:

• Proposals for model vulnerability and security on O'Reilly Ideas
• Can Your Machine Learning Model Be Hacked?! on the h2o.ai blog site.

See main.pdf for AI Risk and Security (AIRS) meetup slides.