This video is part of the Introduction to ML Safety course ([ Ссылка ]) and was recorded by Dan Hendrycks at the Center for AI Safety ([ Ссылка ]).
Check out our other projects:
Community: [ Ссылка ]
Newsletter: [ Ссылка ]
Twitter: [ Ссылка ]
Reddit: [ Ссылка ]
This video covers the following topics in adversarial robustness:
- Optimization pressure
- Projected gradient attack (PGD)
- Untargeted vs targeted attacks
- Adversarial evaluation
- White box vs black box attacks
- Transferability
- Unforeseen attacks
- Text attacks
- Robustness certificates
Ещё видео!