Completed
Hardware Attacks Can Break Mathematically-Proven Guarantees
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
A Sound Mind in a Vulnerable Body - Practical Hardware Attacks on Deep Learning
Automatically move to the next video in the Classroom when playback concludes
- 1 Intro
- 2 Recent Work on Secure Machine Learning
- 3 Conventional View on ML Models' Robustness
- 4 We Propose A New Perspective!
- 5 Hardware Attacks Can Break Mathematically-Proven Guarantees
- 6 (Weak) Hardware Attacks Can Be Exploited in the Cloud
- 7 Prior Work's Perspective on a Model's Robustness
- 8 The Worst-Case Perturbation
- 9 Threat Model - Single-Bit Adversaries
- 10 Evaluate the Weakest Attacker with Multiple Bit-flips
- 11 Our Attack: Reconstruction of DNN Architectures from the Trace
- 12 We Can Identify the Layers Accessed While Computing
- 13 Solution: Generate All Candidate Architectures
- 14 Solution: Eliminate incompatible Candidates
