Applying Privacy-Enhancing Technologies
summary
Hands-on tests compare real Python packages and show what actually works: federated learning stacks perform (Flower is easiest to deploy, PySyft suits research) but need defenses against poisoned clients and flaky participants; differential privacy libraries vary (recommended: TensorFlow Privacy or Opacus) and reveal the expected accuracy–privacy tradeoff; homomorphic encryption and secure multi‑party computation are explored conceptually but not fully implemented here; combining methods (e.g. federated learning + homomorphic encryption) proves necessary for practical deployments.
AI-written summary · about
chapters (9)
- 0:00Privacy-Enhancing Technologies intro
- 3:38What are PETs
- 6:11Federated Learning explained
- 12:11Differential Privacy basics
- 18:11Homomorphic Encryption overview
- 21:01Secure Multi-Party Computation
- 25:36Combining PETs practically
- 28:29Benefits of Privacy-Enhancing Technologies
- 30:37Q&A
related talks
37mDemystifying Privacy Preserving Computing
16mPrivacy Preserving AI
29mPrivacy-preserving methods: Building secure projects
38mDemystifying Privacy Preserving Computing
39mData Privacy Engineering to power Responsible Enterprise AI
43mGuarding the ML Galaxy: Beyond Accuracy to Privacy and Security
43mPanel: Privacy Design Patterns for AI Systems: Threats and Protections
31mFrom Theory to Practice: Advances in Real-World Federated Learning