Secured and Private AI Scholarship Challenge from Facebook
Objective
Train smarter AI models by learning to safely and securely use distributed private data with differential privacy, federated learning, and encrypted computation techniques.
Lessons
- Welcome to Scholarship Challenge
Welcome note to the course and challenge.
- Deep Learning with PyTorch
Hands on tutorials and introdutory notes to PyTorch Library.
- Introducing Differential Privacy
Basics of Differential Privacy, a method for measuring how operations impact the privacy of data.
- Evaluating the Privacy of a Function
Implementing Differential Privacy in Python.
- Introducing Local and Global Differential Privacy
Applying Differential Privacy to arbitrary algorithms by adding noise to the outputs.
- Differential Privacy for Deep Learning
Differential Privacy to Deep Neural Networks
- Federated Learning
Methods for preserving data privacy by training models where the data lives.
- Secruring Federated Learning
Secure models trained with multi-party computation.
- Encrypted Deep Learning
Performing encryted computation. Building an encrypted database, and generate an encrypted prediction with an encryted neural networkon on an encryted database.