PyCon US 2020 · April 2020

Practical Privacy-preserving Machine Learning in Python

A tour of the landscape of Python tools for privacy-preserving machine learning, from federated learning (where data stays on users' devices) to training on encrypted data. Reviews what works, what doesn't, and how tools like TensorFlow Privacy, TensorFlow Encrypted, and PySyft fit into real-world ML pipelines, covering the tradeoffs of each approach and touching on the ethics of using personal data.