Speaker:

Speaker:

Speaker:

Opacus is a library that enables training PyTorch models with differential privacy. It supports training with minimal code changes required on the client, has little impact on training performance and allows the client to online track the privacy budget expended at any given moment. In this talk, Igor and Alex go through the importance of Private ML and showcase how to use Opacus to experiment with Differential Privacy in your PyTorch model training.

Privacy attacks are an important threat vector for machine learning models. This talk explains the main techniques used in Privacy Attacks and showcases the Privacy Linter, a diagnostic tool in PyTorch that stimulates attacks to estimate the associated risk.

Opacus is a library that enables training PyTorch models with differential privacy. In this talk, Igor and Alex will go through the importance of Private ML and showcase how to use Opacus to experiment with Differential Privacy in your PyTorch model training and explain the main techniques used in Privacy Attacks and showcase the Privacy Linter, a diagnostic tool in PyTorch that stimulates attacks to estimate the associated risk.