On Fri, 2019-06-07 at 16:42 +0530, Arjun Bahuguna wrote:
*Topic: *PySyft - an open-source framework for privacy-preserving
machine
learning
*Description: *Machine learning is increasingly being used in various
products and services. What're not being discussed are the serious
privacy
concerns that come with machine learning analysis. In this talk, we
discuss
the ways in which private data can be leaked out of models using
malicious
attacks and show how to protect against these attacks using PySyft
<https://github.com/OpenMined/PySyft>;. We also discuss the work
OpenMined,
an open-source organization focused on private machine learning, is
doing
in the space. We end our talk with a demo
<https://github.com/rohanbanerjee/Morphic> we created for doing data
analysis on private healthcare datasets without compromising user
privacy.
Previous knowledge of ML is not required, we'll cover required basics
in
the introduction.
*Duration: *30mins
*About Yourself:* Sourav Sharan & Arjun Bahuguna are undergraduate
researchers at SRM's Next Tech Lab, and work with statistical
learning and
applied cryptography.