Private AI: Machine Learning on Encrypted Data
Title: Private AI: Machine Learning on Encrypted Data
Abstract: As the world rushes towards the adoption of Artificial
Intelligence (AI) in all aspects of our lives, the promise is great, but
the risks are many. AI can help to improve and simplify our daily lives
and introduce new types of safety measures, but at the cost of sharing
and potentially leaking or misusing our private data. Cloud services
built on AI algorithms need to make use of customer or enterprise data,
to train models and make predictions. It is up to us to protect that
data, in storage and in use. Private AI is a set of tools being
developed by the Cryptography Group at Microsoft Research which can help
protect data while in use by encrypting it. The leading edge of Private
AI is based on Homomorphic Encryption (HE), a new encryption paradigm
which allows the cloud to operate on private data in encrypted form,
without ever decrypting it, enabling private training and private
prediction. This talk will give a snapshot of the state of the art and
show some compelling demos of HE in action.