Fully Homomorphic Encryption for Private Federated Learning

A scalable, Fully Homomorphic Encryption (FHE) pipeline that allows for model inference on encrypted data without the need for decryption.

Collaborative Machine Learning approach to train a mode that classifies a person as smoker or non-smoker based on the user data. The distributed approach of training is done with secure model transmissions to central cloud location where Amazon EC2 instance aggregates the new model based on new training received in Homomorphically Encrypted forms