Homomorphic Encryption: Secure Computing using Machine Learning

Abstract

This project develops a fully homomorphic encryption (FHE) scheme using the Brakerski-Fan-Vercauteren (BFV) algorithm to enable secure machine learning on encrypted data. FHE allows computations to be performed on data without decrypting it first, addressing the Coast Guard's challenge of leveraging AI and ML while navigating strict data use limitations. By implementing an FHE scheme integrated with a linear classifier, this project provides a solution to overcome data trust barriers, allowing the organization to utilize predictive technologies securely.