Computing on Encrypted Data via Secret Dual Codes

Explore advanced cryptographic techniques for computing on encrypted data in this research talk by Yuval Ishai from Technion and AWS. Delve into the secret-key setting where a client uploads encrypted input to an untrusted server and makes unlimited queries while maintaining privacy of both the queries and data using only a secret key. Examine efficient solutions for matrix-vector products and private information retrieval (PIR) under the Learning Parity with Noise assumption and novel assumptions related to learning secret linear subspaces from noisy samples. Discover the core innovation of encoding inputs and queries using secret dual codes while preventing linear algebra attacks through strategic noise addition, inspired by PIR preprocessing techniques. Learn about cutting-edge research that achieves efficiency features not available in existing solutions, presented as part of collaborative work with leading cryptography researchers including Fabrice Benhamouda, Caicai Chen, Shai Halevi, Hugo Krawczyk, Tamer Mour, Tal Rabin, and Alon Rosen.