Abstract
Encryption is very effective at protecting the secrecy of data at rest (storage) and in transit (networks). Could it also protect it during computations on these data ? The lecture introduced the notion of homomorphic encryption, enabling computation on encrypted data without the decryption key, and outlined Gentry's (2009) approach, the first realization of this concept. Homomorphic encryption, along with other cryptographic protocols, provides solutions to the problem of secure multiparty computation, where participants together compute a function of their private data without revealing anything more than the final result. Although still very costly in terms of execution time, these cryptographic approaches provide a mathematically sound answer to the problem of computing on confidential data without information leaks.
