Studies in secure computation [electronic resource] : post-quantum, attribute-based and multi-party
- Valeria Nikolaenko.
- Physical description
- 1 online resource.
Also available at
- The invention of the first public-key cryptosystem four decades ago fired up the development of myriads of secure solutions for protecting data in a large variety of scenarios such as data retrieval, transmission and processing. This dissertation brings together several results on secure computations. We study post-quantum key agreement from generic lattices, attribute-based encryption, and privacy preserving multi-party computations with applications to machine learning. We provide the first instantiation and implementation of a key agreement protocol based on hard problems in random lattices. We show numerous optimizations of the protocol that make it efficient and practical. Next, we build a new lattice-based encryption scheme that lets anyone translate a ciphertext encrypted under a public key x into a ciphertext encrypted under the public key (f(x), f) of the same plaintext for any arithmetic circuit f. We showthat this scheme gives an Attribute-Based Encryption with short keys. Third, we develop systems for privacy preserving data-mining. We design and implement a protocol for computing linear regression model on user data without revealing any other information about the data.
- Publication date
- Submitted to the Department of Computer Science.
- Thesis (Ph.D.)--Stanford University, 2017.