Mariana Raykova, Ph.D., Columbia University; B.A., Bard College

Mariana Raykova's picture
Assistant Professor
AKW 404, 51 Prospect St, New Haven, CT 06511-8937

Dr. Raykova’s research work is in the area of cryptography and applications to security. Her interests included secure computation which enables multiple parties to compute on their private data jointly without revealing their individual sensitive inputs to each other. Her work develops new efficient constructions for secure computation and implementations that aim to bring these techniques closer to practical use. Secure computation has numerous applications in almost any scenario that involves private information such as computation of medical data, running auctions, providing anonymous credentials, processing of financial records and many others. A special case of secure computation with wide applications on its own, which Dr. Raykova has worked on is secure search that allows mutually distrusting parties to share parts of their private data in a controlled manner.

Another area of Dr. Raykova’s research is cryptographic obfuscation and its applications. The goal of a code obfuscation is to hide secrets in software by “obfuscating” the internal mechanisms of a program while preserving access to its functionality which enables the evaluation of the program on any input. Such capabilities would allow to implement proprietary algorithms, e.g., DNA analysis algorithms, in software that enables the use of the algorithm while protecting its inner workings. Hiding secrets into software may also be crucial for the effectiveness of some particular applications such as malware analysis tools that may need to include malware signatures. Dr. Raykova is an author of a breakthrough result that presented the first candidate construction for obfuscation of any program. Her work was featured in Wired, Communications of ACM, IEEE Spectrum and others. Obfuscation is also used for the construction of many new and very useful cryptographic functionalities such as a functional encryption, which is a special type of encryption that reveals only partial information about the encrypted messages, and protocols for efficient computation outsourcing, which allows a computationally powerful party such as the cloud to do work on behalf of a weak client without learning its private inputs. The work of Dr. Raykova has developed constructions for these cryptographic tools.

Verifiable computation considers the setting where a party with limited storage and computation resources outsources its data to a powerful server such as a cloud service and later asks the server to run computation jobs on the outsourced data and report the results. In this setting the client wants to verify very efficiently the correctness of the results that the sever returns while doing much less work than the computation itself. Protocols for verifiable computation aim to achieve this goal. Dr. Raykova has developed one of the most efficient cryptographic approaches for verifiable computation which is used is almost all most recent implemented solutions for this task. Her work “Pinocchio: Nearly Practical Verifiable Computation” received the Best Paper Award at the IEEE Security and Privacy Conference in 2013 and was featured in the MIT Technology Review.

Dr. Raykova is excited to work on new research problems in cryptography and directions that search for applications of cryptography tools to solve practical security questions.

Representative Publications:

  • Secure Two-Party Computation with Sublinear Amortized Work, Dov Gordon, Jonathan Katz, Vladimir Kolesnikov, Tal Malkin, Mariana Raykova, Yevgeniy Vahlis, CCS 2012.
  • Candidate Indistinguishability Obfuscation and Functional Encryption for All Circuits, Sanjam Garg, Craig Gentry, Shai Halevi, Mariana Raykova, Amit Sahai, Brent Waters, FOCS 2013.
  • Pinocchio: Nearly Practical Verifiable Computation, Bryan Parno, Craig Gentry, Jon Howell, Mariana Raykova, IEEE Symposium on Security and Privacy, 2013Best Paper Award.
  • Quadratic Span Programs and Succinct NIZKs without PCPs, Rosario Gennaro, Craig Gentry, Bryan Parno, Mariana Raykova, Eurocrypt 2013.
  • Outsourcing Private RAM Computation, Craig Gentry, Shai Halevi, Mariana Raykova, Daniel Wichs, FOCS 2014.
  • Garbled RAM Revisited, Sanjam Garg, Craig Gentry, Shai Halevi, Steve Lu, Rafail Ostrovsky, Mariana Raykova, Daniel Wichs, Eurocrypt 2014.
  • Semantically Secure Order-Revealing Encryption: Multi-Input Functional Encryption Without Obfuscation, Dan Boneh, Kevin Lewi, Mariana Raykova, Amit Sahai, Mark Zhandry, Joe Zimmerman, Eurocrypt 2015.