PhD Student, UCLA
How to Read an Adversary’s Mind
How can we find out what an adversary knows? This has been an intriguing and important question at the core of most of cryptography.
Most cryptographic protocols are designed to enable collaboration between mutually distrustful participants. In such protocols, it is important to guarantee that an adversary participating in the protocol behaves consistently with some well-defined data that he knows. In fact, the proof of security needs to rely on extracting the data from every participant.
On the other hand, we also want to guarantee privacy of everyone’s data, meaning that real-world adversaries should be unable to obtain the data used by honest participants. Thus, we have two opposing objectives — on the one hand, we should be able to extract the adversary’s data, on the other, it should be impossible for the adversary to extract the honest parties’ data. How can we possibly achieve both of these goals, when we do not know who is honest and who is adversarial?
Prior techniques used to resolve this dilemma required the introduction of additional interaction. This meant that participants were required to go back-and-forth several times in their communication. However, such increased interaction introduced additional delays due to latency or unavailability of participants.
We develop new techniques that allow us to prove that the adversary is operating with knowledge of well-formed data, while reducing such back-and-forth to as little as a single message sent by every participant. We use these techniques to obtain several new minimally interactive protocols in the realm of secure computation, many of which were believed to be impossible so far.
This is based on multiple joint works with Abhishek Jain, Yael Kalai, Ron Rothblum and Amit Sahai
Dakshita Khurana is a PhD student at UCLA, interested in Cryptography and Theoretical Computer Science. She received a Masters degree from UCLA in 2014, and completed her Bachelors from the Indian Institute of Technology, Delhi in 2012.
The primary focus of her research is on ensuring data integrity, and computing on data while preserving privacy. The fundamental challenge at the heart of cryptographic protocol design is: how can we guarantee privacy in situations where an adversary can learn information about our secrets? Dakshita is particularly fascinated by this challenge in one of the core areas in cryptography: secure computation. She is interested in various aspects of the design of secure protocols. Her research so far has involved devising techniques to construct trim protocols with provable security against man-in-the-middle attacks, zero-knowledge proofs, as well as communication-efficient protocols.
Dakshita has previously pursued internships with Microsoft Research, at their India and New England offices. She is a recipient of the 2017-18 Dissertation Year Fellowship at UCLA, as well as the Cisco Outstanding Graduate Student Research Award. She was also recently chosen as a Heidelberg Laureate Forum Young Researcher.