Adversarial Two-Party Quantum Interactions in Cryptography and Machine Learning

dc.contributor.authorBansal, Akshayen
dc.contributor.committeechairSikora, Jamieen
dc.contributor.committeememberMittal, Rajaten
dc.contributor.committeememberMatthews, Gretchen L.en
dc.contributor.committeememberJi, Boen
dc.contributor.committeememberMantri, Atulen
dc.contributor.departmentComputer Science and#38; Applicationsen
dc.date.accessioned2025-06-11T08:03:21Zen
dc.date.available2025-06-11T08:03:21Zen
dc.date.issued2025-06-10en
dc.description.abstractTwo-party adversarial interactions are of fundamental importance in the study of cryptography and machine learning, as they allow us to work in dynamic environments to model and adapt to various challenges, fostering reliability and robustness. In this thesis, we first consider such interactions in the domain of information-theoretic cryptography, where we propose new protocols and security bounds involving various two-party primitives such as coin flipping, bit commitment, and variants of oblivious transfer. We begin by developing the mathematical framework of stochastic selection, which is potentially useful for devising new protocols with improved security for a combination of various tasks. We then use this idea to develop the first protocols for Rabin oblivious transfer that offer partial information-theoretic security and also provide a non-trivial lower bound on its security. We also relax some of the more commonly used assumptions to consider the security of protocols in broader compositions and independently, allowing for devices that may be potentially malicious. In the former setting, we demonstrate that the weak coin flipping task -- which enjoys almost perfect standalone security -- may not be immune to attacks when used in larger compositions. In the latter, we establish the impossibility of determining the inner workings of a shared quantum mechanical device -- a phenomenon commonly known as self-testing -- in the presence of an adversary. Finally, we consider a slightly different task in a similar setting, where we discuss the classical online learnability of various quantum objects. We prove a sublinear regret bound for learning over general subsets of positive semidefinite matrices via the regularized follow-the-leader algorithm and apply it to various settings involving the learning of quantum objects. For concrete applications, we present a sublinear regret bound for learning quantum states, effects, channels, interactive measurements, strategies, co-strategies, and collections of inner products of pure states. In proving our regret bound, we establish various matrix analysis results useful in quantum information theory, including a generalization of Pinsker's inequality for arbitrary positive semidefinite operators with possibly different traces, which may be of independent interest in the study of more general classes of divergences.en
dc.description.abstractgeneralTwo-party interactions are the fundamental building blocks in many cryptographic systems and learning environments alike. In the cryptographic setting, an interaction is said to be adversarial if a malicious party attempts to achieve its desired objective by deviating from the prescribed protocol. On the other hand, a learning environment is deemed adversarial if it aims to restrict the learning prowess of the learner. The focus of this study is on adversarial two-party interactions in both the related domains of cryptography and machine learning, which is crucial for making systems robust, reliable, and adaptive to rapidly changing environments. We first consider such interactions from the lens of unconditional security, or information-theoretic security, where the adversary is assumed to be unbounded by any computational resources. We begin by developing the mathematical framework of stochastic selection, where one of the two parties randomly selects a protocol (from a finite set of known protocols) after some initial communication with the other party. We demonstrate that this framework is potentially useful for devising new protocols with improved security for a combination of various cryptographic tasks such as coin flipping (generating a common binary outcome), bit commitment (revealing the committed bit after a certain hiding phase), and variants of oblivious transfer (sending information without knowing what was sent). We later use this idea of stochastic selection to develop the first protocols for Rabin oblivious transfer that offer partial information-theoretic security and also provide a non-trivial lower bound on its security. We also relax some of the more commonly used assumptions to consider the security of protocols when used as subroutines in a larger system comprising multiple protocols, and independently, allowing for devices that may be potentially malicious. In the former setting, we demonstrate that the task of weak coin flipping -- which enjoys almost perfect information-theoretic security when implemented in isolation -- may not be immune to attacks when used in a larger system or composition. In the latter, we establish the impossibility of determining whether a shared quantum mechanical device satisfies a certain device specification (shared quantum state and measurements) -- a phenomenon commonly known as self-testing -- in the presence of an adversary. In the final part of the thesis, we discuss the online learnability of various useful quantum objects, where estimates of the classical representation of the unknown object are updated adaptively -- generally based on the continuously evolving choice of loss functions by the adversary. We present new results showing that such learning is possible, even in complex scenarios, and we prove several new mathematical results along the way that could be of independent interest in quantum information theory.en
dc.description.degreeDoctor of Philosophyen
dc.format.mediumETDen
dc.identifier.othervt_gsexam:44332en
dc.identifier.urihttps://hdl.handle.net/10919/135470en
dc.language.isoenen
dc.publisherVirginia Techen
dc.rightsIn Copyrighten
dc.rights.urihttp://rightsstatements.org/vocab/InC/1.0/en
dc.subjecttwo-party cryptographyen
dc.subjectself-testingen
dc.subjectonline learningen
dc.titleAdversarial Two-Party Quantum Interactions in Cryptography and Machine Learningen
dc.typeDissertationen
thesis.degree.disciplineComputer Science & Applicationsen
thesis.degree.grantorVirginia Polytechnic Institute and State Universityen
thesis.degree.leveldoctoralen
thesis.degree.nameDoctor of Philosophyen

Files

Original bundle
Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
Bansal_A_D_2025.pdf
Size:
1.3 MB
Format:
Adobe Portable Document Format