6 papers
Exponentially Fewer-Server PIR from Sparser -Decoding Polynomials
Aparna Gupte, Seyoon Ragavan
We show that under a plausible number-theoretic conjecture, for any constant there exists an -server private information retrieval (PIR) protocol that on an -bit database…
On Best-Possible One-Time Programs
Aparna Gupte, Jiahui Liu, Luowen Qian +3
One-time programs (OTPs) aim to let a user evaluate a program on a single input while revealing nothing else. Classical OTPs require hardware assumptions, and even with quantum inf…
Classical Obfuscation of Quantum Circuits via Publicly-Verifiable QFHE
James Bartusek, Aparna Gupte, Saachi Mutreja +1
A classical obfuscator for quantum circuits is a classical program that, given the classical description of a quantum circuit , outputs the classical description of a functional…
Quantum One-Time Programs, Revisited
Aparna Gupte, Jiahui Liu, Justin Raizes +2
One-time programs (Goldwasser, Kalai and Rothblum, CRYPTO 2008) are functions that can be run on any single input of a user's choice, but not on a second input. Classically, they a…
Sparse Linear Regression and Lattice Problems
Aparna Gupte, Neekon Vafa, Vinod Vaikuntanathan
Sparse linear regression (SLR) is a well-studied problem in statistics where one is given a design matrix and a response vector for a -s…
SGD and Weight Decay Secretly Minimize the Rank of Your Neural Network
Tomer Galanti, Zachary S. Siegel, Aparna Gupte +1
We investigate the inherent bias of Stochastic Gradient Descent (SGD) toward learning low-rank weight matrices during the training of deep neural networks. Our results demonstrate…