A
Anand D. Sarwate
Researcher at Rutgers University
Publications - 179
Citations - 5681
Anand D. Sarwate is an academic researcher from Rutgers University. The author has contributed to research in topics: Differential privacy & Computer science. The author has an hindex of 29, co-authored 166 publications receiving 4718 citations. Previous affiliations of Anand D. Sarwate include Toyota & University of California, Berkeley.
Papers
More filters
Journal ArticleDOI
Differentially Private Empirical Risk Minimization
TL;DR: This work proposes a new method, objective perturbation, for privacy-preserving machine learning algorithm design, and shows that both theoretically and empirically, this method is superior to the previous state-of-the-art, output perturbations, in managing the inherent tradeoff between privacy and learning performance.
Proceedings ArticleDOI
Stochastic gradient descent with differentially private updates
TL;DR: This paper derives differentially private versions of stochastic gradient descent, and test them empirically to show that standard SGD experiences high variability due to differential privacy, but a moderate increase in the batch size can improve performance significantly.
Journal ArticleDOI
Broadcast Gossip Algorithms for Consensus
TL;DR: It is proved that the random consensus value is, in expectation, the average of initial node measurements and that it can be made arbitrarily close to this value in mean squared error sense, under a balanced connectivity model and by trading off convergence speed with accuracy of the computation.
Posted Content
Differentially Private Empirical Risk Minimization
TL;DR: In this article, the authors proposed a new method, objective perturbation, for privacy-preserving machine learning algorithm design, which perturbs the objective function before optimizing over classifiers.
Journal ArticleDOI
Geographic Gossip: Efficient Averaging for Sensor Networks
TL;DR: This work proposes and analyzes an alternative gossiping scheme that exploits geographic information and demonstrates substantial gains over previously proposed gossip protocols by utilizing geographic routing combined with a simple resampling method.