Search Results for author: Abhradeep Thakurta

Found 37 papers, 9 papers with code

Training Private Models That Know What They Don't Know

no code implementations28 May 2023 Stephan Rabanser, Anvith Thudi, Abhradeep Thakurta, Krishnamurthy Dvijotham, Nicolas Papernot

Training reliable deep learning models which avoid making overconfident but incorrect predictions is a longstanding challenge.

Faster Differentially Private Convex Optimization via Second-Order Methods

no code implementations22 May 2023 Arun Ganesh, Mahdi Haghifam, Thomas Steinke, Abhradeep Thakurta

We first develop a private variant of the regularized cubic Newton method of Nesterov and Polyak, and show that for the class of strongly convex loss functions, our algorithm has quadratic convergence and achieves the optimal excess loss.

Second-order methods

How to DP-fy ML: A Practical Guide to Machine Learning with Differential Privacy

1 code implementation1 Mar 2023 Natalia Ponomareva, Hussein Hazimeh, Alex Kurakin, Zheng Xu, Carson Denison, H. Brendan McMahan, Sergei Vassilvitskii, Steve Chien, Abhradeep Thakurta

However, while some adoption of DP has happened in industry, attempts to apply DP to real world complex ML models are still few and far between.

Private (Stochastic) Non-Convex Optimization Revisited: Second-Order Stationary Points and Excess Risks

no code implementations20 Feb 2023 Arun Ganesh, Daogao Liu, Sewoong Oh, Abhradeep Thakurta

The first kind of oracles can estimate the gradient of one point, and the second kind of oracles, less precise and more cost-effective, can estimate the gradient difference between two points.

Why Is Public Pretraining Necessary for Private Model Training?

no code implementations19 Feb 2023 Arun Ganesh, Mahdi Haghifam, Milad Nasr, Sewoong Oh, Thomas Steinke, Om Thakkar, Abhradeep Thakurta, Lun Wang

To explain this phenomenon, we hypothesize that the non-convex loss landscape of a model training necessitates an optimization algorithm to go through two phases.

Transfer Learning

Multi-Task Differential Privacy Under Distribution Skew

no code implementations15 Feb 2023 Walid Krichene, Prateek Jain, Shuang Song, Mukund Sundararajan, Abhradeep Thakurta, Li Zhang

We study the problem of multi-task learning under user-level differential privacy, in which $n$ users contribute data to $m$ tasks, each involving a subset of users.

Multi-Task Learning

Differentially Private Image Classification from Features

no code implementations24 Nov 2022 Harsh Mehta, Walid Krichene, Abhradeep Thakurta, Alexey Kurakin, Ashok Cutkosky

We find that linear regression is much more effective than logistic regression from both privacy and computational aspects, especially at stricter epsilon values ($\epsilon < 1$).

Classification Image Classification +3

Multi-Epoch Matrix Factorization Mechanisms for Private Machine Learning

1 code implementation12 Nov 2022 Christopher A. Choquette-Choo, H. Brendan McMahan, Keith Rush, Abhradeep Thakurta

We formalize the problem of DP mechanisms for adaptive streams with multiple participations and introduce a non-trivial extension of online matrix factorization DP mechanisms to our setting.

Image Classification Language Modelling

Fully Adaptive Composition for Gaussian Differential Privacy

no code implementations31 Oct 2022 Adam Smith, Abhradeep Thakurta

We show that Gaussian Differential Privacy, a variant of differential privacy tailored to the analysis of Gaussian noise addition, composes gracefully even in the presence of a fully adaptive analyst.

Fine-Tuning with Differential Privacy Necessitates an Additional Hyperparameter Search

no code implementations5 Oct 2022 Yannis Cattan, Christopher A. Choquette-Choo, Nicolas Papernot, Abhradeep Thakurta

For instance, we achieve 77. 9% accuracy for $(\varepsilon, \delta)=(2, 10^{-5})$ on CIFAR-100 for a model pretrained on ImageNet.

Privacy Preserving

Recycling Scraps: Improving Private Learning by Leveraging Intermediate Checkpoints

no code implementations4 Oct 2022 Virat Shejwalkar, Arun Ganesh, Rajiv Mathews, Om Thakkar, Abhradeep Thakurta

Empirically, we show that the last few checkpoints can provide a reasonable lower bound for the variance of a converged DP model.

(Nearly) Optimal Private Linear Regression via Adaptive Clipping

no code implementations11 Jul 2022 Prateek Varshney, Abhradeep Thakurta, Prateek Jain

Compared to existing $(\epsilon, \delta)$-DP techniques which have sub-optimal error bounds, DP-AMBSSGD is able to provide nearly optimal error bounds in terms of key parameters like dimensionality $d$, number of points $N$, and the standard deviation $\sigma$ of the noise in observations.


Large Scale Transfer Learning for Differentially Private Image Classification

no code implementations6 May 2022 Harsh Mehta, Abhradeep Thakurta, Alexey Kurakin, Ashok Cutkosky

Moreover, by systematically comparing private and non-private models across a range of large batch sizes, we find that similar to non-private setting, choice of optimizer can further improve performance substantially with DP.

Classification Image Classification +1

Differentially Private Sampling from Rashomon Sets, and the Universality of Langevin Diffusion for Convex Optimization

no code implementations4 Apr 2022 Arun Ganesh, Abhradeep Thakurta, Jalaj Upadhyay

In this paper we provide an algorithmic framework based on Langevin diffusion (LD) and its corresponding discretizations that allow us to simultaneously obtain: i) An algorithm for sampling from the exponential mechanism, whose privacy analysis does not depend on convexity and which can be stopped at anytime without compromising privacy, and ii) tight uniform stability guarantees for the exponential mechanism.


Toward Training at ImageNet Scale with Differential Privacy

1 code implementation28 Jan 2022 Alexey Kurakin, Shuang Song, Steve Chien, Roxana Geambasu, Andreas Terzis, Abhradeep Thakurta

Despite a rich literature on how to train ML models with differential privacy, it remains extremely challenging to train real-life, large neural networks with both reasonable accuracy and privacy.

Image Classification with Differential Privacy

Public Data-Assisted Mirror Descent for Private Model Training

no code implementations1 Dec 2021 Ehsan Amid, Arun Ganesh, Rajiv Mathews, Swaroop Ramaswamy, Shuang Song, Thomas Steinke, Vinith M. Suriyakumar, Om Thakkar, Abhradeep Thakurta

In this paper, we revisit the problem of using in-distribution public data to improve the privacy/utility trade-offs for differentially private (DP) model training.

Federated Learning

Adversary Instantiation: Lower Bounds for Differentially Private Machine Learning

no code implementations11 Jan 2021 Milad Nasr, Shuang Song, Abhradeep Thakurta, Nicolas Papernot, Nicholas Carlini

DP formalizes this data leakage through a cryptographic game, where an adversary must predict if a model was trained on a dataset D, or a dataset D' that differs in just one example. If observing the training algorithm does not meaningfully increase the adversary's odds of successfully guessing which dataset the model was trained on, then the algorithm is said to be differentially private.

BIG-bench Machine Learning

The Flajolet-Martin Sketch Itself Preserves Differential Privacy: Private Counting with Minimal Space

no code implementations NeurIPS 2020 Adam Smith, Shuang Song, Abhradeep Thakurta

We propose an $(\epsilon,\delta)$-differentially private algorithm that approximates $\dist$ within a factor of $(1\pm\gamma)$, and with additive error of $O(\sqrt{\ln(1/\delta)}/\epsilon)$, using space $O(\ln(\ln(u)/\gamma)/\gamma^2)$.

Fast Dimension Independent Private AdaGrad on Publicly Estimated Subspaces

no code implementations14 Aug 2020 Peter Kairouz, Mónica Ribero, Keith Rush, Abhradeep Thakurta

In particular, we show that if the gradients lie in a known constant rank subspace, and assuming algorithmic access to an envelope which bounds decaying sensitivity, one can achieve faster convergence to an excess empirical risk of $\tilde O(1/\epsilon n)$, where $\epsilon$ is the privacy budget and $n$ the number of samples.

Tempered Sigmoid Activations for Deep Learning with Differential Privacy

1 code implementation28 Jul 2020 Nicolas Papernot, Abhradeep Thakurta, Shuang Song, Steve Chien, Úlfar Erlingsson

Because learning sometimes involves sensitive data, machine learning algorithms have been extended to offer privacy for training data.

Privacy Preserving Privacy Preserving Deep Learning

Evading Curse of Dimensionality in Unconstrained Private GLMs via Private Gradient Descent

no code implementations11 Jun 2020 Shuang Song, Thomas Steinke, Om Thakkar, Abhradeep Thakurta

We show that for unconstrained convex generalized linear models (GLMs), one can obtain an excess empirical risk of $\tilde O\left(\sqrt{{\texttt{rank}}}/\epsilon n\right)$, where ${\texttt{rank}}$ is the rank of the feature matrix in the GLM problem, $n$ is the number of data samples, and $\epsilon$ is the privacy parameter.


Making the Shoe Fit: Architectures, Initializations, and Tuning for Learning with Privacy

no code implementations25 Sep 2019 Nicolas Papernot, Steve Chien, Shuang Song, Abhradeep Thakurta, Ulfar Erlingsson

Because learning sometimes involves sensitive data, standard machine-learning algorithms have been extended to offer strong privacy guarantees for training data.

Privacy Preserving

Private Stochastic Convex Optimization with Optimal Rates

no code implementations NeurIPS 2019 Raef Bassily, Vitaly Feldman, Kunal Talwar, Abhradeep Thakurta

A long line of existing work on private convex optimization focuses on the empirical loss and derives asymptotically tight bounds on the excess empirical loss.

Amplification by Shuffling: From Local to Central Differential Privacy via Anonymity

no code implementations29 Nov 2018 Úlfar Erlingsson, Vitaly Feldman, Ilya Mironov, Ananth Raghunathan, Kunal Talwar, Abhradeep Thakurta

We study the collection of such statistics in the local differential privacy (LDP) model, and describe an algorithm whose privacy cost is polylogarithmic in the number of changes to a user's value.

Privacy Amplification by Iteration

no code implementations20 Aug 2018 Vitaly Feldman, Ilya Mironov, Kunal Talwar, Abhradeep Thakurta

In addition, we demonstrate that we can achieve guarantees similar to those obtainable using the privacy-amplification-by-sampling technique in several natural settings where that technique cannot be applied.

Model-Agnostic Private Learning via Stability

no code implementations14 Mar 2018 Raef Bassily, Om Thakkar, Abhradeep Thakurta

We provide a new technique to boost the average-case stability properties of learning algorithms to strong (worst-case) stability properties, and then exploit them to obtain private classification algorithms.

Binary Classification Classification +2

Differentially Private Matrix Completion Revisited

no code implementations ICML 2018 Prateek Jain, Om Thakkar, Abhradeep Thakurta

We provide the first provably joint differentially private algorithm with formal utility guarantees for the problem of user-level privacy-preserving collaborative filtering.

Collaborative Filtering Matrix Completion +1

To Drop or Not to Drop: Robustness, Consistency and Differential Privacy Properties of Dropout

no code implementations6 Mar 2015 Prateek Jain, Vivek Kulkarni, Abhradeep Thakurta, Oliver Williams

Moreover, using the above mentioned stability properties of dropout, we design dropout based differentially private algorithms for solving ERMs.

L2 Regularization

Private Empirical Risk Minimization Beyond the Worst Case: The Effect of the Constraint Set Geometry

1 code implementation20 Nov 2014 Kunal Talwar, Abhradeep Thakurta, Li Zhang

In addition, we show that when the loss function is Lipschitz with respect to the $\ell_1$ norm and $\mathcal{C}$ is $\ell_1$-bounded, a differentially private version of the Frank-Wolfe algorithm gives error bounds of the form $\tilde{O}(n^{-2/3})$.

Differentially Private Empirical Risk Minimization: Efficient Algorithms and Tight Error Bounds

1 code implementation27 May 2014 Raef Bassily, Adam Smith, Abhradeep Thakurta

We provide a separate set of algorithms and matching lower bounds for the setting in which the loss functions are known to also be strongly convex.

Analyze Gauss: Optimal Bounds for Privacy-Preserving Principal Component Analysis

1 code implementation1 May 2014 Cynthia Dwork, Kunal Talwar, Abhradeep Thakurta, Li Zhang

We show that the well-known, but misnamed, randomized response algorithm, with properly tuned parameters, provides a nearly optimal additive quality gap compared to the best possible singular subspace of A.

Privacy Preserving

Cannot find the paper you are looking for? You can Submit a new open access paper.