New method for uncertainty quantification in private SGD.
problem Uncertainty quantification for differentially private SGD.
method Block bootstrap for SGD under local differential privacy.
result Valid method for uncertainty quantification in private SGD.
Gradient clipping helps private SGD converge despite potential bias.
problem Gradient clipping in private SGD can bias convergence.
method Theoretical analysis and empirical evaluation of gradient clipping effects.
result Gradient clipping can prevent convergence to stationary points and introduces bias.
DP-RandP improves privacy-utility tradeoff in DP-SGD by learning priors from random processes.
problem Improving the performance of differentially private stochastic gradient descent (DP-SGD) on private data.
method A three-phase approach that learns priors from images generated by random processes and transfers these priors to private data.
result New state-of-the-art accuracy on CIFAR10, CIFAR100, MedMNIST, and ImageNet for various privacy budgets.
Private adaptive methods improve on traditional SGD for convex optimization.
problem Differential privacy constraints in gradient optimization.
method Differentially private variants of SGD and AdaGrad with adaptive stepsizes and non-isotropic clipping.
result Private AdaGrad outperforms private SGD in high-dimensional problems.
Paper improves privacy in SGD with low noise, achieving optimal risk rates.
problem Privacy-preserving machine learning with good performance.
method Differentially private SGD with low-noise analysis.
result Achieves optimal excess risk rates for non-smooth losses.
DPDMs use DP-SGD to generate private synthetic data.
problem Generating private synthetic data from sensitive datasets.
method Introduced DP-SGD for DMs, investigated DM parameterization and sampling, proposed noise multiplicity.
result Achieved state-of-the-art performance in image generation benchmarks.
Paper bridges statistical inference for DP-SGD, a privacy-preserving machine learning method.
problem Asymptotic statistical inference for Differentially Private Stochastic Gradient Descent (DP-SGD).
method Established asymptotic properties of SGD under randomized subsampling, extended to DP-SGD, proposed methods for constructing valid confidence intervals.
result Valid confidence intervals for DP-SGD output achieve nominal coverage rates while maintaining privacy.
While significant progress has been made separately on analytics systems for scalable stochastic gradient descent (SGD) and private SGD, none of the major scalable analytics frameworks have incorporated differentially private SGD. There are two inter-related issues for this disconnect between research and practice: (1)…
The paper proposes a method to learn differentially private variational autoencoders with term-wise gradient aggregation.
problem Learning variational autoencoders with differential privacy constraints and multiple divergences.
method Term-wise Differentially Private SGD (DP-SGD) that crafts randomized gradients for each loss term, keeping sensitivity at O(1).
result The method reduces the amount of noise needed for differential privacy, allowing better learning.
This work improves privacy-generalization bounds for DP-SGD.
problem Understanding the trade-off between privacy and generalization in machine learning.
method Proved a linear max-information bound for DP-SGD, derived PAC-Bayes and generalization bounds.
result Explicit and controlled complexity terms for DP-SGD-trained models.
Privacy preserving machine learning algorithms are crucial for learning models over user data to protect sensitive information. Motivated by this, differentially private stochastic gradient descent (SGD) algorithms for training machine learning models have been proposed. At each step, these algorithms modify the gradie…
DP-SGD can update fewer coordinates while maintaining privacy.
problem How to update fewer coordinates in DP-SGD without losing optimization signal.
method TP-TopK (Two-Phase TopK DP-SGD), a two-phase method for coordinate-sparse private training.
result Private training can update fewer coordinates without losing optimization signal, scaling noise with active dimension \(k\) instead of full dimension \(d\).
Projective DP-SGD reduces privacy error by identifying low-dimensional gradient subspaces.
problem Differentially private SGD's error rate scales with model's dimensionality, problematic for over-parameterized models.
method Projective DP-SGD, projecting noisy gradients to a low-dimensional subspace identified from a public dataset.
result The method reduces the dependence on model dimensionality, improving accuracy in high privacy regimes.
This paper analyzes convergence of DP-SGD with adaptive quantile clipping.
problem Empirical success of adaptive clipping methods lacks theoretical understanding.
method Comprehensive convergence analysis of SGD with quantile clipping (QC-SGD).
result Establishes theoretical guarantees for DP-QC-SGD, revealing relationships between quantile selection, step size, and convergence.
Paper relaxes SGD privacy and generalization guarantees for non-smooth convex losses.
problem Privacy and generalization in SGD for non-smooth convex losses.
method Relaxes Lipschitz and strong smoothness assumptions to Hölder smoothness, proving (ε,δ)-DP and optimal excess risk. result Noisy SGD with α-Hölder smooth losses achieves optimal excess risk with linear gradient complexity for α≥1/2. DP-SGD provides privacy guarantees for all data points, but we propose output-specific DP to better account for individual examples.
problem Accounting for individual privacy guarantees in DP-SGD.
method Output-specific (ε,δ)-DP and an efficient algorithm to investigate individual privacy across datasets. result Most examples enjoy stronger privacy guarantees than the worst-case bound, and there is a correlation between training loss and privacy parameter.
New method for differentially private optimization with general Lipschitz conditions.
problem Differentially private optimization under general Lipschitz conditions.
method Generalized Lipschitz condition for per-sample gradients, tuning clip norm based on minimum per-sample Lipschitz constant.
result Efficacy of the recommended clip norm tuning method verified on 8 datasets.
AdaDPIGU improves privacy in deep learning by adaptively clipping and pruning gradients.
problem Privacy in deep learning models, especially in high-dimensional settings.
method Importance-based gradient updates, adaptive clipping, differentially private SGD.
result AdaDPIGU achieves high accuracy while maintaining privacy, outperforming non-private models.
Optimizes sparse fine-tuning for privacy in neural networks.
problem Performance gap between DP-SGD and non-private fine-tuning.
method Optimization-based approach using private gradient information for selecting trainable weights.
result Our selection method leads to better prediction accuracy compared to existing approaches.
Differentially private learning avoids iterative optimization in parameter space.
problem Differentially private training of neural networks
method Hypernetworks trained on public datasets
result Significantly reduces noise in parameter space
This work analyzes DP-SGD for online LDP problems with practical convergence rates.
problem Analyzing DP-SGD for online LDP problems with practical convergence rates.
method Developed a general framework for online LDP model in stochastic optimization problems, conducted non-asymptotic convergence analysis.
result Comprehensive non-asymptotic convergence analysis of the proposed estimators in finite-sample situations.
New DP-CD method outperforms DP-SGD in solving composite DP-ERM problems.
problem Privacy-preserving machine learning with differential privacy.
method Differentially Private proximal Coordinate Descent (DP-CD) for composite Empirical Risk Minimization (ERM).
result DP-CD outperforms DP-SGD due to larger step sizes and better gradient exploitation.
Improved DP-SGD on large models achieves high accuracy on image classification tasks.
problem Differentially private image classification often degrades performance.
method Careful hyper-parameter tuning and signal propagation techniques.
result Achieved 81.4% top-1 accuracy on CIFAR-10 under (8, 10^{-5})-DP.
We propose an algorithm for the adaptation of the learning rate for stochastic gradient descent (SGD) that avoids the need for validation set use. The idea for the adaptiveness comes from the technique of extrapolation: to get an estimate for the error against the gradient flow which underlies SGD, we compare the resul…
A(DP)2SGD improves federated learning privacy and efficiency.
problem Privacy and efficiency in federated learning with asynchronous decentralized parallel SGD.
method Differentially private asynchronous decentralized parallel SGD (A(DP)2SGD) using R{é}nyi differential privacy. result Achieves optimal convergence rate and comparable model accuracy to SSGD but faster.
We consider the problem of minimizing a convex risk with stochastic subgradients guaranteeing ε-locally differentially private (ε-LDP). While it has been shown that stochastic optimization is possible with ε-LDP via the standard SGD (Song et al., 2013), its convergence rate largely depends on the learning rate, w…
Expands differential privacy mechanisms to include the Generalized Gaussian mechanism for improved private machine learning.
problem Improving privacy in machine learning algorithms while maintaining utility.
method Introduces and analyzes the Generalized Gaussian (GG) mechanism for differential privacy.
result The GG mechanism provides better performance than the Laplace and Gaussian mechanisms across various values of β.
Improved image generation with private data using perceptual features.
problem Difficulty in training generative models with differential privacy.
method Use pre-trained perceptual features to learn private data distribution.
result Generative models can generate high-quality images with low privacy budget (ϵ≈2). New algorithm corrects bias in LDP-released data for better analysis.
problem Bias in data released under Local Differential Privacy (LDP).
method Inverse Weierstrass Private Stochastic Gradient Descent (IWP-SGD).
result Converges to true population risk minimizer at O(1/n) rate. New privacy bounds for DP-SGD's last iterate, even with cyclic sampling.
problem Privacy of the last iterate in DP-SGD with cyclic sampling.
method Established new RDP upper bounds for the last iterate under realistic assumptions.
result Privacy bounds for DP-SGD's last iterate with cyclic sampling and clipping, even for nonconvex losses.
FedSel uses local differential privacy to protect data privacy in federated SGD.
problem Privacy leakage from gradients in federated SGD.
method Two-stage framework with top-k dimension selection and gradient accumulation.
result FedSel reduces privacy leakage by privately selecting important dimensions.
Large batch sizes reduce gradient variance in DP-SGD, improving privacy.
problem Understanding why large batch sizes work in DP-SGD.
method Decomposed total gradient variance into subsampling and noise-induced variances, proving batch size independence in the limit.
result Large batch sizes reduce effective total gradient variance, improving privacy in DP-SGD.
Paper proposes DP-SGD and DP-NSGD for differentially private non-convex optimization.
problem Mitigating privacy risks in large model learning.
method Clip or normalize per-sample gradients and add noise for differential privacy.
result Achieved convergence rate of gradient norm for non-convex optimization.
While machine learning has achieved remarkable results in a wide variety of domains, the training of models often requires large datasets that may need to be collected from different individuals. As sensitive information may be contained in the individual's dataset, sharing training data may lead to severe privacy conc…
New sampling scheme improves privacy in DP-SGD without sacrificing utility.
problem Suboptimal privacy amplification due to participation variance in Poisson subsampling.
method Balanced Iteration Subsampling (BIS) with structured randomness.
result BIS achieves stronger privacy amplification than Poisson subsampling and is optimal at both extremes of noise spectrum.
Improves zeroth-order optimization for private machine learning with public data.
problem High computation and memory cost of first-order DP methods.
method PAZO (Public Data Assisted Zeroth-order Optimization) framework.
result Achieves superior privacy/utility tradeoffs across tasks.
Differentially private random block coordinate descent improves utility in machine learning.
problem Lack of privacy in classical CD methods when handling sensitive information.
method Proposes a differentially private random block coordinate descent method using sketch matrices and importance sampling.
result Demonstrates improved convergence rates and utility guarantees compared to non-private methods.
DP-SGD analysis shows many datapoints leak less privacy than previously thought.
problem Empirical evidence suggests DP-SGD leaks less privacy than current analysis predicts.
method Developed a per-instance DP analysis of DP-SGD, introducing dependence on dataset distribution.
result Formally shows DP-SGD leaks significantly less privacy for many datapoints on common benchmarks.
Machine learning (ML) models trained by differentially private stochastic gradient descent (DP-SGD) have much lower utility than the non-private ones. To mitigate this degradation, we propose a DP Laplacian smoothing SGD (DP-LSSGD) to train ML models with differential privacy (DP) guarantees. At the core of DP-LSSGD is…
This paper benchmarks privacy-preserving machine learning on medical images.
problem Ensuring privacy in medical image analysis while maintaining model accuracy.
method Comparing Local-DP and DP-SGD for differential privacy in medical imagery.
result Theoretical privacy guarantees do not fully align with real-world performance.
The paper proposes a fair and private decentralized deep learning framework.
problem Ensuring fairness and privacy in collaborative deep learning.
method A reputation system and differential privacy are used. FDPDDL framework is built with two stages: initialisation and update.
result FDPDDL achieves high fairness, comparable accuracy to centralised and distributed frameworks, and better accuracy than standalone.
Large batch training with DP-SGD reduces model performance due to implicit bias.
problem Large batch training with DP-SGD reduces model performance.
method The study analyzes the phenomenon of implicit bias in Noisy-SGD (DP-SGD without clipping) and its theoretical solutions for linear models.
result The implicit bias in large batch training with DP-SGD is amplified by additional noise, similar to SGD.
New bounds for KANs trained with DP-SGD, addressing correlated noise.
problem Risk bounds for Kolmogorov-Arnold Networks trained by DP-SGD with correlated noise.
method Established new optimization and population risk analysis for KANs trained with DP-SGD, addressing correlated noise.
result First optimization and population risk analysis of correlated-noise mechanisms for DP training in non-convex settings, including neural networks.
Develops DP-SCD for stochastic coordinate descent, making it differentially private.
problem Privacy leak in auxiliary information during stochastic coordinate descent training.
method Develops DP-SCD, leveraging independent noise addition and decoupling/parallelizing coordinate updates.
result Demonstrates competitive performance against DP-SGD with less tuning.
Novel algorithm reduces privacy noise in machine learning.
problem High privacy noise in machine learning algorithms.
method Robust statistics, specifically median and trimmed mean, to bound sensitivity of SGD iterates.
result Improved privacy-utility trade-off with reduced noise and computational efficiency.
New stability bounds for SGD on nonsmooth convex losses.
problem Understanding stability of SGD on nonsmooth convex losses.
method Sharp upper and lower bounds for SGD and full-batch GD on nonsmooth convex losses.
result SGD can be less stable but still useful for generalization bounds.
New method optimizes privacy and compute trade-offs for deep learning.
problem Privacy and compute trade-offs in deep learning training.
method Decoupling privacy analysis and experimental behavior, using TAN and scaling laws for DP-SGD.
result Stronger privacy guarantees with significant reduction in computational budget.
Improved DMs with DP-SGD for generating private images.
problem Low privacy of DMs and lack of good privacy-utility tradeoff.
method Adopt LDMs with DP-SGD on attention modules of LDMs.
result Generates high-quality DP images with reduced parameters.