arXiv subjects

Wanrong Zhang

Publications and source records attributed to Wanrong Zhang.

At least 19 recordsLinked to original sources

arXiv2026

An Improved Privacy and Utility Analysis of Differentially Private SGD with Bounded Domain and Smooth Losses

Differentially Private Stochastic Gradient Descent (DPSGD) is widely used to protect sensitive data during the training of machine learning models, but its privacy guarantee often comes at a large cost of model performance due to the lack of tight theoretical bounds quantifying privacy loss. While recent efforts have achieved more accurate privacy guarantees, they still impose some assumptions prohibited from practical applications, such as convexity and complex parameter requirements, and rarely investigate in-depth the impact of privacy mechanisms on the model's utility. In this paper, we provide a rigorous privacy characterization for DPSGD with general L-smooth and non-convex loss functions, revealing converged privacy loss with iteration in bounded-domain cases. Specifically, we track the privacy loss over multiple iterations, leveraging the noisy smooth-reduction property, and further establish comprehensive convergence analysis in different scenarios. In particular, we show that for DPSGD with a bounded domain, (i) the privacy loss can still converge without the convexity assumption, (ii) a smaller bounded diameter can improve both privacy and utility simultaneously under certain conditions, and (iii) the attainable big-O order of the privacy utility trade-off for DPSGD with gradient clipping (DPSGD-GC) and for DPSGD-GC with bounded domain (DPSGD-DC) and mu-strongly convex population risk function, respectively. Experiments via membership inference attack (MIA) in a practical setting validate insights gained from the theoretical results.

cs.LG↗

arXiv2025

Estimating Privacy Leakage of Augmented Contextual Knowledge in Language Models

Language models (LMs) rely on their parametric knowledge augmented with relevant contextual knowledge for certain tasks, such as question answering. However, the contextual knowledge can contain private information that may be leaked when answering queries, and estimating this privacy leakage is not well understood. A straightforward approach of directly comparing an LM's output to the contexts can overestimate the privacy risk, since the LM's parametric knowledge might already contain the augmented contextual knowledge. To this end, we introduce *context influence*, a metric that builds on differential privacy, a widely-adopted privacy notion, to estimate the privacy leakage of contextual knowledge during decoding. Our approach effectively measures how each subset of the context influences an LM's response while separating the specific parametric knowledge of the LM. Using our context influence metric, we demonstrate that context privacy leakage occurs when contextual knowledge is out of distribution with respect to parametric knowledge. Moreover, we experimentally demonstrate how context influence properly attributes the privacy leakage to augmented contexts, and we evaluate how factors -- such as model size, context size, generation position, etc. -- affect context privacy leakage. The practical implications of our results will inform practitioners of the privacy risk associated with augmented contextual knowledge.

cs.CL↗

arXiv2025

Click Without Compromise: Online Advertising Measurement via Per User Differential Privacy

Online advertising is a cornerstone of the Internet ecosystem, with advertising measurement playing a crucial role in optimizing efficiency. Ad measurement entails attributing desired behaviors, such as purchases, to ad exposures across various platforms, necessitating the collection of user activities across these platforms. As this practice faces increasing restrictions due to rising privacy concerns, safeguarding user privacy in this context is imperative. Our work is the first to formulate the real-world challenge of advertising measurement systems with real-time reporting of streaming data in advertising campaigns. We introduce AdsBPC, a novel user-level differential privacy protection scheme for online advertising measurement results. This approach optimizes global noise power and results in a non-identically distributed noise distribution that preserves differential privacy while enhancing measurement accuracy. Through experiments on both real-world advertising campaigns and synthetic datasets, AdsBPC achieves a 33% to 95% increase in accuracy over existing streaming DP mechanisms applied to advertising measurement. This highlights our method's effectiveness in achieving superior accuracy alongside a formal privacy guarantee, thereby advancing the state-of-the-art in privacy-preserving advertising measurement.

cs.CR↗

arXiv2025

Interpreting Differential Privacy in Terms of Disclosure Risk

As the use of differential privacy (DP) becomes widespread, the development of effective tools for reasoning about the privacy guarantee becomes increasingly critical. In pursuit of this goal, we demonstrate novel relationships between DP and measures of statistical disclosure risk. We suggest how experts and non-experts can use these results to explain the DP guarantee, interpret DP composition theorems, select and justify privacy parameters, and identify worst-case adversary prior probabilities.

cs.CR↗

arXiv2025

TokenShapley: Token Level Context Attribution with Shapley Value

Large language models (LLMs) demonstrate strong capabilities in in-context learning, but verifying the correctness of their generated responses remains a challenge. Prior work has explored attribution at the sentence level, but these methods fall short when users seek attribution for specific keywords within the response, such as numbers, years, or names. To address this limitation, we propose TokenShapley, a novel token-level attribution method that combines Shapley value-based data attribution with KNN-based retrieval techniques inspired by recent advances in KNN-augmented LLMs. By leveraging a precomputed datastore for contextual retrieval and computing Shapley values to quantify token importance, TokenShapley provides a fine-grained data attribution approach. Extensive evaluations on four benchmarks show that TokenShapley outperforms state-of-the-art baselines in token-level attribution, achieving an 11-23% improvement in accuracy.

cs.CL↗

arXiv2025

Concurrent Composition for Interactive Differential Privacy with Adaptive Privacy-Loss Parameters

In this paper, we study the concurrent composition of interactive mechanisms with adaptively chosen privacy-loss parameters. In this setting, the adversary can interleave queries to existing interactive mechanisms, as well as create new ones. We prove that every valid privacy filter and odometer for noninteractive mechanisms extends to the concurrent composition of interactive mechanisms if privacy loss is measured using $(ε, δ)$-DP, $f$-DP, or Rényi DP of fixed order. Our results offer strong theoretical foundations for enabling full adaptivity in composing differentially private interactive mechanisms, showing that concurrency does not affect the privacy guarantees. We also provide an implementation for users to deploy in practice.

cs.CR↗

arXiv2024

When Focus Enhances Utility: Target Range LDP Frequency Estimation and Unknown Item Discovery

Local Differential Privacy (LDP) protocols enable the collection of randomized client messages for data analysis, without the necessity of a trusted data curator. Such protocols have been successfully deployed in real-world scenarios by major tech companies like Google, Apple, and Microsoft. In this paper, we propose a Generalized Count Mean Sketch (GCMS) protocol that captures many existing frequency estimation protocols. Our method significantly improves the three-way trade-offs between communication, privacy, and accuracy. We also introduce a general utility analysis framework that enables optimizing parameter designs. {Based on that, we propose an Optimal Count Mean Sketch (OCMS) framework that minimizes the variance for collecting items with targeted frequencies.} Moreover, we present a novel protocol for collecting data within unknown domain, as our frequency estimation protocols only work effectively with known data domain. Leveraging the stability-based histogram technique alongside the Encryption-Shuffling-Analysis (ESA) framework, our approach employs an auxiliary server to construct histograms without accessing original data messages. This protocol achieves accuracy akin to the central DP model while offering local-like privacy guarantees and substantially lowering computational costs.

cs.CR↗

arXiv2024

Meeting Utility Constraints in Differential Privacy: A Privacy-Boosting Approach

Data engineering often requires accuracy (utility) constraints on results, posing significant challenges in designing differentially private (DP) mechanisms, particularly under stringent privacy parameter $ε$. In this paper, we propose a privacy-boosting framework that is compatible with most noise-adding DP mechanisms. Our framework enhances the likelihood of outputs falling within a preferred subset of the support to meet utility requirements while enlarging the overall variance to reduce privacy leakage. We characterize the privacy loss distribution of our framework and present the privacy profile formulation for $(ε,δ)$-DP and Rényi DP (RDP) guarantees. We study special cases involving data-dependent and data-independent utility formulations. Through extensive experiments, we demonstrate that our framework achieves lower privacy loss than standard DP mechanisms under utility constraints. Notably, our approach is particularly effective in reducing privacy loss with large query sensitivity relative to the true answer, offering a more practical and flexible approach to designing differentially private mechanisms that meet specific utility constraints.

cs.CR↗

arXiv2024

Membership Inference Attacks and Privacy in Topic Modeling

Recent research shows that large language models are susceptible to privacy attacks that infer aspects of the training data. However, it is unclear if simpler generative models, like topic models, share similar vulnerabilities. In this work, we propose an attack against topic models that can confidently identify members of the training data in Latent Dirichlet Allocation. Our results suggest that the privacy risks associated with generative modeling are not restricted to large neural models. Additionally, to mitigate these vulnerabilities, we explore differentially private (DP) topic modeling. We propose a framework for private topic modeling that incorporates DP vocabulary selection as a pre-processing step, and show that it improves privacy while having limited effects on practical utility.

cs.CR↗

arXiv2024

Continual Release of Differentially Private Synthetic Data from Longitudinal Data Collections

Motivated by privacy concerns in long-term longitudinal studies in medical and social science research, we study the problem of continually releasing differentially private synthetic data from longitudinal data collections. We introduce a model where, in every time step, each individual reports a new data element, and the goal of the synthesizer is to incrementally update a synthetic dataset in a consistent way to capture a rich class of statistical properties. We give continual synthetic data generation algorithms that preserve two basic types of queries: fixed time window queries and cumulative time queries. We show nearly tight upper bounds on the error rates of these algorithms and demonstrate their empirical performance on realistically sized datasets from the U.S. Census Bureau's Survey of Income and Program Participation.

cs.DS↗

arXiv2024

Advancing Differential Privacy: Where We Are Now and Future Directions for Real-World Deployment

In this article, we present a detailed review of current practices and state-of-the-art methodologies in the field of differential privacy (DP), with a focus of advancing DP's deployment in real-world applications. Key points and high-level contents of the article were originated from the discussions from "Differential Privacy (DP): Challenges Towards the Next Frontier," a workshop held in July 2022 with experts from industry, academia, and the public sector seeking answers to broad questions pertaining to privacy and its implications in the design of industry-grade systems. This article aims to provide a reference point for the algorithmic and design decisions within the realm of privacy, highlighting important challenges and potential research directions. Covering a wide spectrum of topics, this article delves into the infrastructure needs for designing private systems, methods for achieving better privacy/utility trade-offs, performing privacy attacks and auditing, as well as communicating privacy with broader audiences and stakeholders.

cs.CR↗

arXiv2024

Safeguarding Data in Multimodal AI: A Differentially Private Approach to CLIP Training

The surge in multimodal AI's success has sparked concerns over data privacy in vision-and-language tasks. While CLIP has revolutionized multimodal learning through joint training on images and text, its potential to unintentionally disclose sensitive information necessitates the integration of privacy-preserving mechanisms. We introduce a differentially private adaptation of the Contrastive Language-Image Pretraining (CLIP) model that effectively addresses privacy concerns while retaining accuracy. Our proposed method, Dp-CLIP, is rigorously evaluated on benchmark datasets encompassing diverse vision-and-language tasks such as image classification and visual question answering. We demonstrate that our approach retains performance on par with the standard non-private CLIP model. Furthermore, we analyze our proposed algorithm under linear representation settings. We derive the convergence rate of our algorithm and show a trade-off between utility and privacy when gradients are clipped per-batch and the loss function does not satisfy smoothness conditions assumed in the literature for the analysis of DP-SGD.

cs.LG↗

arXiv2023

DP-Fast MH: Private, Fast, and Accurate Metropolis-Hastings for Large-Scale Bayesian Inference

Bayesian inference provides a principled framework for learning from complex data and reasoning under uncertainty. It has been widely applied in machine learning tasks such as medical diagnosis, drug design, and policymaking. In these common applications, data can be highly sensitive. Differential privacy (DP) offers data analysis tools with powerful worst-case privacy guarantees and has been developed as the leading approach in privacy-preserving data analysis. In this paper, we study Metropolis-Hastings (MH), one of the most fundamental MCMC methods, for large-scale Bayesian inference under differential privacy. While most existing private MCMC algorithms sacrifice accuracy and efficiency to obtain privacy, we provide the first exact and fast DP MH algorithm, using only a minibatch of data in most iterations. We further reveal, for the first time, a three-way trade-off among privacy, scalability (i.e. the batch size), and efficiency (i.e. the convergence rate), theoretically characterizing how privacy affects the utility and computational cost in Bayesian inference. We empirically demonstrate the effectiveness and efficiency of our algorithm in various experiments.

cs.LG↗

arXiv2023

Concurrent Composition Theorems for Differential Privacy

We study the concurrent composition properties of interactive differentially private mechanisms, whereby an adversary can arbitrarily interleave its queries to the different mechanisms. We prove that all composition theorems for non-interactive differentially private mechanisms extend to the concurrent composition of interactive differentially private mechanisms, whenever differential privacy is measured using the hypothesis testing framework of $f$-DP, which captures standard $(\eps,δ)$-DP as a special case. We prove the concurrent composition theorem by showing that every interactive $f$-DP mechanism can be simulated by interactive post-processing of a non-interactive $f$-DP mechanism. In concurrent and independent work, Lyu~\cite{lyu2022composition} proves a similar result to ours for $(\eps,δ)$-DP, as well as a concurrent composition theorem for Rényi DP. We also provide a simple proof of Lyu's concurrent composition theorem for Rényi DP. Lyu leaves the general case of $f$-DP as an open problem, which we solve in this paper.

cs.DS↗

arXiv2022

Bandit Change-Point Detection for Real-Time Monitoring High-Dimensional Data Under Sampling Control

In many real-world problems of real-time monitoring high-dimensional streaming data, one wants to detect an undesired event or change quickly once it occurs, but under the sampling control constraint in the sense that one might be able to only observe or use selected components data for decision-making per time step in the resource-constrained environments. In this paper, we propose to incorporate multi-armed bandit approaches into sequential change-point detection to develop an efficient bandit change-point detection algorithm based on the limiting Bayesian approach to incorporate a prior knowledge of potential changes. Our proposed algorithm, termed Thompson-Sampling-Shiryaev-Roberts-Pollak (TSSRP), consists of two policies per time step: the adaptive sampling policy applies the Thompson Sampling algorithm to balance between exploration for acquiring long-term knowledge and exploitation for immediate reward gain, and the statistical decision policy fuses the local Shiryaev-Roberts-Pollak statistics to determine whether to raise a global alarm by sum shrinkage techniques. Extensive numerical simulations and case studies demonstrate the statistical and computational efficiency of our proposed TSSRP algorithm.

stat.ME↗

arXiv2022

Private Sequential Hypothesis Testing for Statisticians: Privacy, Error Rates, and Sample Size

The sequential hypothesis testing problem is a class of statistical analyses where the sample size is not fixed in advance. Instead, the decision-process takes in new observations sequentially to make real-time decisions for testing an alternative hypothesis against a null hypothesis until some stopping criterion is satisfied. In many common applications of sequential hypothesis testing, the data can be highly sensitive and may require privacy protection; for example, sequential hypothesis testing is used in clinical trials, where doctors sequentially collect data from patients and must determine when to stop recruiting patients and whether the treatment is effective. The field of differential privacy has been developed to offer data analysis tools with strong privacy guarantees, and has been commonly applied to machine learning and statistical tasks. In this work, we study the sequential hypothesis testing problem under a slight variant of differential privacy, known as Renyi differential privacy. We present a new private algorithm based on Wald's Sequential Probability Ratio Test (SPRT) that also gives strong theoretical privacy guarantees. We provide theoretical analysis on statistical performance measured by Type I and Type II error as well as the expected sample size. We also empirically validate our theoretical results on several synthetic databases, showing that our algorithms also perform well in practice. Unlike previous work in private hypothesis testing that focused only on the classical fixed sample setting, our results in the sequential setting allow a conclusion to be reached much earlier, and thus saving the cost of collecting additional samples.

stat.ML↗

arXiv2021

Leakage of Dataset Properties in Multi-Party Machine Learning

Secure multi-party machine learning allows several parties to build a model on their pooled data to increase utility while not explicitly sharing data with each other. We show that such multi-party computation can cause leakage of global dataset properties between the parties even when parties obtain only black-box access to the final model. In particular, a ``curious'' party can infer the distribution of sensitive attributes in other parties' data with high accuracy. This raises concerns regarding the confidentiality of properties pertaining to the whole dataset as opposed to individual data records. We show that our attack can leak population-level properties in datasets of different types, including tabular, text, and graph data. To understand and measure the source of leakage, we consider several models of correlation between a sensitive attribute and the rest of the data. Using multiple machine learning models, we show that leakage occurs even if the sensitive attribute is not included in the training data and has a low correlation with other attributes or the target variable.

cs.LG↗

arXiv2021

Attribute Privacy: Framework and Mechanisms

Ensuring the privacy of training data is a growing concern since many machine learning models are trained on confidential and potentially sensitive data. Much attention has been devoted to methods for protecting individual privacy during analyses of large datasets. However in many settings, global properties of the dataset may also be sensitive (e.g., mortality rate in a hospital rather than presence of a particular patient in the dataset). In this work, we depart from individual privacy to initiate the study of attribute privacy, where a data owner is concerned about revealing sensitive properties of a whole dataset during analysis. We propose definitions to capture \emph{attribute privacy} in two relevant cases where global attributes may need to be protected: (1) properties of a specific dataset and (2) parameters of the underlying distribution from which dataset is sampled. We also provide two efficient mechanisms and one inefficient mechanism that satisfy attribute privacy for these settings. We base our results on a novel use of the Pufferfish framework to account for correlations across attributes in the data, thus addressing "the challenging problem of developing Pufferfish instantiations and algorithms for general aggregate secrets" that was left open by \cite{kifer2014pufferfish}.

cs.CR↗