arXiv subjects

Thierry Rakotoarivelo

Publications and source records attributed to Thierry Rakotoarivelo.

At least 19 recordsLinked to original sources

arXiv2025

Enhancing DPSGD via Per-Sample Momentum and Low-Pass Filtering

Differentially Private Stochastic Gradient Descent (DPSGD) is widely used to train deep neural networks with formal privacy guarantees. However, the addition of differential privacy (DP) often degrades model accuracy by introducing both noise and bias. Existing techniques typically address only one of these issues, as reducing DP noise can exacerbate clipping bias and vice-versa. In this paper, we propose a novel method, \emph{DP-PMLF}, which integrates per-sample momentum with a low-pass filtering strategy to simultaneously mitigate DP noise and clipping bias. Our approach uses per-sample momentum to smooth gradient estimates prior to clipping, thereby reducing sampling variance. It further employs a post-processing low-pass filter to attenuate high-frequency DP noise without consuming additional privacy budget. We provide a theoretical analysis demonstrating an improved convergence rate under rigorous DP guarantees, and our empirical evaluations reveal that DP-PMLF significantly enhances the privacy-utility trade-off compared to several state-of-the-art DPSGD variants.

cs.LG↗

arXiv2025

SoK: The Privacy Paradox of Large Language Models: Advancements, Privacy Risks, and Mitigation

Large language models (LLMs) are sophisticated artificial intelligence systems that enable machines to generate human-like text with remarkable precision. While LLMs offer significant technological progress, their development using vast amounts of user data scraped from the web and collected from extensive user interactions poses risks of sensitive information leakage. Most existing surveys focus on the privacy implications of the training data but tend to overlook privacy risks from user interactions and advanced LLM capabilities. This paper aims to fill that gap by providing a comprehensive analysis of privacy in LLMs, categorizing the challenges into four main areas: (i) privacy issues in LLM training data, (ii) privacy challenges associated with user prompts, (iii) privacy vulnerabilities in LLM-generated outputs, and (iv) privacy challenges involving LLM agents. We evaluate the effectiveness and limitations of existing mitigation mechanisms targeting these proposed privacy challenges and identify areas for further research.

cs.CR↗

arXiv2025

Privacy Meets Explainability: Managing Confidential Data and Transparency Policies in LLM-Empowered Science

As Large Language Models (LLMs) become integral to scientific workflows, concerns over the confidentiality and ethical handling of confidential data have emerged. This paper explores data exposure risks through LLM-powered scientific tools, which can inadvertently leak confidential information, including intellectual property and proprietary data, from scientists' perspectives. We propose "DataShield", a framework designed to detect confidential data leaks, summarize privacy policies, and visualize data flow, ensuring alignment with organizational policies and procedures. Our approach aims to inform scientists about data handling practices, enabling them to make informed decisions and protect sensitive information. Ongoing user studies with scientists are underway to evaluate the framework's usability, trustworthiness, and effectiveness in tackling real-world privacy challenges.

cs.HC↗

arXiv2025

Federated Learning with Differential Privacy: An Utility-Enhanced Approach

Federated learning has emerged as an attractive approach to protect data privacy by eliminating the need for sharing clients' data while reducing communication costs compared with centralized machine learning algorithms. However, recent studies have shown that federated learning alone does not guarantee privacy, as private data may still be inferred from the uploaded parameters to the central server. In order to successfully avoid data leakage, adopting differential privacy (DP) in the local optimization process or in the local update aggregation process has emerged as two feasible ways for achieving sample-level or user-level privacy guarantees respectively, in federated learning models. However, compared to their non-private equivalents, these approaches suffer from a poor utility. To improve the privacy-utility trade-off, we present a modification to these vanilla differentially private algorithms based on a Haar wavelet transformation step and a novel noise injection scheme that significantly lowers the asymptotic bound of the noise variance. We also present a holistic convergence analysis of our proposed algorithm, showing that our method yields better convergence performance than the vanilla DP algorithms. Numerical experiments on real-world datasets demonstrate that our method outperforms existing approaches in model utility while maintaining the same privacy guarantees.

cs.LG↗

arXiv2025

Multi-Objective Optimization for Privacy-Utility Balance in Differentially Private Federated Learning

Federated learning (FL) enables collaborative model training across distributed clients without sharing raw data, making it a promising approach for privacy-preserving machine learning. However, ensuring differential privacy (DP) in FL presents challenges due to the trade-off between model utility and privacy protection. Clipping gradients before aggregation is a common strategy to limit privacy loss, but selecting an optimal clipping norm is non-trivial, as excessively high values compromise privacy, while overly restrictive clipping degrades model performance. In this work, we propose an adaptive clipping mechanism that dynamically adjusts the clipping norm using a multi-objective optimization framework. By integrating privacy and utility considerations into the optimization objective, our approach balances privacy preservation with model accuracy. We theoretically analyze the convergence properties of our method and demonstrate its effectiveness through extensive experiments on MNIST, Fashion-MNIST, and CIFAR-10 datasets. Our results show that adaptive clipping consistently outperforms fixed-clipping baselines, achieving improved accuracy under the same privacy constraints. This work highlights the potential of dynamic clipping strategies to enhance privacy-utility trade-offs in differentially private federated learning.

cs.LG↗

arXiv2025

Adaptive Clipping for Privacy-Preserving Few-Shot Learning: Enhancing Generalization with Limited Data

In the era of data-driven machine-learning applications, privacy concerns and the scarcity of labeled data have become paramount challenges. These challenges are particularly pronounced in the domain of few-shot learning, where the ability to learn from limited labeled data is crucial. Privacy-preserving few-shot learning algorithms have emerged as a promising solution to address such pronounced challenges. However, it is well-known that privacy-preserving techniques often lead to a drop in utility due to the fundamental trade-off between data privacy and model performance. To enhance the utility of privacy-preserving few-shot learning methods, we introduce a novel approach called Meta-Clip. This technique is specifically designed for meta-learning algorithms, including Differentially Private (DP) model-agnostic meta-learning, DP-Reptile, and DP-MetaSGD algorithms, with the objective of balancing data privacy preservation with learning capacity maximization. By dynamically adjusting clipping thresholds during the training process, our Adaptive Clipping method provides fine-grained control over the disclosure of sensitive information, mitigating overfitting on small datasets and significantly improving the generalization performance of meta-learning models. Through comprehensive experiments on diverse benchmark datasets, we demonstrate the effectiveness of our approach in minimizing utility degradation, showcasing a superior privacy-utility trade-off compared to existing privacy-preserving techniques. The adoption of Adaptive Clipping represents a substantial step forward in the field of privacy-preserving few-shot learning, empowering the development of secure and accurate models for real-world applications, especially in scenarios where there are limited data availability.

cs.LG↗

arXiv2024

From 5G to 6G: A Survey on Security, Privacy, and Standardization Pathways

The vision for 6G aims to enhance network capabilities with faster data rates, near-zero latency, and higher capacity, supporting more connected devices and seamless experiences within an intelligent digital ecosystem where artificial intelligence (AI) plays a crucial role in network management and data analysis. This advancement seeks to enable immersive mixed-reality experiences, holographic communications, and smart city infrastructures. However, the expansion of 6G raises critical security and privacy concerns, such as unauthorized access and data breaches. This is due to the increased integration of IoT devices, edge computing, and AI-driven analytics. This paper provides a comprehensive overview of 6G protocols, focusing on security and privacy, identifying risks, and presenting mitigation strategies. The survey examines current risk assessment frameworks and advocates for tailored 6G solutions. We further discuss industry visions, government projects, and standardization efforts to balance technological innovation with robust security and privacy measures.

cs.CR↗

arXiv2024

Privacy risk in GeoData: A survey

With the ubiquitous use of location-based services, large-scale individual-level location data has been widely collected through location-awareness devices. The widespread exposure of such location data poses significant privacy risks to users, as it can lead to re-identification, the inference of sensitive information, and even physical threats. In this survey, we analyse different geomasking techniques proposed to protect individuals' privacy in geodata. We propose a taxonomy to characterise these techniques across various dimensions. We then highlight the shortcomings of current techniques and discuss avenues for future research. Our proposed taxonomy serves as a practical resource for data custodians, offering them a means to navigate the extensive array of existing privacy mechanisms and to identify those that align most effectively with their specific requirements.

cs.CR↗

arXiv2024

Decentralized Privacy Preservation for Critical Connections in Graphs

Many real-world interconnections among entities can be characterized as graphs. Collecting local graph information with balanced privacy and data utility has garnered notable interest recently. This paper delves into the problem of identifying and protecting critical information of entity connections for individual participants in a graph based on cohesive subgraph searches. This problem has not been addressed in the literature. To address the problem, we propose to extract the critical connections of a queried vertex using a fortress-like cohesive subgraph model known as $p$-cohesion. A user's connections within a fortress are obfuscated when being released, to protect critical information about the user. Novel merit and penalty score functions are designed to measure each participant's critical connections in the minimal $p$-cohesion, facilitating effective identification of the connections. We further propose to preserve the privacy of a vertex enquired by only protecting its critical connections when responding to queries raised by data collectors. We prove that, under the decentralized differential privacy (DDP) mechanism, one's response satisfies $(\varepsilon, δ)$-DDP when its critical connections are protected while the rest remains unperturbed. The effectiveness of our proposed method is demonstrated through extensive experiments on real-life graph datasets.

cs.CR↗

arXiv2024

Privacy at a Price: Exploring its Dual Impact on AI Fairness

The worldwide adoption of machine learning (ML) and deep learning models, particularly in critical sectors, such as healthcare and finance, presents substantial challenges in maintaining individual privacy and fairness. These two elements are vital to a trustworthy environment for learning systems. While numerous studies have concentrated on protecting individual privacy through differential privacy (DP) mechanisms, emerging research indicates that differential privacy in machine learning models can unequally impact separate demographic subgroups regarding prediction accuracy. This leads to a fairness concern, and manifests as biased performance. Although the prevailing view is that enhancing privacy intensifies fairness disparities, a smaller, yet significant, subset of research suggests the opposite view. In this article, with extensive evaluation results, we demonstrate that the impact of differential privacy on fairness is not monotonous. Instead, we observe that the accuracy disparity initially grows as more DP noise (enhanced privacy) is added to the ML process, but subsequently diminishes at higher privacy levels with even more noise. Moreover, implementing gradient clipping in the differentially private stochastic gradient descent ML method can mitigate the negative impact of DP noise on fairness. This mitigation is achieved by moderating the disparity growth through a lower clipping threshold.

cs.LG↗

arXiv2024

{A New Hope}: Contextual Privacy Policies for Mobile Applications and An Approach Toward Automated Generation

Privacy policies have emerged as the predominant approach to conveying privacy notices to mobile application users. In an effort to enhance both readability and user engagement, the concept of contextual privacy policies (CPPs) has been proposed by researchers. The aim of CPPs is to fragment privacy policies into concise snippets, displaying them only within the corresponding contexts within the application's graphical user interfaces (GUIs). In this paper, we first formulate CPP in mobile application scenario, and then present a novel multimodal framework, named SeePrivacy, specifically designed to automatically generate CPPs for mobile applications. This method uniquely integrates vision-based GUI understanding with privacy policy analysis, achieving 0.88 precision and 0.90 recall to detect contexts, as well as 0.98 precision and 0.96 recall in extracting corresponding policy segments. A human evaluation shows that 77% of the extracted privacy policy segments were perceived as well-aligned with the detected contexts. These findings suggest that SeePrivacy could serve as a significant tool for bolstering user interaction with, and understanding of, privacy policies. Furthermore, our solution has the potential to make privacy notices more accessible and inclusive, thus appealing to a broader demographic. A demonstration of our work can be accessed at https://cpp4app.github.io/SeePrivacy/

cs.CR↗

arXiv2023

Learn to Unlearn: A Survey on Machine Unlearning

Machine Learning (ML) models have been shown to potentially leak sensitive information, thus raising privacy concerns in ML-driven applications. This inspired recent research on removing the influence of specific data samples from a trained ML model. Such efficient removal would enable ML to comply with the "right to be forgotten" in many legislation, and could also address performance bottlenecks from low-quality or poisonous samples. In that context, machine unlearning methods have been proposed to erase the contributions of designated data samples on models, as an alternative to the often impracticable approach of retraining models from scratch. This article presents a comprehensive review of recent machine unlearning techniques, verification mechanisms, and potential attacks. We further highlight emerging challenges and prospective research directions (e.g. resilience and fairness concerns). We aim for this paper to provide valuable resources for integrating privacy, equity, andresilience into ML systems and help them "learn to unlearn".

cs.LG↗

arXiv2022

PADDLES: Phase-Amplitude Spectrum Disentangled Early Stopping for Learning with Noisy Labels

Convolutional Neural Networks (CNNs) have demonstrated superiority in learning patterns, but are sensitive to label noises and may overfit noisy labels during training. The early stopping strategy averts updating CNNs during the early training phase and is widely employed in the presence of noisy labels. Motivated by biological findings that the amplitude spectrum (AS) and phase spectrum (PS) in the frequency domain play different roles in the animal's vision system, we observe that PS, which captures more semantic information, can increase the robustness of DNNs to label noise, more so than AS can. We thus propose early stops at different times for AS and PS by disentangling the features of some layer(s) into AS and PS using Discrete Fourier Transform (DFT) during training. Our proposed Phase-AmplituDe DisentangLed Early Stopping (PADDLES) method is shown to be effective on both synthetic and real-world label-noise datasets. PADDLES outperforms other early stopping methods and obtains state-of-the-art performance.

cs.CV↗

arXiv2022

Private Graph Data Release: A Survey

The application of graph analytics to various domains has yielded tremendous societal and economical benefits in recent years. However, the increasingly widespread adoption of graph analytics comes with a commensurate increase in the need to protect private information in graph data, especially in light of the many privacy breaches in real-world graph data that was supposed to preserve sensitive information. This paper provides a comprehensive survey of private graph data release algorithms that seek to achieve the fine balance between privacy and utility, with a specific focus on provably private mechanisms. Many of these mechanisms are natural extensions of the Differential Privacy framework to graph data, but we also investigate more general privacy formulations like Pufferfish Privacy that address some of the limitations of Differential Privacy. We also provide a wide-ranging survey of the applications of private graph data release mechanisms to social networks, finance, supply chain, and health care. This survey paper and the taxonomy it provides should benefit practitioners and researchers alike in the increasingly important area of private analytics and data release.

cs.CR↗

arXiv2021

Enhancing Utility in the Watchdog Privacy Mechanism

This paper is concerned with enhancing data utility in the privacy watchdog method for attaining information-theoretic privacy. For a specific privacy constraint, the watchdog method filters out the high-risk data symbols through applying a uniform data regulation scheme, e.g., merging all high-risk symbols together. While this method entirely trades the symbols resolution off for privacy, we show that the data utility can be greatly improved by partitioning the high-risk symbols set and individually privatizing each subset. We further propose an agglomerative merging algorithm that finds a suitable partition of high-risk symbols: it starts with a singleton high-risk symbol, which is iteratively fused with others until the resulting subsets are private.~Numerical simulations demonstrate the efficacy of this algorithm in privately achieving higher utilities in the watchdog scheme.

cs.IT↗

arXiv2021

Realistic Differentially-Private Transmission Power Flow Data Release

For the modeling, design and planning of future energy transmission networks, it is vital for stakeholders to access faithful and useful power flow data, while provably maintaining the privacy of business confidentiality of service providers. This critical challenge has recently been somewhat addressed in [1]. This paper significantly extends this existing work. First, we reduce the potential leakage information by proposing a fundamentally different post-processing method, using public information of grid losses rather than power dispatch, which achieve a higher level of privacy protection. Second, we protect more sensitive parameters, i.e., branch shunt susceptance in addition to series impedance (complete pi-model). This protects power flow data for the transmission high-voltage networks, using differentially private transformations that maintain the optimal power flow consistent with, and faithful to, expected model behaviour. Third, we tested our approach at a larger scale than previous work, using the PGLib-OPF test cases [10]. This resulted in the successful obfuscation of up to a 4700-bus system, which can be successfully solved with faithfulness of parameters and good utility to data analysts. Our approach addresses a more feasible and realistic scenario, and provides higher than state-of-the-art privacy guarantees, while maintaining solvability, fidelity and feasibility of the system.

cs.CR↗

arXiv2020

Part I: Improving Computational Efficiency of Communication for Omniscience

Communication for omniscience (CO) refers to the problem where the users in a finite set $V$ observe a discrete multiple random source and want to exchange data over broadcast channels to reach omniscience, the state where everyone recovers the entire source. This paper studies how to improve the computational complexity for the problem of minimizing the sum-rate for attaining omniscience in $V$. While the existing algorithms rely on the submodular function minimization (SFM) techniques and complete in $O(|V|^2 \cdot \text{SFM}(|V|)$ time, we prove the strict strong map property of the nesting SFM problem. We propose a parametric (PAR) algorithm that utilizes the parametric SFM techniques and reduce the the complexity to $O(|V| \cdot \text{SFM}(|V|)$. The output of the PAR algorithm is in fact the segmented Dilworth truncation of the residual entropy for all minimum sum-rate estimates $α$, which characterizes the principal sequence of partitions (PSP) and solves some related problems: It not only determines the secret capacity, a dual problem to CO, and the network strength of a graph, but also outlines the hierarchical solution to a combinatorial clustering problem. \end{abstract}

cs.IT↗

arXiv2020

On Properties and Optimization of Information-theoretic Privacy Watchdog

We study the problem of privacy preservation in data sharing, where $S$ is a sensitive variable to be protected and $X$ is a non-sensitive useful variable correlated with $S$. Variable $X$ is randomized into variable $Y$, which will be shared or released according to $p_{Y|X}(y|x)$. We measure privacy leakage by \emph{information privacy} (also known as \emph{log-lift} in the literature), which guarantees mutual information privacy and differential privacy (DP). Let $\Xepsc \subseteq \X$ contain elements n the alphabet of $X$ for which the absolute value of log-lift (abs-log-lift for short) is greater than a desired threshold $\eps$. When elements $x\in \Xepsc$ are randomized into $y\in \Y$, we derive the best upper bound on the abs-log-lift across the resultant pairs $(s,y)$. We then prove that this bound is achievable via an \emph{$X$-invariant} randomization $p(y|x) = R(y)$ for $x,y\in\Xepsc$. However, the utility measured by the mutual information $I(X;Y)$ is severely damaged in imposing a strict upper bound $\eps$ on the abs-log-lift. To remedy this and inspired by the probabilistic ($\eps$, $δ$)-DP, we propose a relaxed ($\eps$, $δ$)-log-lift framework. To achieve this relaxation, we introduce a greedy algorithm which exempts some elements in $\Xepsc$ from randomization, as long as their abs-log-lift is bounded by $\eps$ with probability $1-δ$. Numerical results demonstrate efficacy of this algorithm in achieving a better privacy-utility tradeoff.

cs.IT↗