Privacy Mechanisms Research Articles

FedLAP-DP: Federated Learning by Sharing Differentially Private Loss Approximations

Conventional gradient-sharing approaches for federated learning (FL), such as FedAvg, rely on aggregation of local models and often face performance degradation under differential privacy (DP) mechanisms or data heterogeneity, which can be attributed to the inconsistency between the local and global objectives. To address this issue, we propose FedLAP-DP, a novel privacy-preserving approach for FL. Our formulation involves clients synthesizing a small set of samples that approximate local loss landscapes by simulating the gradients of real images within a local region. Acting as loss surrogates, these synthetic samples are aggregated on the server side to uncover the global loss landscape and enable global optimization. Building upon these insights, we offer a new perspective to enforce record-level differential privacy in FL. A formal privacy analysis demonstrates that FedLAP-DP incurs the same privacy costs as typical gradient-sharing schemes while achieving an improved trade-off between privacy and utility. Extensive experiments validate the superiority of our approach across various datasets with highly skewed distributions in both DP and non-DP settings. Beyond the promising performance, our approach presents a faster convergence speed compared to typical gradient-sharing methods and opens up the possibility of trading communication costs for better performance by sending a larger set of synthetic images. The source is available at https://github.com/a514514772/FedLAP-DP.

Distributed differentially private energy management of virtual power plants

Prosumers with flexible distributed energy resources (DERs) can be aggregated as a virtual power plant (VPP) to participate in the electricity market. However, the VPP’s energy management requires the prosumers to share individual data, which poses privacy concerns. This paper proposes a distributed differentially private energy management strategy for the VPP to maximize its profit by coordinating the prosumers and at the same time mitigate the privacy risks of local prosumers. Specifically, the coordination of prosumers is first formulated as an optimization problem, which includes the operation constraints of each individual prosumer and the VPP coordination process. On this basis, this paper proposes to solve the optimization problem with a two-level privacy protection strategy. For the individual level, a distributed solution framework is proposed to keep the private information of each prosumer preserved locally. For the communication level, a differential privacy mechanism is integrated into the information exchange process thus reducing the privacy leakage risk. Both theoretical and practical results are provided to verify the performance of the proposed method in terms of optimality and privacy protection levels.

Decentralized IoT data sharing: A blockchain-based federated learning approach with joint optimizations for efficiency and privacy

Blockchain-based Federated Learning (BCFL) is gaining significant attention as a promising decentralized data sharing technology with privacy protection. Most existing BCFL frameworks loosely couple blockchain and Federated Learning (FL). FL transforms data sharing into model sharing, while blockchain decentralizes the model aggregation and handles security verification. However, this simplistic overlay of the two technologies often involves third-party blockchain peers, leading to inefficient workflows and potential privacy attacks from malicious blockchain peers. Some studies have proposed differential privacy with noise addition to prevent privacy attacks, yet most do not allow clients to customize privacy budgets, impacting data availability and limiting BCFL’s application in data sharing. To address these challenges, we first introduce a novel tightly-coupled BCFL framework that integrates training and mining at the client side. Under this framework, a totally decentralized data sharing process is established. Moreover, a Personalised Differential Privacy (PDP) mechanism is devised, enabling clients to add Laplace noise to model gradients based on custom privacy budgets. To achieve optimal privacy budgets, a privacy optimization mechanism based on Stackelberg games is proposed. It establishes utility functions for data requesters and providers, models the process of utility optimization as a Stackelberg game process, and obtains the optimal privacy budget while maximizing utility for all participants. This facilitates flexible and on-demand privacy-protected data sharing. Extensive experiments validate the effectiveness of our approach in enhancing system efficiency and facilitating flexible on-demand privacy-protected data sharing, further solidifying BCFL’s potential in decentralized data sharing scenarios.

Differentially Private Average Consensus for Networks With Positive Agents.

This research paper addresses the problem of achieving differentially private average consensus for multiagent systems (MASs) consisting of positive agents. A novel randomized mechanism is introduced that employs nondecaying positive multiplicative truncated Gaussian noises to maintain the positivity and randomness of the state information over time. A time-varying controller is developed to achieve mean-square positive average consensus, and convergence accuracy is evaluated. The proposed mechanism is shown to preserve (ϵ,δ) -differential privacy of MASs, and the privacy budget is derived. Numerical examples are provided to illustrate the effectiveness of the proposed controller and privacy mechanism.

Nonlinear Analysis and Topological Approaches towards a Deep Intelligent Framework for Privacy Assurance of Autonomous IoT Systems

The proliferation of autonomous Internet of Things (IoT) systems powered by deep learning and artificial intelligence has ushered in a new era of data-driven convenience and automation. However, this innovation comes hand in hand with heightened concerns regarding data privacy. This paper presents a comprehensive framework for Privacy Assurance in Autonomous IoT Systems (PAIS), which amalgamates cutting-edge technologies and best practices to safeguard individual privacy in the era of pervasive connectivity and autonomous decision-making. The PAIS framework comprises multifaceted strategies to address privacy challenges in autonomous IoT ecosystems. It leverages advanced encryption techniques, robust access control mechanisms, and anonymization protocols to ensure data confidentiality. Moreover, differential privacy mechanisms are deployed to protect the identities of individuals within data streams. An innovative aspect of PAIS is the integration of AI-driven privacy monitoring, which constantly evaluates data for potential breaches and triggers immediate responses when anomalies are detected. Ensuring regulatory compliance is a paramount facet of the PAIS framework, as it aligns with evolving data protection regulations globally. Users are afforded control and transparency through intuitive interfaces, enabling them to manage their data usage preferences effectively. The ethical implications of AI in privacy preservation are also examined within the framework, emphasizing the importance of fairness and bias mitigation. PAIS promotes a privacy-by-design approach, where privacy considerations are integral to the inception and development of IoT systems. Regular risk assessments are performed to identify potential privacy vulnerabilities, ensuring that the framework adapts to emerging threats. Education and training programs are provided to stakeholders to foster awareness and adherence to privacy best practices.

Privacy-Preserving Gaze Data Streaming in Immersive Interactive Virtual Reality: Robustness and User Experience.

Eye tracking is routinely being incorporated into virtual reality (VR) systems. Prior research has shown that eye tracking data, if exposed, can be used for re-identification attacks [14]. The state of our knowledge about currently existing privacy mechanisms is limited to privacy-utility trade-off curves based on data-centric metrics of utility, such as prediction error, and black-box threat models. We propose that for interactive VR applications, it is essential to consider user-centric notions of utility and a variety of threat models. We develop a methodology to evaluate real-time privacy mechanisms for interactive VR applications that incorporate subjective user experience and task performance metrics. We evaluate selected privacy mechanisms using this methodology and find that re-identification accuracy can be decreased to as low as 14% while maintaining a high usability score and reasonable task performance. Finally, we elucidate three threat scenarios (black-box, black-box with exemplars, and white-box) and assess how well the different privacy mechanisms hold up to these adversarial scenarios. This work advances the state of the art in VR privacy by providing a methodology for end-to-end assessment of the risk of re-identification attacks and potential mitigating solutions. f.

Privacy-preserving data integration and sharing in multi-party IoT environments: An entity embedding perspective

The increasing prevalence of IoT applications highlights the urgency for insightful data fusion and information acquisition, boosting data integration and sharing needs. However, challenges arise in multi-party data sharing due to inherent data heterogeneity and privacy concerns. To address these issues, this paper discusses the feasibility of using embedding vectors as the semantic representation, aiming to enhance interoperability across diverse data sources and lay the foundation for natural language-based data querying. At the specific method level, this paper proposes an improved entity tree embedding algorithm to reduce information loss and ameliorate the representation of entity semantics. Additionally, a privacy preservation mechanism based on the entity embedding approach is introduced to provide privacy protection for text-based data. Experimental results on address data demonstrate the mechanism’s efficacy in achieving privacy protection comparable to the widely adopted 2D Laplace plane noise method. Furthermore, incorporating the entity tree embedding into the privacy mechanism could yield more robust and reasonable results regarding location privacy and service quality, signifying the validity of the entity embedding results.

Finding ε and δ of Traditional Disclosure Control Systems

This paper analyzes the privacy of traditional Statistical Disclosure Control (SDC) systems under a differential privacy interpretation. SDCs, such as cell suppression and swapping, promise to safeguard the confidentiality of data and are routinely adopted in data analyses with profound societal and economic impacts. Through a formal analysis and empirical evaluation of demographic data from real households in the U.S., the paper shows that widely adopted SDC systems not only induce vastly larger privacy losses than classical differential privacy mechanisms, but, they may also come at a cost of larger accuracy and fairness.

Data Security and Privacy Mechanism Based on Key Exposure Method over Cloud

Recent news reveal a powerful attacker which breaks data confidentiality by acquiring cryptographic keys, by means of coercion or backdoors in cryptographic software. Once the encryption key is exposed, the only viable measure to preserve data confidentiality is to limit the attacker’s access to the ciphertext. This may be achieved, for example, by spreading ciphertext blocks across servers in multiple administrative domains—thus assuming that the adversary cannot compromise all of them. Nevertheless, if data is encrypted with existing schemes, an adversary equipped with the encryption key, can still compromise a single server and decrypt the ciphertext blocks stored therein. In this paper, we study data confidentiality against an adversary which knows the encryption key and has access to a large fraction of the ciphertext blocks. To this end, we propose Bastion, a novel and efficient scheme that guarantees data confidentiality even if the encryption key is leaked and the adversary has access to almost all ciphertext blocks. We analyze the security of Bastion, and we evaluate its performance by means of a prototype implementation. We also discuss practical insights with respect to the integration of Bastion in commercial dispersed storage systems.

Fed-MPS: Federated learning with local differential privacy using model parameter selection for resource-constrained CPS

In Cyber-Physical Systems (CPS), distributed learning is essential for efficiently handling complex tasks when sufficient resources are available. However, when resources are limited, traditional distributed learning struggles to complete even simple tasks and presents a risk of privacy leakage. As a promising distributed learning paradigm, federated learning only requires the client to send the trained model to the server instead of private data, thereby preserving the client’s privacy to some extent. However, with the rapid development of artificial intelligence technology, attack methods such as inference attacks still cause privacy leakage for clients participating in federated learning. Moreover, due to its distributed learning nature, federated learning cannot escape the dilemma of model accuracy being constrained by resources. To address the aforementioned problems, this paper proposes a Federated local differential privacy scheme using Model Parameter Selection, named Fed-MPS, for resource-constrained CPS. Specifically, to resolve the issue of limited CPS resources, Fed-MPS adopts an update direction consistency-based parameter selection algorithm in federated learning to extract parameters that enhance model accuracy for subsequent training, thereby improving the final model accuracy and reducing communication overhead. Furthermore, Fed-MPS applies the local differential privacy mechanism to further enhance clients’ privacy. By adding noise only to the chosen parameters, the privacy budget is significantly reduced while ensuring model accuracy. Through privacy analysis, we prove that the proposed Fed-MPS scheme satisfies (ϵ,δ)−DP. Additionally, convergence analysis guarantees that Fed-MPS will converge to the global optimum with a convergence ratio of O(1T2) within T rounds of federated learning. Extensive experiments on prominent benchmark datasets Cifar10, Mnist, and FashionMNIST demonstrate that, compared with baseline schemes, the proposed Fed-MPS provides higher model accuracy for CPS under resource constraints.

Mechanisms for Robust Local Differential Privacy.

We consider privacy mechanisms for releasing data X=(S,U), where S is sensitive and U is non-sensitive. We introduce the robust local differential privacy (RLDP) framework, which provides strong privacy guarantees, while preserving utility. This is achieved by providing robust privacy: our mechanisms do not only provide privacy with respect to a publicly available estimate of the unknown true distribution, but also with respect to similar distributions. Such robustness mitigates the potential privacy leaks that might arise from the difference between the true distribution and the estimated one. At the same time, we mitigate the utility penalties that come with ordinary differential privacy, which involves making worst-case assumptions and dealing with extreme cases. We achieve robustness in privacy by constructing an uncertainty set based on a Rényi divergence. By analyzing the structure of this set and approximating it with a polytope, we can use robust optimization to find mechanisms with high utility. However, this relies on vertex enumeration and becomes computationally inaccessible for large input spaces. Therefore, we also introduce two low-complexity algorithms that build on existing LDP mechanisms. We evaluate the utility and robustness of the mechanisms using numerical experiments and demonstrate that our mechanisms provide robust privacy, while achieving a utility that is close to optimal.

Proof-of-authority-based secure and efficient aggregation with differential privacy for federated learning in industrial IoT

The industrial internet of things (IIoT) uses connected devices and sensors to improve efficiency in industry, but increased reliance on these systems makes them prone to faults. To ensure the reliability of IIoT systems, robust fault detection is necessary, which can be provided by artificial intelligence (AI). The current state-of-the-art approach for AI in IIoT is mainly focused on centralized learning, which is inefficient due to its high communication cost. Federated learning (FL) addresses this limitation by enabling distributed training without exposing individual information. In order to provide secure and efficient FL for the IIoT environment, several research studies have investigated the use of blockchain networks. However, most of these studies ignore the processing time metric, which is crucial for IIoT networks. In this paper, we propose a secure and efficient parameter aggregation technique that enhances trust, security, and privacy for IIoT. We use a lightweight smart contract deployed with a proof-of-authority (PoA)-based blockchain in combination with a Gaussian differential privacy mechanism. We evaluate our proposed system using a real bearing fault dataset and demonstrate that it is able to provide a secure aggregation process with an accuracy of 94.00% and a processing time of 1.54 s, which is suitable for the IIoT environment with a private blockchain network.

The Bounded Gaussian Mechanism for Differential Privacy

The Gaussian mechanism is one differential privacy mechanism commonly used to protect numerical data. However, it may be ill-suited to some applications because it has unbounded support and thus can produce invalid numerical answers to queries, such as negative ages or human heights in the tens of meters. One can project such private values onto valid ranges of data, though such projections lead to the accumulation of private query responses at the boundaries of such ranges, thereby harming accuracy. Motivated by the need for both privacy and accuracy over bounded domains, we present a bounded Gaussian mechanism for differential privacy, which has support only on a given region. We present both univariate and multivariate versions of this mechanism and illustrate a significant reduction in variance relative to comparable existing work.

Towards mitigating uncann(eye)ness in face swaps via gaze-centric loss terms

Advances in face swapping have enabled the automatic generation of highly realistic faces. Yet face swaps are perceived differently than when looking at real faces, with key differences in viewer behavior surrounding the eyes. Face swapping algorithms generally place no emphasis on the eyes, relying on pixel or feature matching losses that consider the entire face to guide the training process. We further investigate viewer perception of face swaps, focusing our analysis on the presence of an uncanny valley effect. We additionally propose a novel loss equation for the training of face swapping models, leveraging a pretrained gaze estimation network to directly improve representation of the eyes. We confirm that viewed face swaps do elicit uncanny responses from viewers. Our proposed improvements significant reduce viewing angle errors between face swaps and their source material. Our method additionally reduces the prevalence of the eyes as a deciding factor when viewers perform deepfake detection tasks. Our findings have implications on face swapping for special effects, as digital avatars, as privacy mechanisms, and more; negative responses from users could limit effectiveness in said applications. Our gaze improvements are a first step towards alleviating negative viewer perceptions via a targeted approach.

Enhancing Patient Outcome Predictions Through Differential Privacy with Optimized RNN Model Applied to Electronic Health Record

This study introduces a novel method for predicting patient admission and discharge events using Electronic Health Records (EHR) while prioritizing the protection of personal privacy. Our approach utilizes a Simple Recurrent Neural Network (RNN) model enhanced with a Differential Privacy mechanism to strike a balance between high-accuracy outcome predictions and the confidentiality of patient data. At the core of our methodology is the application of a Simple RNN to EHR data, which facilitates the prediction of whether a patient will be admitted to or discharged from a healthcare facility. To strengthen data confidentiality, we integrate Differential Privacy by injecting controlled noise into the dataset, ensuring that our model’s predictions preserve the privacy of individual patient records. We conducted experiments using six different classifiers to implement our privacy-preserving prediction strategy. Among these, the Random Forest classifier emerged as the most accurate, proving the effectiveness of our method in providing reliable predictions without compromising privacy. In contrast, the XG Boost classifier showed the least precision, indicating some limitations in its ability to balance privacy with predictive accuracy. This research significantly advances the field of healthcare informatics by presenting a sophisticated solution that combines cutting-edge predictive models with stringent privacy safeguards. Our findings highlight the critical need to maintain a delicate balance between achieving precise clinical predictions and upholding the moral responsibility to protect patient privacy in the modern landscape of digital health records.

Private bipartite consensus control for multi-agent systems: A hierarchical differential privacy scheme

This paper addresses the differentially private bipartite consensus control problem for general linear discrete-time multi-agent systems (MASs), where the initial states of the agents are the sensitive data to be protected from potential eavesdroppers. In response to the cooperative–competitive relationship among the agents, a novel hierarchical differential privacy mechanism (HDPM) is proposed, which consists of a classifier and a pair of input and output noise injectors. The HDPM and the bipartite consensus controller are co-designed to ensure the ultimate mean-square bipartite consensus while preserving differential privacy. Finally, numerical simulations are conducted to illustrate the effectiveness of the proposed algorithm.

Shedding Light on the Inadequate Protection of Internet Users’ Right to Privacy under Kuwaiti Law

Abstract Digital technology such as AI has enabled private companies to collect and analyse people’s personal information, which may infer hidden private data about the individual, such as medical data or psychological state. The inferred private information can be sold to third parties or used to target advertisements without the individual’s knowledge and explicit consent. Private companies, therefore, pose a risk to internet users’ right to privacy in the digital world. As such, the legal system should have a comprehensive Data Privacy law that provides internet users with adequate legal protection for preserving their right to privacy against private companies. Still, Kuwait lacks comprehensive Data Privacy. The only two data privacy mechanisms are embodied in chapter 7 under the Electronic Transaction Act (ETA) No. 20 of 2014 and the Regulation of User Protection and Privacy issued by the Communication and Information Technology Regulatory Authority (CITRA). This paper finds that the fragmented Kuwaiti laws concerning the individual’s right to privacy inadequately protect internet users’ right to privacy against the risk posed by private companies, and therefore, legislative interventions are needed. This paper concludes with proposing essential recommendations for the Kuwaiti Legislative Authority to issue a comprehensive data privacy law, using European Union’s General Data Protection Regulation as a guidance framework (GDPR), to strengthen internet users’ right to privacy (as control over their personal information) against the risk private companies pose.

GeoPM-DMEIRL: A deep inverse reinforcement learning security trajectory generation framework with serverless computing

Vehicle trajectory data is essential for traffic management and location-based services. However, the release of trajectories raises serious privacy concerns because they contain sensitive information, such as home addresses and workplaces, etc, making it indispensable to consider privacy protection when releasing trajectory data. It is urgent to reconcile data utility with privacy in trajectory generation calls for advanced methods that adhere to real-world traffic rules. Leveraging serverless computing for deep learning provides an efficient solution, bypassing infrastructure complexities to handle the scale trajectory data. In this paper, a Geo-Piecewise Mechanism Deep Maximum Entropy Inverse Reinforcement Learning (GeoPM-DMEIRL) is proposed for secure trajectory generation, which consists of three key components, namely GeoPM, A2C reinforcement learning and DMEIRL. Firstly, GeoPM is a novel local differential privacy mechanism that incorporates Geo-aware gridding techniques. The Piecewise Mechanism is used to perturb the user trajectory while ensuring that the perturbed trajectory conforms to real-world traffic rules. Secondly, an A2C reinforcement learning network is refined to train the optimal trajectory generation strategy. Thirdly, Deep Maximum Entropy Inverse Reinforcement Learning is improved to train the weights of the A2C reward function. Finally, real-world data are collected in the experiments and the serverless computing platform AWS Lambda is used for training the reinforcement learning models. Experimental results show that our proposed GeoPM-DMEIRL framework can effectively resist user re-identification attacks, which can improve the utility by an average of 54.605% and enhance the privacy by an average of 30.678%. Meanwhile, GeoPM-DMEIRL is able to maintain the utility of the data while protecting the privacy of the user’s trajectories.

Differentially private distributed online optimization via push-sum one-point bandit dual averaging

This paper focuses on the distributed online optimization problem in multi-agent systems considering privacy preservation. Each agent exchanges local information with neighboring agents on the strongly connected time-varying directed graphs. Since the process of information transmission is prone to information leakage, a distributed push-sum dual averaging algorithm based on the differential privacy mechanism is proposed to protect the privacy of the data. In addition, to handle situations where the gradient information of the node cost function is unknown, the one-point gradient estimation is designed to calculate the true gradient information and guide the update of the decision variables. With the appropriate choice of the stepsizes and the exploration parameters, the algorithm can effectively protect the privacy of agents while achieving sublinear regret with the convergence rate O(T34). Furthermore, this paper also explores the effect of one-point estimation parameters on the regret in the online setting and investigates the relation between the convergence effect of individual regret and differential privacy levels. Finally, several federated learning experiments were conducted to verify the efficacy of the algorithm.

Graph rewriting primitives for semantic graph databases sanitization

Due to the rapid proliferation of data online, an important quantity of private or sensitive informations is being stored as linked data in graph databases (e.g., represented as RDF). For such databases to be shared without jeopardizing privacy, they must first undergo a process known as database sanitization. During this process, databases are transformed following graph transformations that are usually described informally or through ad-hoc processes. However, a more thourough formalization of these transformations would aid in analysing the sanitization process, ensuring its correctness, and demonstrating the resulting privacy guarantees. This paper is an effort toward bridging the gap between the rigorous graph rewriting approaches and graph sanitization. We propose a graph transformation language to serve as a basis for constructing various sanitization mechanisms. This language relies on a set of elementary transformation operators formalized using a generic algebraic graph rewriting approach. Our language takes into account semantic and supports the equivalent of WHERE and EXCEPT clauses. As a proof of concept, we use these operators to implement two mechanisms from the literature, one generic (Local Differential Privacy) and one specifically introduced for semantic graph databases (sensitive attribute masking through anatomization). We propose an open-sourced tool implementing the elementary operators and the privacy mechanisms we derive from them relying on the Attributed Graph Grammar System (AGG) and its java API, providing a concrete tool implementing formal graph rewriting mechanisms to sanitize semantic graph databases. We present experimental results on this implementation regarding both proposed schemes and discuss its efficiency and scalability.