Strong Privacy Research Articles

Recommendations in Smart Devices Using Federated Tensor Learning

Recommendations based on prediction of user preferences from partial information are widely used in various applications. However, recommendations using smart devices have some challenges related to limited data device resources, data sparsity, and data privacy. Since there are many multidimensional data in smart devices, recommendations may collect a large amount of user private data. In this article, we study privacy-preserving recommendations with high-dimensional tensor data in smart devices. First, we propose a federated tensor completion scheme to infer the user’s preferences and we prove that this scheme satisfies the differential privacy guarantee. Our scheme consists of a global update and a local update, which reduce information exposure and guarantee local data privacy. Second, we mathematically analyze the privacy and utility of the proposed algorithm. Third, we provide empirical evaluations on synthetic data sets and real-world data sets. Results show that our scheme has a low recovery error and provides strong privacy protection.

Expert insights on digital contact tracing: interviews with contact tracing policy professionals in New Zealand.

Digital contact tracing (DCT) is the application of digital tools to assist with identifying and informing close contacts of a COVID-19 case. DCT is a potential solution to capacity constraints of current manual contact tracing processes. Expert opinion from contact tracing professionals rarely informs public discourse on the benefits and limitations of DCT solutions. Three focus groups were undertaken in New Zealand to understand benefits and limitations of DCT solutions from contact tracing professionals. One was with the National Investigation and Tracing Centre (NITC) and two were with Public Health Units (PHUs). Participants highlighted four key themes including: (i) equity, (ii) privacy, (iii) communication and public perception and (iv) the operational model. Participants were concerned DCT solutions could exacerbate existing health inequities due to lack of access to, or familiarity with, technology. Poor communication and public understanding of DCT were seen as a major threat to both the efficacy of DCT solutions and the wider COVID-19 response. Most importantly, end-users were cautious of the operational model for DCT data that might: (i) attempt to replace manual processes that cannot or should not be automated by technology (case investigations, follow-ups); (ii) place undue burden on citizens and (iii) increase the workload for the current system beyond its capacity, for unproven or limited benefit. To be effective, contact tracing professionals believed DCT technologies must have strong privacy safeguards, a clear and simple communication strategy, interoperability with the existing contact tracing system and a foundation of health equity.

Toward Scalable and Privacy-preserving Deep Neural Network via Algorithmic-Cryptographic Co-design

Deep Neural Networks (DNNs) have achieved remarkable progress in various real-world applications, especially when abundant training data are provided. However, data isolation has become a serious problem currently. Existing works build privacy-preserving DNN models from either algorithmic perspective or cryptographic perspective. The former mainly splits the DNN computation graph between data holders or between data holders and server, which demonstrates good scalability but suffers from accuracy loss and potential privacy risks. In contrast, the latter leverages time-consuming cryptographic techniques, which has strong privacy guarantee but poor scalability. In this article, we propose SPNN—a Scalable and Privacy-preserving deep Neural Network learning framework, from an algorithmic-cryptographic co-perspective. From algorithmic perspective, we split the computation graph of DNN models into two parts, i.e., the private-data-related computations that are performed by data holders and the rest heavy computations that are delegated to a semi-honest server with high computation ability. From cryptographic perspective, we propose using two types of cryptographic techniques, i.e., secret sharing and homomorphic encryption, for the isolated data holders to conduct private-data-related computations privately and cooperatively. Furthermore, we implement SPNN in a decentralized setting and introduce user-friendly APIs. Experimental results conducted on real-world datasets demonstrate the superiority of our proposed SPNN.

BlockMaze: An Efficient Privacy-Preserving Account-Model Blockchain Based on zk-SNARKs

The disruptive blockchain technology is expected to have broad applications in many areas due to its advantages of transparency, fault tolerance, and decentralization, but the open nature of blockchain also introduces severe privacy issues. Since anyone can deduce private information about relevant accounts, different privacy-preserving techniques have been proposed for cryptocurrencies under the UTXO model, e.g., Zerocash and Monero. However, it is more challenging to protect privacy for account-model blockchains (e.g., Ethereum) since it is much easier to link accounts in the account-model blockchain. In this article, we propose <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">BlockMaze</i> , an efficient privacy-preserving account-model blockchain based on zk-SNARKs. Along with dual-balance model, BlockMaze achieves strong privacy guarantees by hiding account balances, transaction amounts, and linkage between senders and recipients. Moreover, we provide formal security definitions and prove the security of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">BlockMaze</i> . Finally, we implement a prototype of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">BlockMaze</i> based on Libsnark and Go-Ethereum, and conduct extensive experiments to evaluate its performance. Our 300-node experiment results show that BlockMaze has high efficiency in computation and transaction throughput: one transaction verification takes about 14.2 ms, one transaction generation takes 6.1-18.6 seconds, and its throughput is around 20 TPS.

A Novel Privacy Paradigm for Improving Serial Data Privacy.

Protecting the privacy of individuals is of utmost concern in today’s society, as inscribed and governed by the prevailing privacy laws, such as GDPR. In serial data, bits of data are continuously released, but their combined effect may result in a privacy breach in the whole serial publication. Protecting serial data is crucial for preserving them from adversaries. Previous approaches provide privacy for relational data and serial data, but many loopholes exist when dealing with multiple sensitive values. We address these problems by introducing a novel privacy approach that limits the risk of privacy disclosure in republication and gives better privacy with much lower perturbation rates. Existing techniques provide a strong privacy guarantee against attacks on data privacy; however, in serial publication, the chances of attack still exist due to the continuous addition and deletion of data. In serial data, proper countermeasures for tackling attacks such as correlation attacks have not been taken, due to which serial publication is still at risk. Moreover, protecting privacy is a significant task due to the critical absence of sensitive values while dealing with multiple sensitive values. Due to this critical absence, signatures change in every release, which is a reason for attacks. In this paper, we introduce a novel approach in order to counter the composition attack and the transitive composition attack and we prove that the proposed approach is better than the existing state-of-the-art techniques. Our paper establishes the result with a systematic examination of the republication dilemma. Finally, we evaluate our work using benchmark datasets, and the results show the efficacy of the proposed technique.

Sharing personal ECG time-series data privately.

Emerging technologies (eg, wearable devices) have made it possible to collect data directly from individuals (eg, time-series), providing new insights on the health and well-being of individual patients. Broadening the access to these data would facilitate the integration with existing data sources (eg, clinical and genomic data) and advance medical research. Compared to traditional health data, these data are collected directly from individuals, are highly unique and provide fine-grained information, posing new privacy challenges. In this work, we study the applicability of a novel privacy model to enable individual-level time-series data sharing while maintaining the usability for data analytics. We propose a privacy-protecting method for sharing individual-level electrocardiography (ECG) time-series data, which leverages dimensional reduction technique and random sampling to achieve provable privacy protection. We show that our solution provides strong privacy protection against an informed adversarial model while enabling useful aggregate-level analysis. We conduct our evaluations on 2 real-world ECG datasets. Our empirical results show that the privacy risk is significantly reduced after sanitization while the data usability is retained for a variety of clinical tasks (eg, predictive modeling and clustering). Our study investigates the privacy risk in sharing individual-level ECG time-series data. We demonstrate that individual-level data can be highly unique, requiring new privacy solutions to protect data contributors. The results suggest our proposed privacy-protection method provides strong privacy protections while preserving the usefulness of the data.

MLCT: A multi‐level contact tracing scheme with strong privacy

AbstractWith the outbreak of Covid‐19, both people's health and the world economy are facing great challenges. Contact tracing scheme based on Bluetooth of smartphones has been regarded as a viable way to mitigate the spread of Covid‐19. The existing schemes mainly belong to the centralized or the decentralized structure, both of which have their own limitations. It is infeasible for the existing schemes to balance the different demands of governments and users for user privacy and tracing efficiency at different periods of the epidemic. In this paper, we propose a hybrid contact tracing scheme named MLCT (multi‐level contact tracing scheme) which is mainly based on short group signature. MLCT provides multiple privacy levels by applying anonymous credential technology and secret sharing technology to desensitize user identity privacy and encounter privacy. Comparing to the previous schemes, MLCT fully considers the different demands of the government, patients, and close contacts for user privacy and tracing efficiency in the different stages of Covid‐19. The experimental results show viability in terms of the required resource from both server and mobile phone perspectives. And the security analysis demonstrates that MLCT can achieve the five targets security goals. It is expected that MLCT can contribute to the design and development of contact tracing schemes.

Public support for counterterrorism efforts using probabilistic computing technologies to decipher terrorist communication on the internet.

Advancements in big data analytics offer new avenues for the analysis and deciphering of suspicious activities on the internet. One promising new technology to increase the identification of terrorism threats is based on probabilistic computing. The technology promises to provide more efficient problem solutions in encryption and cybersecurity. Probabilistic computing technologies use large amounts of data, though, which raises potential privacy concerns. A study (N = 1,023) was conducted to survey public support for using probabilistic computing technologies to increase counterterrorism efforts. Overall, strong support was found for the use of publicly available personal information (e.g., personal websites). Regarding private personal information (e.g., online conversations), respondents perceived it to be more appropriate to use information from out-group members (non-American citizens) than from in-group members (American citizens). In line with a social-identity account, this form of in-group favoritism was strongest among respondents displaying a combination of strong national identities and strong privacy concerns.

Secure and Privacy-Preserving Federated Learning via Co-Utility

The decentralized nature of federated learning, that often leverages the power of edge devices, makes it vulnerable to attacks against privacy and security. The privacy risk for a peer is that the model update she computes on her private data may, when sent to the model manager, leak information on those private data. Even more obvious are security attacks, whereby one or several malicious peers return wrong model updates in order to disrupt the learning process and lead to a wrong model being learned. In this article, we build a federated learning framework that offers privacy to the participating peers as well as security against the Byzantine and poisoning attacks. Our framework consists of several protocols that provide strong privacy to the participating peers via unlinkable anonymity and that are rationally sustainable based on the co-utility property. In other words, no rational party is interested in deviating from the proposed protocols. We leverage the notion of co-utility to build a decentralized co-utile reputation management system that provides incentives for parties to adhere to the protocols. Unlike privacy protection via differential privacy, our approach preserves the values of model updates and, hence, the accuracy of plain federated learning; unlike privacy protection via update aggregation, our approach preserves the ability to detect bad model updates while substantially reducing the computational overhead compared to methods based on homomorphic encryption.

Federated learning and differential privacy for medical image analysis

The artificial intelligence revolution has been spurred forward by the availability of large-scale datasets. In contrast, the paucity of large-scale medical datasets hinders the application of machine learning in healthcare. The lack of publicly available multi-centric and diverse datasets mainly stems from confidentiality and privacy concerns around sharing medical data. To demonstrate a feasible path forward in medical image imaging, we conduct a case study of applying a differentially private federated learning framework for analysis of histopathology images, the largest and perhaps most complex medical images. We study the effects of IID and non-IID distributions along with the number of healthcare providers, i.e., hospitals and clinics, and the individual dataset sizes, using The Cancer Genome Atlas (TCGA) dataset, a public repository, to simulate a distributed environment. We empirically compare the performance of private, distributed training to conventional training and demonstrate that distributed training can achieve similar performance with strong privacy guarantees. We also study the effect of different source domains for histopathology images by evaluating the performance using external validation. Our work indicates that differentially private federated learning is a viable and reliable framework for the collaborative development of machine learning models in medical image analysis.

Hiding payments in lightning network with approximate differentially private payment channels

Payment Channel Networks (PCN) form a class of techniques created to solve the scalability problems that permissionless blockchains such as Bitcoin face. The Lightning Network (LN) is the biggest PCN to date. As of September 2021, LN is an interconnected network with 27 thousand nodes and 71 thousand channels, representing a total payment capacity of 2500 bitcoins (114M USD). LN nodes establish trustless payment channels by locking funds in a funding transaction. The distribution of the funds can be modified with near-instant transactions without broadcasting them to the blockchain. By chaining channels together, it is possible for two nodes to transact without having a direct channel between them. This results in far fewer transactions needing to be broadcast and a theoretical number of off-chain transactions per second that can rival established, centralized payment systems. With off-chain transactions one can obtain more privacy than with on-chain transaction: individual payments through LN should be able to stay private. However, recent research suggests that this is not the case. Several works have shown that channel balances can be tracked with the Balance Discovery Attack (BDA). Monitoring the changes in channel balances over time enables an adversary to track individual payments in LN. Since payments can travel through channels outside the scope of influence of the payer or the payee, these BDA’s form an Interdependent Privacy (IDP) risk for the participants of the LN network. This work applies approximate differential privacy to hide payment amounts of a realistic size in LN. To our knowledge, this is the first time that strong privacy guarantees in the sense of approximate differential privacy are achieved in the setting of LN. We prove the feasibility with our prototype, compatible with LN today. Our solution is evaluated in terms of cost and utility and we show that the latter is not affected by our solution.

Pulse Oximeter App Privacy Policies During COVID-19: Scoping Assessment

BackgroundPulse oximeter apps became of interest to consumers during the COVID-19 pandemic, particularly when traditional over-the-counter pulse oximeter devices were in short supply. Yet, no study to date has examined or scoped the state of privacy policies and notices for the top-rated and most downloaded pulse oximeter apps during COVID-19.ObjectiveThe aim of this study was to examine, through a high-level qualitative assessment, the state and nature of privacy policies for the downloaded and top-rated pulse oximeter apps during the COVID-19 pandemic to (1) compare findings against comparable research involving other mobile health (mHealth) apps and (2) begin discussions on opportunities for future research or investigation.MethodsDuring August-October 2020, privacy policies were reviewed for pulse oximeter apps that had either at least 500 downloads (Google Play Store apps only) or a three out of five-star rating (Apple Store apps only). In addition to determining if the apps had an accessible privacy policy, other key privacy policy–related details that were extracted included, but were not limited to, app developer location (country); whether the app was free or required paid use/subscription; whether an ads disclosure was provided on the app’s site; the scope of personal data collected; proportionality, fundamental rights, and data protection and privacy issues; and privacy safeguards.ResultsSix pulse oximeter apps met the inclusion criteria and only 33% (n=2) of the six apps had an accessible privacy policy that was specific to the pulse oximeter app feature (vs the app developer’s website or at all). Variation was found in both the regulatory nature and data privacy protections offered by pulse oximeter apps, with notable privacy protection limitations and gaps, although each app provided at least some information about the scope of personal data collected upon installing the app.ConclusionsPulse oximeter app developers should invest in offering stronger privacy protections for their app users, and should provide more accessible and transparent privacy policies. This is a necessary first step to ensure that the data privacy of mHealth consumers is not exploited during public health emergency situations such as the COVID-19 pandemic, where over-the-counter personal health monitoring devices could be in short supply and patients and consumers may, as a result, turn to mHealth apps to fill such supply gaps. Future research considerations and recommendations are also suggested for mHealth technology and privacy researchers who are interested in examining privacy implications associated with the use of pulse oximeter apps during and after the COVID-19 pandemic.

Research on the Privacy Security of Face Recognition Technology.

To solve the problem that privacy data are easy to leak in the application of face recognition technology in apps, a method which is based on differential privacy for privacy security protection is proposed. Firstly, Bayesian GAN is conducted to obtain the training data with the same distribution as the privacy data, and the algorithm of differential privacy is conducted to train the training data to obtain these labels with privacy protection. Then, based on the proposed lightface lightweight face recognition model, the tag with noise is generated, and the gradient descent is conducted on the recovered face feature vector from the attack. Finally, through the analysis of privacy loss, an accurate privacy protection boundary is provided. From the results of experiments, it could be known that the proposed privacy security protection method can effectively protect the parameter information of the face recognition model under the face recognition technology and reduce the recognition accuracy of the image recovered by the attacker. Compared with the privacy protection methods such as DPSGD and PATE, it has strong privacy protection ability and can be applied to the privacy protection of practical APP.

FastPRS: Augmenting Fast and Hidden Query in EPR Systems via Online/Offline Puncturable Search

Electronic patient record (EPR) gains popularity in the health care system with easy availability and usually employs searchable encryption to guarantee both record confidentiality and record retrieval. However, traditional searchable encryption schemes cost a linear complexity of database and even reveals the query index. Both problems hinder EPR applications thereby balancing privacy and search efficiency becomes challenging. In this article, we design a cryptographically searchable EPR system named FastPRS which supports strong privacy and high performance. FastPRS is built on top of online/offline puncturable search (OOPS), a newly defined primitive which splits the search process into the offline phase and the online phase and embeds an index in puncturable pseudorandom sets. Leveraging a combination of linearly homomorphic encryption and oblivious transfer, the query index is punctured. OOPS, therefore, hides the query index obliviously and reduces the search complexity to a sublinear size. We also implement a FastPRS prototype system that supports more than 60 000 keywords and evaluation results show that FastPRS optimizes the search efficiency to 417 search operations per millisecond.

LDPCD: A Novel Method for Locally Differentially Private Community Detection.

As one of the cores of data analysis in large social networks, community detection has become a hot research topic in recent years. However, user's real social relationship may be at risk of privacy leakage and threatened by inference attacks because of the semitrusted server. As a result, community detection in social graphs under local differential privacy has gradually aroused the interest of industry and academia. On the one hand, the distortion of user's real data caused by existing privacy-preserving mechanisms can have a serious impact on the mining process of densely connected local graph structure, resulting in low utility of the final community division. On the other hand, private community detection requires to use the results of multiple user-server interactions to adjust user's partition, which inevitably leads to excessive allocation of privacy budget and large error of perturbed data. For these reasons, a new community detection method based on the local differential privacy model (named LDPCD) is proposed in this paper. Due to the introduction of truncated Laplace mechanism, the accuracy of user perturbation data is improved. In addition, the community divisive algorithm based on extremal optimization (EO) is also refined to reduce the number of interactions between users and the server. Thus, the total privacy overhead is reduced and strong privacy protection is guaranteed. Finally, LDPCD is applied in two commonly used real-world datasets, and its advantage is experimentally validated compared with two state-of-the-art methods.

SecEDMO: Enabling Efficient Data Mining with Strong Privacy Protection in Cloud Computing

Frequent itemsets mining and association rules mining are among the top used algorithms in the area of data mining. Secure outsourcing of data mining tasks to the third-party cloud is an effective option for data owners. However, due to the untrust cloud and the distrust between data owners, the traditional algorithms which only work over plaintext should be re-considered to take security and privacy concerns into account. For example, each data owner may not be willing to disclose their own private data to others during the cooperative data mining process. The previous solutions are either not sufficiently secure or not efficient. Therefore, we propose a <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Sec</u> ure and <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">E</u> fficient <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">D</u> ata <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">M</u> ining <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">O</u> utsourcing (SecEDMO) scheme for secure outsourcing of frequent itemsets mining and association rules mining over the joint database (i.e., database aggregated from multiple data owners) in the paradigm of cloud computing. Based on our customized lightweight symmetric homomorphic encryption algorithm and a secure comparison algorithm, SecEDMO can ensure strong privacy protection and low data mining latency simultaneously. Moreover, the well-designed virtual transaction insertion algorithm can hide the information of the original database while still preserving the cloud’s ability to perform data mining over the obfuscated data. By evaluation of a numerical experiment and theoretical comparisons, the correctness, security, and efficiency of SecEDMO are confirmed.

“O poder da voz”

Voice is a feature that individualizes and identifies us. Also, a voice is a biometric data, which means that once a voice has been processed, its owner can be identified. This allows us to state that the voice can be used as a means to authenticate the signature of an electronic contract. Therefore, biometric voice recognition techniques will make it possible to identify the owner of the voice, who will thus be giving his contractual consent. The issue could become more complicated when not only the “voiceprint” is used as a signature, but also the counterpart is a virtual voice assistant. These assistants, designed to both “answer” and “listen”, should contain strong privacy policies regarding the processing of the voices interacting with them. The purpose of this paper is to analyze the consent given both for electronic contracting by voice and for the processing of these voices by virtual assistants.

Privacy-Preserving Analytics on Decentralized Social Graphs: The Case of Eigendecomposition

Analytics over social graphs allows to extract valuable knowledge and insights for many fields like community detection, fraud detection, and interest mining. In practice, decentralized social graphs frequently arise, where the social graph is not available to a single entity and is decentralized among a large number of users, each holding only a limited local view about the whole graph. Collecting the local views for analytics of decentralized social graphs raises critical privacy concerns, as they encode private information about the social interactions among individuals. In this paper, we design, implement, and evaluate PrivGED, a new system aimed at privacy-preserving analytics over decentralized social graphs. PrivGED focuses on the support for eigendecomposition, one popular and fundamental graph analytics task producing eigenvalues/eigenvectors over the adjacency matrix of a social graph and benefits various practical applications. PrivGED is built from a delicate synergy of insights on graph analytics, lightweight cryptography, and differential privacy, allowing users to securely contribute their local views on a decentralized social graph for a cloud-based eigendecomposition analytics service while gaining strong privacy protection. Extensive experiments over real-world social graph datasets demonstrate that PrivGED achieves accuracy comparable to the plaintext domain, with practically affordable performance superior to prior art.

Differential Privacy for Deep and Federated Learning: A Survey

Differential privacy (DP) is considered a de-facto standard for protecting users' privacy in data analysis, machine, and deep learning. Existing DP-based privacy-preserving training approaches consist of adding noise to the clients' gradients before sharing them with the server. However, implementing DP on the gradient is not efficient as the privacy leakage increases by increasing the synchronization training epochs due to the composition theorem. Recently researchers were able to recover images used in the training dataset using Generative Regression Neural Network (GRNN) even when the gradient was protected by DP. In this paper, we propose two layers of privacy protection approach to overcome the limitations of the existing DP-based approaches. The first layer reduces the dimension of the training dataset based on Hensel's Lemma. We are the first to use Hensel's Lemma for reducing the dimension (i.e., compress) of a dataset. The new dimensionality reduction method allows reducing the dimension of a dataset without losing information since Hensel's Lemma guarantees uniqueness. The second layer applies DP to the compressed dataset generated by the first layer. The proposed approach overcomes the problem of privacy leakage due to composition by applying DP only once before the training; clients train their local model on the privacy-preserving dataset generated by the second layer. Experimental results show that the proposed approach ensures strong privacy protection while achieving good accuracy. The new dimensionality reduction method achieves an accuracy of 97%, with only 25 % of the original data size.

Offset-Symmetric Gaussians for Differential Privacy

The Gaussian distribution is widely used in mechanism design for differential privacy (DP). Thanks to its sub-Gaussian tail, it significantly reduces the chance of outliers when responding to queries. However, it can only provide approximate (&#x03B5;,&#x03B4;(&#x03B5;))-DP. In practice, &#x03B4;(&#x03B5;) must be much smaller than the size of the dataset, which may limit the use of the Gaussian mechanism for large datasets with strong privacy requirements. In this paper, we introduce and analyze a new distribution for use in DP that is based on the Gaussian distribution, but has improved privacy performance. The so-called offset-symmetric Gaussian tail (OSGT) distribution is obtained through using the normalized tails of two symmetric Gaussians around zero. Consequently, it can still have sub-Gaussian tail and lend itself to analytical derivations. We analytically derive the variance of the OSGT random variable and &#x03B4;(&#x03B5;) of the single-dimensional OSGT mechanism. We extend the OSGT mechanism to <i>k</i>-dimensional queries, iteratively compute its &#x03B4;_<i>k</i>(&#x03B5;), derive its R&#x00E9;nyi differential privacy, and study its composition. Numerical results show the OSGT mechanism can offer better privacy-utility performance compared to the Gaussian and Laplace mechanisms. We also derive a method for post processing the output of the OSGT mechanism to approximate the query based on the minimum mean square error (MMSE) estimation technique. The simulation results of such processing confirm the efficacy of the OSGT mechanism over the Gaussian mechanism.