Privacy Cost Research Articles

Цифровая трансформация промышленных предприятий и экономический рост

The article examines the features of the digital transformation of industrial enterprises in the Russian Federation at the present stage of their development. The empirical part of the study is focused on examining the growth rates of large enterprises representing various sectors of the country’s national economy (EvrazHolding, Sukhoi Company, and Gazprom Neft) and their transformation as part of the digitalization of the domestic economic system. Their benefits from the introduction of digital technologies into production processes, including ecosystem purposes, which simplify the process of selling products without loss of quality, ensuring a high level of cooperation between platform participants and high market transparency, are considered. The indicators of profitability and business activity of the above companies (2017–2021) were studied as part of the digital transformation of these enterprises. The problems of introducing digital technologies are identified: high risks of data leakage when introducing online technologies; high costs of introducing digital technologies; changes in the labor market; customer privacy costs; errors in device operation, calculations, and software. In accordance with the above-identified problems, ways to resolve them have been developed based on a strategic approach to the digital transformation of the financial and economic activities of enterprises: preparation of production in accordance with the requirements of the digital passport of the Ministry of Industry and Trade; drawing up an investment schedule; preparation of a business process reengineering schedule; determination of strategic directions for R&D to improve the quality of products and increase labor productivity through digital technologies.

Privacy Cost Optimization of Smart Meters Using URLLC and Demand Side Energy Trading

Privacy Cost Optimization of Smart Meters Using URLLC and Demand Side Energy Trading

Adopting blockchain-based traceability in the fruit supply chain in a developing economy: facilitators and barriers

PurposeBlockchain-based traceability technology (BTT) is an emerging digital technology that claims to have the potential to fulfil the demand for traceability to safeguard fruit safety. Drawing on the technological, organisational and environmental (TOE) framework, this study aims to investigate the perceived facilitators and barriers that influence the behavioural intentions of multiple stakeholders in the Vietnamese fruit supply chain (i.e., farmers, trading enterprises and consumers) to adopt BTT.Design/methodology/approachThis study utilised a qualitative approach of semi-structured in-depth interviews with 60 stakeholders in the Vietnamese fruit supply chain to achieve the research objectives. NVivo 12 was employed to analyse the collected data using content and thematic analysis.FindingsThe findings identify several perceived facilitators that motivate BTT adoption. These include trust, transparency, business performance, the formation of alliances, consumer awareness of food safety and ethical agricultural practices, fruit branding and the pivotal role of farmers' cooperatives. Meanwhile, the perceived barriers to BTT adoption include a lack of digital literacy amongst the stakeholders, poor organisational culture, the high cost of traceability-enabled products and data privacy and security governance.Practical implicationsThis study suggests that technology awareness and perceived facilitators and barriers should be incorporated into the design and deployment of blockchain-based traceability technology in the agri-food supply chain in developing countries.Originality/valueTo the best of the authors' knowledge, this study is the first qualitative paper that attempts to fill the research gap of understanding the perceived facilitators and barriers that influence the intentions of multiple stakeholders in the fruit supply chain to adopt BTT in the context of a developing economy.

Membership inference attack on differentially private block coordinate descent.

The extraordinary success of deep learning is made possible due to the availability of crowd-sourced large-scale training datasets. Mostly, these datasets contain personal and confidential information, thus, have great potential of being misused, raising privacy concerns. Consequently, privacy-preserving deep learning has become a primary research interest nowadays. One of the prominent approaches adopted to prevent the leakage of sensitive information about the training data is by implementing differential privacy during training for their differentially private training, which aims to preserve the privacy of deep learning models. Though these models are claimed to be a safeguard against privacy attacks targeting sensitive information, however, least amount of work is found in the literature to practically evaluate their capability by performing a sophisticated attack model on them. Recently, DP-BCD is proposed as an alternative to state-of-the-art DP-SGD, to preserve the privacy of deep-learning models, having low privacy cost and fast convergence speed with highly accurate prediction results. To check its practical capability, in this article, we analytically evaluate the impact of a sophisticated privacy attack called the membership inference attack against it in both black box as well as white box settings. More precisely, we inspect how much information can be inferred from a differentially private deep model's training data. We evaluate our experiments on benchmark datasets using AUC, attacker advantage, precision, recall, and F1-score performance metrics. The experimental results exhibit that DP-BCD keeps its promise to preserve privacy against strong adversaries while providing acceptable model utility compared to state-of-the-art techniques.

Privacy-Preserving Data Synthesis via Differentially Private Normalizing Flows with Application to Electronic Health Records Data

Medical data often contain sensitive personal information about individuals, posing significant limitations to it being shared or released for downstream learning and inferential tasks. We use normalizing flows (NF), a family of deep generative models, to estimate the probability density of a dataset with differential privacy (DP) guarantees, from which privacy-preserving synthetic data are generated and released. We apply the technique to an electronic health records dataset containing patients with pulmonary hypertension. We assess the learning and inferential utility of synthetic data by comparing the accuracy of hypertension predictions and the variational posterior distribution of the parameters in a physics-based model. The results suggest that synthetic data generated via NF with DP can yield good utility at a reasonable privacy cost. Our study provides evidence and adds to the growing literature on the feasibility of generating synthetic medical data for sharing or obtaining inferences from medical data using deep generate models with formal privacy guarantees.

Distributed Clustering for Secure Multitask Estimation Based on Adaptive Task Switching

Consider adversarial multi-task networks with a high cost of sharing genuine information, where agents feed others unreliable information according to a time-varying probability caused by a high privacy cost. This makes agents in the network tend to be selfish, and it greatly increases the difficulty of effective information sharing and distributed multi-task estimation without the prior cluster information. To address these problems, we propose an adaptive task-switching strategy that is governed by our formulated information sharing model, which cooperates with the proposed DATSLCS-AM algorithm. This distributed algorithm guides agents to cooperate reasonably following reputation updating and pseudo-clustering, which can improve the effectiveness of information sharing and also increase the security of estimation over adversarial multi-task networks. The theoretical analysis of the adaptive pseudo-clustering threshold is provided. Finally, extensive simulations validate the superior performance of our distributed algorithm.

Game relations between platform enterprises and participants of digital interaction-sharing platforms in the maritime logistics industry

Digital platforms simplify the complexities of day-to-day transactions in the maritime logistics industry and facilitate participants to seamlessly connect with each other. However, platform enterprises find it difficult to involve participants in their platforms. It is also challenging for these enterprises to share data out of consideration for participants’ privacy concerns, rights, and interests. To promote data sharing, this study intends to understand the strategies of both platform enterprises and participants based on a dynamic game model to provide policy implications to platform enterprises and the government. The main findings are as follows. First, when platform enterprises move first, they should subsidize key participants with high privacy costs to compensate for the negative externality of data. Second, in the early stages of building a platform, when supporting platform enterprises move first, they must subsidize both key participants with low privacy costs and general participants. Third, governments must introduce regulations and laws to govern the delimitation of data property rights.

FEDAUX: Leveraging Unlabeled Auxiliary Data in Federated Learning.

Federated distillation (FD) is a popular novel algorithmic paradigm for Federated learning (FL), which achieves training performance competitive to prior parameter averaging-based methods, while additionally allowing the clients to train different model architectures, by distilling the client predictions on an unlabeled auxiliary set of data into a student model. In this work, we propose FEDAUX, an extension to FD, which, under the same set of assumptions, drastically improves the performance by deriving maximum utility from the unlabeled auxiliary data. FEDAUX modifies the FD training procedure in two ways: First, unsupervised pre-training on the auxiliary data is performed to find a suitable model initialization for the distributed training. Second, ε, δ)-differentially private certainty scoring is used to weight the ensemble predictions on the auxiliary data according to the certainty of each client model. Experiments on large-scale convolutional neural networks (CNNs) and transformer models demonstrate that our proposed method achieves remarkable performance improvements over state-of-the-art FL methods, without adding appreciable computation, communication, or privacy cost. For instance, when training ResNet8 on non-independent identically distributed (i.i.d.) subsets of CIFAR10, FEDAUX raises the maximum achieved validation accuracy from 30.4% to 78.1%, further closing the gap to centralized training performance. Code is available at https://github.com/fedl-repo/fedaux.

Smart services and privacy, can we have both?

Smart services’ data-intensive nature means their convenience for users comes with privacy concerns. This poses challenges for the supply of smart services. This paper analyses the impact of user-differentiated privacy costs on smart service supply strategies. It also discusses the effects of privacy protections and data utilisation capabilities on the improvement of smart service levels. We find that providers tend to segment the market based on user privacy sensitivity to capture the maximum consumer surplus. Strict privacy protection requirements may cause providers to sacrifice service quality. Privacy protection is conducive to improving the level of smart services only when users fully recognise these services’ value. Interestingly, our study finds improvements in service levels by providers are driven by two factors. On the one hand, providers choose to improve their service levels to obtain more data and create more profits. On the other hand, to reduce the opportunity costs of early investments to develop services, providers need to improve their capabilities to utilise the value of data, ultimately leading to improved service levels. This study brings a new perspective to the supply of smart services and offers suggestions for how to resolve the contradiction between smart service use and privacy preservation.

Robin Hood to the Rescue: Sustainable Revenue‐Allocation Schemes for Data Cooperatives

The promise of consumer data along with advances in information technology has spurred innovation not only in the way firms conduct their business operations but also in the manner in which data are collected. A prominent institutional structure that has recently emerged is a data cooperative—an organization that collects data from its members, and processes and monetizes the pooled data. A characteristic of consumer data is the externality it generates: Data shared by an individual reveal information about other similar individuals; thus, the marginal value of pooled data increases in both the quantity and quality of the data. A key challenge faced by a data cooperative is the design of a revenue‐allocation scheme for sharing revenue with its members. An effective scheme generates a beneficial cycle: It incentivizes members to share high‐quality data, which in turn results in high‐quality pooled data—this increases the attractiveness of the data for buyers and hence the cooperative's revenue, ultimately resulting in improved compensation for the members. While the cooperative naturally wishes to maximize its total surplus, two other important desirable properties of an allocation scheme are individual rationality and coalitional stability. We first examine a natural proportional allocation scheme—which pays members based on their individual contribution—and show that it simultaneously achieves individual rationality, the first‐best outcome, and coalitional stability, when members' privacy costs are homogeneous. Under heterogeneity in privacy costs, we analyze a novel hybrid allocation scheme and show that it achieves both individual rationality and the first‐best outcome, but may not satisfy coalitional stability. Finally, our RobinHood allocation scheme—which uses a fraction of the revenue to ensure coalitional stability and allocates the remaining based on the hybrid scheme—achieves all the desirable properties.

An analysis of strategies for adopting blockchain in green supply chains under corporate social responsibility.

In recent years, green products have received increasing attention by firms and consumers, but consumers often have uncertainty about the green degree of products. Many firms adopt blockchain technology to solve this problem, but blockchain adoption will bring privacy concerns to consumers. Meanwhile, corporate social responsibility has become one of the hot topics concerned by firms. Based on this, a Stackelberg game model dominated by the manufacturer is constructed to study the strategies of blockchain adoption for green supply chains under corporate social responsibility. The interaction of corporate social responsibility awareness and blockchain adoption in different models is verified through the calculation and simulation analysis of the optimal decision of the supply chain members. The research results show that no matter whether supply chain members have corporate social responsibility awareness, the manufacturer should adopt blockchain technology only when the privacy cost of consumers is low. After blockchain technology is adopted, the profit and utility of retailer, the manufacturer's utility, consumer surplus, and social welfare will be higher. However, when the manufacturer has corporate social responsibility awareness, the manufacturer's profit may be damaged by blockchain adoption. In addition, when supply chain members have corporate social responsibility awareness, the manufacturer is more likely to adopt blockchain technology. And with the enhancement of corporate social responsibility awareness, blockchain technology is more likely to be adopted. It provides a reference for the blockchain adoption strategies of the green supply chain under corporate social responsibility.

Optimal Nonlinear Pricing with Data-Sensitive Consumers

We study monopolistic screening when some consumers are data sensitive and incur a privacy cost if their purchase reveals information to the monopolist. The monopolist discriminates between data-sensitive and classical consumers using privacy mechanisms that consist of a direct mechanism and a privacy option. A privacy mechanism is optimal for large privacy costs and leaves classical consumers better off than data-sensitive consumers with the same valuation. When privacy preferences become public information, data-sensitive consumers and the monopolist gain, whereas classical consumers lose. Our results are relevant for policies targeting consumers’ data awareness, such as the European General Data Protection Regulation. (JEL D11, D42, D82, D83, L12)

Exploring the Relationship Between Privacy and Utility in Mobile Health: Algorithm Development and Validation via Simulations of Federated Learning, Differential Privacy, and External Attacks.

Although evidence supporting the feasibility of large-scale mobile health (mHealth) systems continues to grow, privacy protection remains an important implementation challenge. The potential scale of publicly available mHealth applications and the sensitive nature of the data involved will inevitably attract unwanted attention from adversarial actors seeking to compromise user privacy. Although privacy-preserving technologies such as federated learning (FL) and differential privacy (DP) offer strong theoretical guarantees, it is not clear how such technologies actually perform under real-world conditions. Using data from the University of Michigan Intern Health Study (IHS), we assessed the privacy protection capabilities of FL and DP against the trade-offs in the associated model's accuracy and training time. Using a simulated external attack on a target mHealth system, we aimed to measure the effectiveness of such an attack under various levels of privacy protection on the target system and measure the costs to the target system's performance associated with the chosen levels of privacy protection. A neural network classifier that attempts to predict IHS participant daily mood ecological momentary assessment score from sensor data served as our target system. An external attacker attempted to identify participants whose average mood ecological momentary assessment score is lower than the global average. The attack followed techniques in the literature, given the relevant assumptions about the abilities of the attacker. For measuring attack effectiveness, we collected attack success metrics (area under the curve [AUC], positive predictive value, and sensitivity), and for measuring privacy costs, we calculated the target model training time and measured the model utility metrics. Both sets of metrics are reported under varying degrees of privacy protection on the target. We found that FL alone does not provide adequate protection against the privacy attack proposed above, where the attacker's AUC in determining which participants exhibit lower than average mood is over 0.90 in the worst-case scenario. However, under the highest level of DP tested in this study, the attacker's AUC fell to approximately 0.59 with only a 10% point decrease in the target's R2 and a 43% increase in model training time. Attack positive predictive value and sensitivity followed similar trends. Finally, we showed that participants in the IHS most likely to require strong privacy protection are also most at risk from this particular privacy attack and subsequently stand to benefit the most from these privacy-preserving technologies. Our results demonstrated both the necessity of proactive privacy protection research and the feasibility of the current FL and DP methods implemented in a real mHealth scenario. Our simulation methods characterized the privacy-utility trade-off in our mHealth setup using highly interpretable metrics, providing a framework for future research into privacy-preserving technologies in data-driven health and medical applications.

The cost of privacy on blockchain: A study on sealed-bid auctions

In order to preserve privacy in a blockchain ecosystem, the main objective is to keep a transaction's data private, such as the sender, the receiver, and the amount transferred. The current work studies the cryptographic tools commonly used to achieve this type of privacy, primarily focusing on the Ethereum blockchain. Such tools usually require many computational and storage resources, leading to additional fees. An anonymous auction protocol was developed as a case study to explore these costs, where hiding the identity and the amount of the bids utilizes a variety of cryptographic primitives. The proposed implementation was compared against three sealed-bid auction protocols, which utilize similar cryptographic tools for preserving privacy throughout the auction process. The results show that providing an additional level of anonymity, such as hiding someone's identity, can increase the gas cost significantly, up to 2.5 times, depending on the choice of the cryptographic tools, which determine the usage of the blockchain's storage and computational resources. By adjusting the level of decentralization on the application level by moving some operations off-chain and maintaining the role of the auctioneer, we show that we can maintain anonymity while reducing the gas cost by 40%.

Digital Privacy: GDPR and Its Lessons for Australia

AbstractAustralia's Privacy Act 1988 is under review with a view to bringing Australia's privacy laws into the digital era, more in line with the European Union's General Data Protection Regulation (GDPR). This article discusses how the GDPR can be refined and standardised to be more effective in protecting privacy in the digital era while not adversely affecting the digital economy that relies heavily on data. We argue that an ideal data policy should be informative and transparent about potential privacy costs while giving consumers a menu of opt‐in choices into which they can self‐select.

Differentially Private LQ Control

As multi-agent systems proliferate and more user data, new approaches are needed to protect sensitive data while still enabling system operation. To address this need, this article presents a private multiagent LQ control framework. Agents’ state trajectories can be sensitive, and we therefore protect them using differential privacy. We quantify the impact of privacy along three dimensions: the amount of information shared under privacy, the control-theoretic cost of privacy, and the tradeoffs between privacy and performance. These analyses are done in conventional control-theoretic terms, which we use to develop guidelines for calibrating privacy as a function of system parameters. Numerical results indicate that system performance remains within desirable ranges, even under strict privacy requirements.

The Cost of Privacy. The Impact of the California Consumer Protection Act on Mortgage Markets.

The Cost of Privacy. The Impact of the California Consumer Protection Act on Mortgage Markets.

The Review of Technology in Monitoring the Heart Health of the Elderly

This paper reviews the technology to assist the elderly in monitoring their heart health. Post-COVID19 pandemic, heart disease is back as the main cause of death among the global citizens, especially the elderly. Technology plays a big role in ensuring that elderly can monitor their heart health more easily and effectively. Among the technologies that can be used are wearable devices, mobile ECG devices, mobile applications, pacemaker implants, and IoT sensors. Some of the potentials of heart health monitoring technology include helping to monitor health independently, in real time, remotely, schedule medication intake, schedule heart rate recording, schedule light physical activity, and communicate directly with specialist doctors to discuss the status and level of their heart health. However, the potential of such technology has several challenges such as usability, affordances, cost, and trust in data confidentiality and privacy.

Differentially private block coordinate descent

Deep learning models have revolutionized AI tasks by producing accurate predictions. These models′ success largely depends on precise training using large-scale datasets, mostly crowdsourced from the target population. The training datasets may contain sensitive personal information, and the model parameters can encode this information on internal wirings of hidden layers, thus bearing the risk of a privacy breach. The modern trend of sharing trained models has grown the privacy breach risk manifold. The performance of existing privacy-preserving deep learning models trying to address this issue is unsatisfactory. Consequently, only a marginal proportion of these privacy-preserved models have been adopted in the industry. Therefore, we developed the first differentially private version of the block coordinate descent (BCD) algorithm. Our proposed mechanism considerably cuts the privacy cost by injecting an appropriate amount of noise in block variables. It achieves high accuracy comparable to its non-private equivalents. It improves the convergence speed and provides a provable privacy guarantee by performing privacy accounting using advanced composition and moments accountant methods. We empirically evaluate the robustness of our proposed mechanism on benchmark datasets. The results demonstrate the competitive performance in terms of both privacy cost reduction and speedy convergence against the state-of-the-art differential privacy-based mechanisms.

Factors Influencing Use of Fitness Apps by Adults under Influence of COVID-19.

During the coronavirus disease 2019 (COVID-19) pandemic, many countries imposed restrictions and quarantines on the population, which led to a decrease in people's physical activity (PA) and severely damaged their mental health. As a result, people engaged in fitness activities with the help of fitness apps, which improved their resistance to the virus and reduced the occurrence of psychological problems, such as anxiety and depression. However, the churn rate of fitness apps is high. As such, our purpose in this study was to analyze the factors that influence the use of fitness apps by adults aged 18-65 years in the context of COVID-19, with the aim of contributing to the analysis of mobile fitness user behavior and related product design practices. We constructed a decision target program model using the analytic hierarchy process (AHP), and we analyzed and inductively screened 11 evaluation indicators, which we combined with an indicator design questionnaire. We distributed 420 questionnaires; of the respondents, 347 knew about or used fitness apps. Among these 347, we recovered 310 valid questionnaires after removing invalid questionnaires with a short completion time, for an effective questionnaire recovery rate of 89.33%. We used the AHP and entropy method to calculate and evaluate the weight coefficient of each influencing factor and to determine an influencing factor index. Our conclusions were as follows: first, the effect of perceived usefulness on the use of fitness apps by the study groups was the most notable. Second, personal motivation and perceived ease of use considerably influenced the adult group's willingness to use fitness apps. Finally, the perceived cost had relatively little effect on the use of fitness apps by adults, and the study group was much more concerned with the privacy cost than the expense cost.