Loss Of Privacy Research Articles

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.

Local Differentially Private Heavy Hitter Detection in Data Streams with Bounded Memory

Top-k frequent items detection is a fundamental task in data stream mining. Many promising solutions are proposed to improve memory efficiency while still maintaining high accuracy for detecting the Top-k items. Despite the memory efficiency concern, the users could suffer from privacy loss if participating in the task without proper protection, since their contributed local data streams may continually leak sensitive individual information. However, most existing works solely focus on addressing either the memory-efficiency problem or the privacy concerns but seldom jointly, which cannot achieve a satisfactory tradeoff between memory efficiency, privacy protection, and detection accuracy. In this paper, we present a novel framework HG-LDP to achieve accurate Top-k item detection at bounded memory expense, while providing rigorous local differential privacy (LDP) protection. Specifically, we identify two key challenges naturally arising in the task, which reveal that directly applying existing LDP techniques will lead to an inferior "accuracy-privacy-memory efficiency" tradeoff. Therefore, we instantiate three advanced schemes under the framework by designing novel LDP randomization methods, which address the hurdles caused by the large size of the item domain and by the limited space of the memory. We conduct comprehensive experiments on both synthetic and real-world datasets to show that the proposed advanced schemes achieve a superior "accuracy-privacy-memory efficiency" tradeoff, saving 2300× memory over baseline methods when the item domain size is 41,270. Our code is anonymously open-sourced via the link.

Where you go is who you are: a study on machine learning based semantic privacy attacks

Concerns about data privacy are omnipresent, given the increasing usage of digital applications and their underlying business model that includes selling user data. Location data is particularly sensitive since they allow us to infer activity patterns and interests of users, e.g., by categorizing visited locations based on nearby points of interest (POI). On top of that, machine learning methods provide new powerful tools to interpret big data. In light of these considerations, we raise the following question: What is the actual risk that realistic, machine learning based privacy attacks can obtain meaningful semantic information from raw location data, subject to inaccuracies in the data? In response, we present a systematic analysis of two attack scenarios, namely location categorization and user profiling. Experiments on the Foursquare dataset and tracking data demonstrate the potential for abuse of high-quality spatial information, leading to a significant privacy loss even with location inaccuracy of up to 200 m. With location obfuscation of more than 1 km, spatial information hardly adds any value, but a high privacy risk solely from temporal information remains. The availability of public context data such as POIs plays a key role in inference based on spatial information. Our findings point out the risks of ever-growing databases of tracking data and spatial context data, which policymakers should consider for privacy regulations, and which could guide individuals in their personal location protection measures.

Valparaíso: Touristification and displacement in a UNESCO city

ABSTRACT Tourism has become a sector that is transforming historic areas of the current city. Its impact on the quality of life of communities, local economy, access to affordable housing, urban landscape, is being analyzed from studies on gentrification and touristification. This article analyzes the case of Valparaíso (Chile) and how the declaration of its historic area as a UNESCO World Heritage site in 2003, triggered a process of touristification. Through public programmes and private investment, houses were transformed into tourist shops and hotels, shaping a tourist bubble in these residential areas. How inhabitants felt and experienced the tourist transformation of their neighborhoods was analyzed through the use of interviews. The residents reports material and symbolic changes in their living conditions, which are experienced as pressure by displacement from their neighborhood. A pressure expressed in the loss of privacy, depopulation and loss of community networks, the displacement of neighborhood shops, the increase in property taxes, and the absence of public subsidies to rehabilitate their homes. The results place pressure by displacement as one of the main mechanisms of dispossession in historic areas of Latin American cities and suggest that in cities configured from neoliberal urban policies the UNESCO designation acts as a device that strengthens this dispossession.

Numerical Composition of Differential Privacy

We give a fast algorithm to optimally compose privacy guarantees of differentially private (DP) algorithms to arbitrary accuracy. Our method is based on the notion of \emph{privacy loss random variables} to quantify the privacy loss of DP algorithms.The running time and memory needed for our algorithm to approximate the privacy curve of a DP algorithm composed with itself $k$ times is $\tilde{O}(\sqrt{k})$. This improves over the best prior method by Koskela et al. (2021) which requires $\tilde{\Omega}(k^{1.5})$ running time. We demonstrate the utility of our algorithm by accurately computing the privacy loss of DP-SGD algorithm of Abadi et al. (2016) and showing that our algorithm speeds up the privacy computations by a few orders of magnitude compared to prior work, while maintaining similar accuracy.

Differentially Private Guarantees for Analytics and Machine Learning on Graphs: A Survey of Results

We study the applications of differential privacy (DP) in the context of graph-structured data and discuss the formulations of DP applicable to the publication of graphs and their associated statistics as well as machine learning on graph-based data, including graph neural networks (GNNs). Interpreting DP guarantees in the context of graph-structured data can be challenging, as individual data points are interconnected (often non-linearly or sparsely). This connectivity complicates the computation of individual privacy loss in differentially private learning. The problem is exacerbated by an absence of a single, well-established formulation of DP in graph settings. This issue extends to the domain of GNNs, rendering private machine learning on graph-structured data a challenging task. A lack of prior systematisation work motivated us to study graph-based learning from a privacy perspective. In this work, we systematise different formulations of DP on graphs, discuss challenges and promising applications, including the GNN domain. We compare and separate works into graph analytics tasks and graph learning tasks with GNNs. We conclude our work with a discussion of open questions and potential directions for further research in this area.

Privacy-Preserving Federated Learning for Value-Added Service Model in Advanced Metering Infrastructure

Advanced metering infrastructure (AMI) is the backbone of the next generation smart city and smart grid; it not only provides near real-time two-way communication between the consumers and the energy systems, but also enables the third parties (TPs) to provide relevant value-added services to the consumers to improve user satisfaction. However, the existing services are implemented in a centralized manner, which has potential and associated security and privacy risks also increased with Internet-of-Things (IoT) devices. To better balance the quality of the services and ensure users’ privacy, a TP AMI service model based on differentially private federated learning (FL) is proposed in this article. Instead of sending the private energy data to the cloud server, the proposed service model trains the neural network models locally, and only model parameters are shared with the central server. Moreover, the identity of individuals is eliminated by adding random Gaussian noise during the secure aggregation. Furthermore, an attention-based bidirectional long short-term memory (ATT-BLSTM) neural network model is adopted to solve the long-range dependency problem of conventional neural networks. In the case study, a residential short-term load forecasting (STLF) task is implemented to evaluate the performance of the proposed model. Compared with other state-of-the-art energy service models, the proposed one can achieve similar accuracy as the typical centralized model and balances the trade-off between privacy loss and prediction accuracy flexibly.

AI-assisted Blockchain-enabled Smart and Secure E-prescription Management Framework

Traditional medical prescriptions based on physical paper-based documents are prone to manipulation, errors, and unauthorized reproduction due to their format. Addressing the limitations of the traditional prescription system, e-prescription systems have been introduced in several countries. However, e-prescription systems lead to several concerns like the risk of privacy loss, the problem of double-spending prescriptions, lack of interoperability, and single point of failure, all of which need to be addressed immediately. We propose an AI-assisted blockchain-enabled smart and secure e-prescription management framework to address these issues. Our proposed system overcomes the problems of the centralized e-prescription systems and enables efficient consent management to access prescriptions by incorporating blockchain-based smart contracts. Our work incorporates the Umbral proxy re-encryption scheme in the proposed system, avoiding the need for repeated encryption and decryption of the prescriptions when transferred between different entities in the network. In our work, we employ two different machine learning models(Random Forest classifier and LightGBM classifier) to assist the doctor in prescribing medicines. One is a drug recommendation model, which is aimed at providing drug recommendations considering the medical history of the patients and the general prescription pattern for the particular ailment of the patient. We have fine-tuned the SciBERT model for adverse drug reaction detection. The extensive experimentation and results show that the proposed e-prescription framework is secure, scalable, and interoperable. Further, the proposed machine learning models produce results higher than 95%.

Imprisonment for South Ethiopian people living with HIV presents a double health burden: lived experiences of prisoners.

Optimal adherence is crucial for ensuring both therapeutic and preventative benefits of antiretroviral therapy (ART). Sub-optimal adherence is common in prisoners and little information is available about its predisposing circumstances in resource-limited settings. We explored lived experiences of inmates living with HIV (ILWH) and experiential accounts of service providers in South Ethiopia to identify barriers to and facilitators of HIV care use in the prison context. We conducted qualitative in-depth interviewing with eleven ILWH and eleven service providers. Audio recorded interview data were transcribed verbatim in Amharic language, translated into English and coded based on emerging concepts. We employed a descriptive phenomenological approach to abstract meaning attributed to the prisoners' lived experiences in relation to HIV care use and service providers' experiential account regarding care provision as presented to our consciousness. Several concepts emerged as barriers to HIV care use amongst ILWH in South Ethiopia including: limited access to standard care, insufficient health staff support, uncooperative security system, loss of patient privacy, a lack of status disclosure due to social stigma, and food supply insufficiency. In addition to a unique opportunity offered by an imprisonment for some ILWH to refrain from health damaging behaviours, the presence of social support in the prison system facilitated care use. This study identified important structural and social contexts that can both hinder and enhance HIV care use amongst ILWH in South Ethiopia. Given the disproportionate burden of HIV in prisoners and the potential of transmission to others during and after incarceration, development of contextually-responsive strategies is required to address the barriers and to also strengthen the enablers.

Collective privacy recovery: Data-sharing coordination via decentralized artificial intelligence.

Collective privacy loss becomes a colossal problem, an emergency for personal freedoms and democracy. But, are we prepared to handle personal data as scarce resource and collectively share data under the doctrine: as little as possible, as much as necessary? We hypothesize a significant privacy recovery if a population of individuals, the data collective, coordinates to share minimum data for running online services with the required quality. Here, we show how to automate and scale-up complex collective arrangements for privacy recovery using decentralized artificial intelligence. For this, we compare for the first time attitudinal, intrinsic, rewarded, and coordinated data sharing in a rigorous living-lab experiment of high realism involving real data disclosures. Using causal inference and cluster analysis, we differentiate criteria predicting privacy and five key data-sharing behaviors. Strikingly, data-sharing coordination proves to be a win-win for all: remarkable privacy recovery for people with evident costs reduction for service providers.

CTAB-GAN+: enhancing tabular data synthesis.

The usage of synthetic data is gaining momentum in part due to the unavailability of original data due to privacy and legal considerations and in part due to its utility as an augmentation to the authentic data. Generative adversarial networks (GANs), a paragon of generative models, initially for images and subsequently for tabular data, has contributed many of the state-of-the-art synthesizers. As GANs improve, the synthesized data increasingly resemble the real data risking to leak privacy. Differential privacy (DP) provides theoretical guarantees on privacy loss but degrades data utility. Striking the best trade-off remains yet a challenging research question. In this study, we propose CTAB-GAN+ a novel conditional tabular GAN. CTAB-GAN+ improves upon state-of-the-art by (i) adding downstream losses to conditional GAN for higher utility synthetic data in both classification and regression domains; (ii) using Wasserstein loss with gradient penalty for better training convergence; (iii) introducing novel encoders targeting mixed continuous-categorical variables and variables with unbalanced or skewed data; and (iv) training with DP stochastic gradient descent to impose strict privacy guarantees. We extensively evaluate CTAB-GAN+ on statistical similarity and machine learning utility against state-of-the-art tabular GANs. The results show that CTAB-GAN+ synthesizes privacy-preserving data with at least 21.9% higher machine learning utility (i.e., F1-Score) across multiple datasets and learning tasks under given privacy budget.

Willingness to Keep or to Give Up the Social Media for a Certain Monetary Compensation

Millions of people around the world use social media every day for various reasons, whether it is sharing information, self-presentation, filling free time, or communicating with friends. The use of social networks brings with it advantages, but also risks associated with information overload, loss of privacy, as well as fraud or criminal acts. Mostly young people are connected to social networks almost constantly and find it more difficult to give up using social networks compared to older age groups. The aim of the present paper is to find out the willingness of social network users to give up social network such as Facebook and Instagram for certain monetary compensation. The necessary data were collected based on a questionnaire survey within 4 months. A total of 2 095 responses were collected. The static methods Chi-Square test of independence, logistic regression and the statistical program SAS Enterprise Guide were used to evaluate the results of the questionnaire survey. Analysis of Maximum Likelihood Estimates was tested for five predictors: Age, Gender, Income, Activity on the social network, Number of subscribers on the social network, especially for the social network Facebook and Instagram. Based on the results of the questionnaire survey, it can be concluded that Facebook users (in all age groups and regardless of gender) demand a higher average monthly financial compensation for giving up the social network compared to Instagram users. The same conclusion follows when categorizing respondents into age categories. Regarding gender, women demand lower monetary compensation than men in relation to both social networks. There was a significant difference in the Activity predictor, where active users asked for a higher amount of money for the Instagram social network and, conversely, inactive users asked for a higher average monthly payment for the Facebook social network.
 
 Received: 25 July 2023 / Accepted: 20 November 2023 / Published: 5 January 2024

Right to be Forgotten as a Legal Protection for The Victims of Electronic Sexual Violence Cases

The 2020 National Commission on Violence Against Women's Annual Records identified the rise of a new type of violence against gender in Indonesia, specifically electronic-based sexual violence (Kekerasan Seksual Berbasis Elektronik/KSBE) which is part of gender-based violence in the digital spaces or Online Gender-Based Violence (OGBV). In 2020, KSBE cases increased 300% compared to the previous year. So far, the arrangements for tackling KSBE perpetrators in Indonesia have been inadequate. Existing laws do not contain procedures that are sensitive to victims and lack a human rights and gender perspective. Currently, Indonesia has adopted a new regulation related to handling sexual violence, namely Law No. 12 of 2022 on Sexual Violence Crimes. The authors analyse laws and regulations relevant to online sexual violence and its extension (involving physical action), from the perspective of human rights and women's rights. Thus, several significant regulations apart from the Sexual Violence Crimes Law are discussed, namely the Criminal Code, the Information and Transaction Law, the Anti-Pornography Law, and the Child Protection Law. We argue that protection of KSBE victims must be made the main goal of law enforcement. The loss of the victim's privacy and the dissemination of personal data violate victims’ human rights guaranteed by the state, and therefore derivative regulations regarding the UU TPKS (Sexual Violence Law) need to be issued and existing laws harmonized with it. Moreover, additional special victim rights regulations regarding electronic-based sexual violence are necessary, especially for KSBE acts that require a quick response, such as removing sexually explicit content.

ЗАСТОСУВАННЯ ШТУЧНОГО ІНТЕЛЕКТУ В ФІНАНСОВО-ЕКОНОМІЧНОМУ АНАЛІЗІ ДІЯЛЬНОСТІ ПІДПРИЄМСТВА

The article examines the peculiarities of artificial intelligence development and the using its achievements for the future development of Ukrainian economy, in particular in accordance with the Concept of Artificial Intelligence Development in Ukraine. The major tasks of AI using in economical spare of Ukraine are: stimulating the development of entrepreneurship in the field of artificial intelligence; motivation of business to introduce artificial intelligence technologies to increase their own efficiency; development of the Roadmap to retrain people whose work can be automated in the nearest future; introduction of the state order for artificial intelligence systems, IT specialists and data researchers; stimulation of partnership between the state and business in the field of innovative projects, as well as improvement of legislation in the relevant field. Using of information technologies allows to monitor business projects, influence the results of business operations, which enables to warn possible risks and make effective decisions. AI is a modern tool capable to increase business efficiency through rapid analysis of available data, that lets choose the best development strategy. The main advantages of using AI include: forecasting financial and market-commercial risks; optimization of development strategies; creation the system for monitoring business results in real time; automation of accounting and financial and economic analysis; analysis of market and economic trends. The key problems in the economic sector that affect the development of AI in Ukraine are separated, special attention is paid to the advantages and risks of using artificial intelligence technologies in business activities, among them low-quality data for analysis, opacity of the obtained results, as well as the risk of privacy loss. In order to minimize the possible risks connected with the loss of confidential data, it is proposed to use of software that would allow, taking into account the industry specifics of each enterprise's activity, to maintain, monitor and analyze significant data. That would allow to monitor the effectiveness of business processes in the real time mode and make efficient management decisions without fear of leakage of important confidential information.

Big Data, Digitization and New Surveillance Practices: New Decision-Making Culture and Panoptic Classification

Big data brings along a new decision-making culture through data mining aided decision-making mechanisms. One of the biggest sources of this mechanism is meta-data and that creates a suitable condition for mass-monitoring surveillance. The belief in these systems creates a discussion about usage of big data as a social classification tool, big data hubris, and big data divide. The aim of this study is to discuss the possible destructive results of big data especially as a social classification tool and new surveillance practices. In that point, this study focuses on the idea that big data paves the way to the new surveillance or dragnet surveillance practices and the possibility of social classification, and these might construct a world of increasing disadvantage. Through this panoptic sort, it comes with the problem of a serious power imbalance, simplification and decontextualization. It might result in the danger of loss of privacy and erosion of long-term privacy norms, a routine classification, manipulation of the future behaviors, antidemocratic control system, panoptic missort, data-antisubordination, abuse of civil rights and security problems.

AI-Enabled Medical Education: Threads of Change, Promising Futures, and Risky Realities Across Four Potential Future Worlds.

The rapid trajectory of artificial intelligence (AI) development and advancement is quickly outpacing society's ability to determine its future role. As AI continues to transform various aspects of our lives, one critical question arises for medical education: what will be the nature of education, teaching, and learning in a future world where the acquisition, retention, and application of knowledge in the traditional sense are fundamentally altered by AI? The purpose of this perspective is to plan for the intersection of health care and medical education in the future. We used GPT-4 and scenario-based strategic planning techniques to craft 4 hypothetical future worlds influenced by AI's integration into health care and medical education. This method, used by organizations such as Shell and the Accreditation Council for Graduate Medical Education, assesses readiness for alternative futures and effectively manages uncertainty, risk, and opportunity. The detailed scenarios provide insights into potential environments the medical profession may face and lay the foundation for hypothesis generation and idea-building regarding responsible AI implementation. The following 4 worlds were created using OpenAI's GPT model: AI Harmony, AI conflict, The world of Ecological Balance, and Existential Risk. Risks include disinformation and misinformation, loss of privacy, widening inequity, erosion of human autonomy, and ethical dilemmas. Benefits involve improved efficiency, personalized interventions, enhanced collaboration, early detection, and accelerated research. To ensure responsible AI use, the authors suggest focusing on 3 key areas: developing a robust ethical framework, fostering interdisciplinary collaboration, and investing in education and training. A strong ethical framework emphasizes patient safety, privacy, and autonomy while promoting equity and inclusivity. Interdisciplinary collaboration encourages cooperation among various experts in developing and implementing AI technologies, ensuring that they address the complex needs and challenges in health care and medical education. Investing in education and training prepares professionals and trainees with necessary skills and knowledge to effectively use and critically evaluate AI technologies. The integration of AI in health care and medical education presents a critical juncture between transformative advancements and significant risks. By working together to address both immediate and long-term risks and consequences, we can ensure that AI integration leads to a more equitable, sustainable, and prosperous future for both health care and medical education. As we engage with AI technologies, our collective actions will ultimately determine the state of the future of health care and medical education to harness AI's power while ensuring the safety and well-being of humanity.

CONECTAR: A MINDSET TO INCREASE HISPANIC ENGAGEMENT IN DIGITAL TECHNOLOGY RESEARCH

Abstract In-home digital technology affords the opportunity for more precise and ecologically valid assessments of daily activity engagement. However, concerns about loss of privacy can hinder recruitment, especially within communities already underrepresented in research. Literature on effective, culturally tailored recruitment strategies is limited. Herein, we describe strategies employed for an in-home digital technology study with 100% Mexican American representation. Hispanic participants were recruited from the greater San Antonio area. Direct recruitment strategies conducted in both English and Spanish included community outreach events, a participant repository, cross-enrollment with other research studies, and by word of mouth. Mass recruitment strategies included flyer dissemination, newsletters, and digital announcements through community organizations (i.e., Salud America!) and national (i.e., Research Match) and institutional websites. Between September 2022 and February 2023, 31 participants were enrolled out of 154 individuals who were contacted. Of these, 29% were recruited from community events, 22.6% from word of mouth, 19.4% from cross-enrollment, 12.9% from the research repositories, and 16.1% from mass recruitment strategies. The most common reasons for non-enrollment included lack of interest (n=49), screen failure (n=29), and privacy concerns (n=12).Direct recruitment efforts achieved higher rates of enrollment, while mass media campaigns yielded lower responses and resulted in more unsuccessful contact attempts. Understanding the relationship between social networks and research beliefs may be particularly relevant given the Hispanic culture’s collectivist and family-oriented nature. Hence, social network theories, combined with a diffusion of innovation framework, may be advantageous for increasing trust and may consequently improve representation of diverse groups in research.

DProvDB: Differentially Private Query Processing with Multi-Analyst Provenance

Recent years have witnessed the adoption of differential privacy (DP) in practical database systems like PINQ, FLEX, and PrivateSQL. Such systems allow data analysts to query sensitive data while providing a rigorous and provable privacy guarantee. However, the existing design of these systems does not distinguish data analysts of different privilege levels or trust levels. This design can have an unfair apportion of the privacy budget among the data analyst if treating them as a single entity, or waste the privacy budget if considering them as non-colluding parties and answering their queries independently. In this paper, we propose DProvDB, a fine-grained privacy provenance framework for the multi-analyst scenario that tracks the privacy loss to each single data analyst. Under this framework, when given a fixed privacy budget, we build algorithms that maximize the number of queries that could be answered accurately and apportion the privacy budget according to the privilege levels of the data analysts.

Price and variety in the Salop model

AbstractUsing a Salop circle model, this research analyzes the welfare implications of firm/product entry with information provision by consumers. While firms use consumer information to target sales efforts, consumers face privacy trade‐offs when providing their personal information. We show that (i) price and profit first increase, then decrease with more varieties; (ii) consumer welfare, affected by price, sales effort, privacy loss, and matching effects, first decreases, then increases with firm entry; (iii) equilibrium information is socially optimal given the number of varieties; and (iv) if the variable cost of providing sales assistance is low (high), free entry leads to too much (few) varieties and too little (more) information, from a social welfare standpoint.

Sino-futurism and alternative imaginaries of Digital China

This paper investigates the role of digital technologies in transforming China's self-image. It focuses on the use of extended reality (XR) in ceremonial events and art exhibitions. The paper offers two case studies. The first study, Beijing 8-Minute Show (2018), a multimedia performance at the closing ceremony of the Pyeongchang Winter Olympic Games, set the scene for what was called the Science and Technology Winter Olympics in 2022. The second study, Blueprints (2020), was a multimedia exhibition in the UK by the artist Cao Fei, which drew attention to a future of increased alienation, loss of privacy and digital surveillance. In the paper, the framework of a ‘techno-cultural imaginary’ shows how China's self-image is increasingly tied to modernisation. The paper demonstrates how the ‘two cultures’, science and the arts, have converged in policy thinking. In this reset, China's so-called cultural confidence is re-energised by digital platforms, echoing the description of Digital China. Sino-futurism, originally used in relation to Chinese sci-fi literature, provides a stepping-off point to imagine the future, which is alternatively characterised as techno-utopian (within China) and dystopian (in the West). Drawing on the qualitative analysis of publicly available interviews, media reports, online comments and close reading of the art content, the paper argues that XR allows the government to present Digital China as a positive blueprint for human progress. Meanwhile, XR is capable of generating critical stories about China, which contradict the message the government seeks to cultivate with its public diplomacy and propaganda campaigns.