Privacy Enhancing Research Articles

Privacy enhancing and generalizable deep learning with synthetic data for mediastinal neoplasm diagnosis

The success of deep learning (DL) relies heavily on training data from which DL models encapsulate information. Consequently, the development and deployment of DL models expose data to potential privacy breaches, which are particularly critical in data-sensitive contexts like medicine. We propose a new technique named DiffGuard that generates realistic and diverse synthetic medical images with annotations, even indistinguishable for experts, to replace real data for DL model training, which cuts off their direct connection and enhances privacy safety. We demonstrate that DiffGuard enhances privacy safety with much less data leakage and better resistance against privacy attacks on data and model. It also improves the accuracy and generalizability of DL models for segmentation and classification of mediastinal neoplasms in multi-center evaluation. We expect that our solution would enlighten the road to privacy-preserving DL for precision medicine, promote data and model sharing, and inspire more innovation on artificial-intelligence-generated-content technologies for medicine.

Private compression for intermediate feature in IoT-supported mobile cloud inference

In the emerging Internet of Things (IoT) paradigm, mobile cloud inference serves as an efficient application framework that relieves the computation and storage burden on resource-constrained mobile devices by offloading the workload to cloud servers. However, mobile cloud inference encounters computation, communication, and privacy challenges to ensure efficient system inference and protect the privacy of mobile users’ collected information. To address the deployment of deep neural networks (DNN) with large capacity, we propose splitting computing (SC) where the entire model is divided into two parts, to be executed on mobile and cloud ends respectively. However, the transmission of intermediate data poses a bottleneck to system performance. This paper initially demonstrates the privacy issue arising from the machine analysis-oriented intermediate feature. We conduct a preliminary experiment to intuitively reveal the latent potential for enhancing the privacy-preserving ability of the initial feature. Motivated by this, we propose a framework for privacy-preserving intermediate feature compression, which addresses the limitations in both compression and privacy that arise in the original extracted feature data. Specifically, we propose a method that jointly enhances privacy and encoding efficiency, achieved through the collaboration of the encoding feature privacy enhancement module and the privacy feature ordering enhancement module. Additionally, we develop a gradient-reversal optimization strategy based on information theory to ensure the utmost concealment of core privacy information throughout the entire codec process. We evaluate the proposed method on two DNN models using two datasets, demonstrating its ability to achieve superior analysis accuracy and higher privacy preservation than HEVC. Furthermore, we provide an application case of a wireless sensor network to validate the effectiveness of the proposed method in a real-world scenario.

Privacy Preserving Technology using Federated Learning and Blockchain in protecting against Adversarial Attacks for Retinal Imaging

PurposeCollaboration provides valuable data for robust artificial intelligence (AI) model development. Federated learning (FL) is a privacy enhancing technology that allows collaboration while respecting privacy via the development of models without raw data transfer. However state-of-the-art FL models still face challenges in non-independent and identically distributed (non-i.i.d.) health care settings and remain susceptible to privacy breaches. We propose a novel FL framework coupled with blockchain technology to address these challenges. DesignRetrospective multicohort study Subjects and Methods27,145 images from Singapore, China and Taiwan were used to design a novel FL aggregation method for the detection of myopic macular degeneration (MMD) from fundus photographs and macular disease from optical coherence tomography (OCT) scans in feature distribution skew and label distribution imbalance scenarios. We further performed adversarial attacks (label flipping and clean label). As proof of concept, blockchain was incorporated into FL to demonstrate secure transfer of model updates across collaborating sites. Main Outcome MeasuresWe evaluated our FL model performance in MMD and OCT classification and compared our model against state-of the-art FL and centralized models. ResultsOur FL model showed robust performance with areas under the receiving operating characteristic curves (AUC) of 0.868±0.009 for MMD detection and 0.970±0.012 for OCT macular disease classification. In label flipping attack, our FL model had an AUC of 0.861±0.019, similar to the centralized model (AUC 0.856± 0.015) and higher than other FL models (AUC 0.578–0.819) In clean label attack, our FL model had an AUC of 0.878±0.006 which was comparable to the centralized model (AUC 0.878±0.001) and superior to other state-of-the-art FL models with AUC of 0.529-0.838. Simulation showed that the additional time with blockchain in one global epoch was around 5 seconds. The addition of blockchain to the FL framework was feasible with a minimal impact on model development time. ConclusionsOur proposed FL algorithm overcomes the shortcoming of the traditional FL in non i.i.d. situations and remains robust to against adversarial attacks. The integration of blockchain adds further security during the transfer of model updates. Blockchain-enabled FL can be a trusted platform for collaborative health AI research.

A New Framework for Enhancing VANETs through Layer 2 DLT Architectures with Multiparty Threshold Key Management and PETs

This work proposes a new architectural approach to enhance the security, privacy, and scalability of VANETs through threshold key management and Privacy Enhancing Technologies (PETs), such as homomorphic encryption and secure multiparty computation, integrated with Decentralized Ledger Technologies (DLTs). These advanced mechanisms are employed to eliminate centralization and protect the privacy of transferred and processed information in VANETs, thereby addressing privacy concerns. We begin by discussing the weaknesses of existing VANET architectures concerning trust, privacy, and scalability and then introduce a new architectural framework that shifts from centralized to decentralized approaches. This transition applies a decentralized ledger mechanism to ensure correctness, reliability, accuracy, and security against various known attacks. The use of Layer 2 DLTs in our framework enhances key management, trust distribution, and data privacy, offering cost and speed advantages over Layer 1 DLTs, thereby enabling secure vehicle-to-everything (V2X) communication. The proposed framework is superior to other frameworks as it improves decentralized trust management, adopts more efficient PETs, and leverages Layer 2 DLT for scalability. The integration of multiparty threshold key management and homomorphic encryption also enhances data confidentiality and integrity, thus securing against various existing cryptographic attacks. Finally, we discuss potential future developments to improve the security and reliability of VANETs in the next generation of networks, including 5G networks.

Privacy‐preserving collaboration in blockchain‐enabled IoT: The synergy of modified homomorphic encryption and federated learning

SummaryThe proliferation of network devices capable of gathering, transmitting, and receiving data over the Internet has spurred the widespread adoption of Internet of Things (IoT) devices, particularly in resource‐oriented applications. Integrating blockchain, IoT, homomorphic encryption, and federated learning requires a balance between computational requirements and real‐time performance. Secure key management is crucial to maintain data privacy and integrity. Compliance with privacy regulations requires careful implementation of privacy‐preserving mechanisms in blockchain‐enabled IoT environments, which can be subjected to various attacks. Addressing these challenges requires interdisciplinary expertise, research, and innovation to develop more efficient and effective privacy‐preserving techniques tailored to the unique characteristics of such environments. This research introduces the Modified Homomorphic Encryption Federated‐based Adaptive Hybrid Dandelion Search (MHEF‐AHDS) algorithm as an effective framework to enhance security in blockchain‐enabled IoT systems. The amalgamation of Modified Homomorphic Encryption (MHE) and Federated Learning (FL) constitutes a potent alliance that addresses privacy concerns within collaborative and decentralized machine learning environments. This facilitates secure and adaptable data collaboration, effectively mitigating privacy risks associated with sensitive information. The integration of quantum machine learning into security applications presents an exciting opportunity for distinctive progress and innovation. Within this work, the Adaptive Hybrid Dandelion optimization algorithm, featuring an Initial search strategy, is employed for hyperparameter optimization thereby elevating the performances of the proposed MHEF‐AHDS method. Furthermore, the integration of smart contracts and Blockchain‐based IoT enhances the overall security of the proposed method. MHEF‐AHDS comprehensively tackles privacy, security, and scalability challenges through robust security measures and privacy enhancements. The performance evaluation of the MHEF‐AHDS method encompasses a thorough analysis based on key metrics such as throughput, latency, scalability, energy consumption, accuracy, precision, recall, and f1‐score. Comparative assessments against existing methods are conducted to gauge the effectiveness of the proposed method in addressing security, privacy, and scalability concerns.

On the evaluation of synthetic longitudinal electronic health records

BackgroundSynthetic Electronic Health Records (EHRs) are becoming increasingly popular as a privacy enhancing technology. However, for longitudinal EHRs specifically, little research has been done into how to properly evaluate synthetically generated samples. In this article, we provide a discussion on existing methods and recommendations when evaluating the quality of synthetic longitudinal EHRs.MethodsWe recommend to assess synthetic EHR quality through similarity to real EHRs in low-dimensional projections, accuracy of a classifier discriminating synthetic from real samples, performance of synthetic versus real trained algorithms in clinical tasks, and privacy risk through risk of attribute inference. For each metric we discuss strengths and weaknesses, next to showing how it can be applied on a longitudinal dataset.ResultsTo support the discussion on evaluation metrics, we apply discussed metrics on a dataset of synthetic EHRs generated from the Medical Information Mart for Intensive Care-IV (MIMIC-IV) repository.ConclusionsThe discussion on evaluation metrics provide guidance for researchers on how to use and interpret different metrics when evaluating the quality of synthetic longitudinal EHRs.

Exploring security and privacy enhancement technologies in the Internet of Things: A comprehensive review

Exploring security and privacy enhancement technologies in the Internet of Things: A comprehensive review

PrivacyProber: Assessment and Detection of Soft–Biometric Privacy–Enhancing Techniques

PrivacyProber: Assessment and Detection of Soft–Biometric Privacy–Enhancing Techniques

A novel dynamic batching energy management strategy for multi-microgrid networks with enhanced privacy

This article presents a novel energy management approach termed Dynamic Batching EMS (DBEMS) designed specifically for day-ahead scheduling within multi-microgrid (MMG) networks. The prime aim of this strategy is to curtail operational costs within the MMG network, fortify the privacy of individual microgrids, and optimize the utilization of available resources, surpassing existing methodologies like centralized, decentralized, and nested management strategies. The emphasis is chiefly on orchestrating cost-effective operations while bolstering data privacy for each microgrid within the complex MMG framework. The efficacy of the proposed strategy is evaluated across both grid-connected and islanded modes, focusing on a network comprising four microgrids. The model formulation is executed utilizing the Mixed Integer Linear Programming method. In the islanded mode, optimization is conducted for scenarios both inclusive and exclusive of Battery Energy Storage Systems (BESS). Comparative assessments are made against centralized and nested strategies, considering parameters including operational costs, privacy enhancements, load-shedding, computation time, and computational burden.

Dynamic Value-Based Subscriptions: A Novel Approach to Digital Services Leveraging FirstParty Data, Machine Learning, and Privacy Enhancement in the Post-Cookie Era

Dynamic Value-Based Subscriptions: A Novel Approach to Digital Services Leveraging FirstParty Data, Machine Learning, and Privacy Enhancement in the Post-Cookie Era

Regulatory sandboxes for trustworthy artificial intelligence – global and Latin American experiences

ABSTRACT This paper explores how regulatory sandboxes can be used to promote the development of trustworthy artificial intelligence (AI) systems. The analysis focus on the experiences of Data Protection Authorities (DPAs) who have been experimenting with sandboxes to foster the implementation of privacy by design principles in AI systems. Throughout bibliographic research, this study (i) highlights privacy-related elements on international AI regulatory frameworks; (ii) explores the concept of privacy by design (PbD) and some of its strategies, patterns and techniques, such as Privacy Enhancing Technologies (PETs); (iii) reflects upon the impact of regulatory sandboxes; and (iv) analyses how regulatory sandboxes are being discussed in the context of AI regulatory frameworks. While this study brings global perspective on regulatory sandboxes, it also does relevant analysis in the Latin American context by presenting different stages of development of initiatives in the region. One of the main findings of this paper is that literature currently lacks an analysis of how privacy by design could be used in the context of AI systems design and governance. It also suggests that regulatory sandboxes may be able to integrate PbD principles to foster the development of trustworthy AI, although further research would be needed on that matter.

High-fidelity synthetic patient data applications and privacy considerations

This paper explores the potential applications of high-fidelity synthetic patient data in the context of healthcare research, including challenges and benefits. The paper starts by defining synthetic data, types of synthetic data and approaches to generating synthetic data. It then discusses the potential applications of synthetic data in addition to as a privacy enhancing technology and current debates around whether synthetic data should be considered personal data and,therefore, should be subjected to privacy controls to minimise reidentification risks. This will be followed by a discussion of privacy preservation approaches and privacy metrics that can be applied in the context of synthetic data. The paper includes a case study based on synthetic electronic healthcare record data from the Clinical Practice Research Datalink on how privacy concerns due to reidentification have been addressed in order to make this data available for research purposes. The authors conclude that synthetic data, particularly high-fidelity synthetic patient data, has the potential to add value over and above real data for public health and that it is possible to address privacy concerns to make synthetic data available via a combination of privacy measures applied during the synthetic data generation process and post-generation reidentification risk assessments as part of data protection impact assessments.

Digital transitions in healthcare: the need for transdisciplinary research to overcome barriers of privacy enhancing technologies uptake

PurposeAdopting Privacy Enhancing Technologies (PETs) is key to accelerating digitisation of the healthcare sector while simultaneously upholding data protection rights and increasing cybersecurity. While such technologies are market-ready, the uptake of PETs in healthcare is lagging behind. In this study we explore how conflicting logics and disciplinary disparities impact PET adoption, and how transdisciplinary (TD) methods can facilitate collaboration and mutual learning to overcome these hurdles.MethodsA sequential mixed-methods case analysis is employed, focusing on a specific multidisciplinary partnership within the healthcare sector. TD methods are used to examine challenges and underlying value systems related to PET adoption.ResultsConflicting institutional logics in healthcare, driven by missions and value systems, hinder collaborative efforts. TD methods promote awareness of disciplinary disparities, fostering an appreciative attitude toward diverse viewpoints, with an active facilitator playing a crucial role. Mutual learning aids in identifying collective actions to address challenges but may lead to tensions when issues are framed differently. TD methods, however, fall short in guiding decision-making when preferences diverge.ConclusionsSustainable PET adoption in healthcare requires addressing underlying value systems, effective communication, interdisciplinary consciousness and mutual learning. Acknowledging and managing tensions, particularly in diverse data governance contexts, proves important for successful PET adoption. Our research highlights the complexity of multidisciplinary partnerships, urging additional research to guide decision-making and governmental interventions. Ultimately, understanding these dynamics is a constructive approach for improving healthcare practices and outcomes through PET adoption.

A Differentially Private Framework for the Dynamic Heterogeneous Redundant Architecture System in Cyberspace

With the development of information technology, tremendous vulnerabilities and backdoors have evolved, causing inevitable and severe security problems in cyberspace. To fix them, the endogenous safety and security (ESS) theory and one of its practices, the Dynamic Heterogeneous Redundant (DHR) architecture, are proposed. In the DHR architecture, as an instance of the multi-heterogeneous system, a decision module is designed to obtain intermediate results from heterogeneous equivalent functional executors. However, privacy-preserving is not paid attention to in the architecture, which may cause privacy breaches without compromising the ESS theory. In this paper, based on differential privacy (DP), a theoretically rigorous privacy tool, we propose a privacy-preserving DHR framework called DP-DHR. Gaussian random noise is injected into each (online) executor output in DP-DHR to guarantee DP, but it also makes the decision module unable to choose the final result because each executor output is potentially correct even if it is compromised by adversaries. To weaken this disadvantage, we propose the advanced decision strategy and the hypersphere clustering algorithm to classify the perturbed intermediate results into two categories, candidates and outliers, where the former is closer to the correct value than the latter. Finally, the DP-DHR is proven to guarantee DP, and the experimental results also show that the utility is not sacrificed for the enhancement of privacy by much (a ratio of 4–7% on average), even in the condition of some executors (less than one-half) being controlled by adversaries.

A TEE-Based Federated Privacy Protection Method: Proposal and Implementation

With the continuous enhancement of privacy protection globally, there is a problem for the traditional machine learning paradigm, which is that training data cannot be obtained from a single place. Federated learning is considered a viable technique for preserving privacy that can train deep models with decentralized data. Aiming at two-party vertical federated learning, and at common attack problems such as model inversion, gradient leakage, and data theft, we provide a formal definition of Intel SGX’s trusted computing base, remote attestation, integrity verification, and encrypted storage, and propose a general federated learning privacy enhancement algorithm in the scenario of a malicious adversary model, and we extend this method to support horizontal federated learning, secure outsourced computation, etc. Furthermore, the method is developed in a Fedlearner framework of open-sourced machine learning to achieve privacy protection of the training data and model without any modification to the existing neural network and algorithm running on the framework. The experimental results show that this scheme substantially improves on the existing schemes in terms of training efficiency, without losing model accuracy.

Mix-Zones as an Effective Privacy Enhancing Technique in Mobile and Vehicular Ad-hoc Networks

Intelligent Transportation Systems (ITS) promise significant increases in throughput and reductions in trip delay. ITS makes extensive use of Connected and Autonomous Vehicles (CAV) frequently broadcasting location, speed, and intention information. However, with such extensive communication comes the risk to privacy. Preserving privacy while still exchanging vehicle state information has been recognized as an important problem. Mix zones have emerged as a potentially effective way of protecting user privacy in ITS. CAVs are assigned pseudonyms to mask their identity; a mix zone is an area where CAVs can change their pseudonyms to resist being tracked. In order to be effective, mix zone placement must take account of traffic flows. Also, since a mix zone can degrade throughput, mix zones must be used sparingly. Determining the number and placement of mix zones is a difficult dynamic optimization problem. This paper outlines the various approaches recently taken by researchers to deal with this problem.

Enhancement of Security and Privacy of Smart Contracts in Blockchain

Smart contracts, leveraging the power of blockchain technology, have revolutionized the execution and enforcement of agreements. However, their adoption also brings forth substantial challenges in terms of security and privacy. This research paper aims to identify the recent areas of focus and provide a comprehensive perspective on blockchain applications and smart contracts, highlighting their main issues and corresponding solutions. Furthermore, it seeks to address the gaps in current research and outline future avenues of investigation. The primary objective is to assess the security and privacy concerns associated with smart contracts in blockchain and propose effective measures to enhance their robustness. By conducting a thorough analysis of vulnerabilities, attack vectors, and privacy considerations, this study offers valuable insights into the risks involved in smart contracts. It also puts forth practical solutions and best practices to mitigate these risks, ensuring a more secure and privacy-preserving environment for the deployment and execution of smart contracts.

Advanced Surveillance System with Encrypted Alerts and Privacy Enhancement

The proposed project "Advanced Surveillance System with Encrypted Alerts and Privacy Enhancement" addresses the critical need to increase public safety through a sophisticated security surveillance system. This multi-faceted system integrates core functions including real-time object detection, encrypted image capture, secure email alerts and comprehensive security measures. Different algorithms are used for face detection, which allows the system to continuously evolve and improve its object detection capabilities and adapt to real-world scenarios. The primary goal is to identify specific objects in public spaces and thus contribute to safer communities. The project further prioritizes the protection of captured and transmitted data, implementing strict security measures to protect against attempts to hack email accounts and decrypt images. This project represents a significant leap forward in security and surveillance, offering a versatile and adaptable solution designed to effectively address evolving public safety challenges.. Keywords — Encryption, Decryption, Cryptography, Object Detection, RNS, Machine learning algorithms.

A Secure Selective Image and Voice Encryption (SSIVE): Privacy Enhancing Strategy

A Secure Selective Image and Voice Encryption (SSIVE): Privacy Enhancing Strategy

Differentially Private Clustering with Dynamic Noise Adjustment (DPC-DNA) based Fusion Anonymity and Privacy Enhancement in Big Data

Differentially Private Clustering with Dynamic Noise Adjustment (DPC-DNA) based Fusion Anonymity and Privacy Enhancement in Big Data