Data Perturbation Techniques Research Articles

PPRT: Privacy Preserving and Reliable Trust-Aware Platoon Recommendation Scheme in IoV

Vehicular platoon is an evolved driving pattern in Internet of Vehicles (IoV) that consists of a platoon head (PH) vehicle and several member vehicles to reduce fuel consumption and enhance communication efficiency. In such vehicular platoon system, PH vehicles' reputation is calculated according to feedback reports provided by its member vehicles. Hence, it is essential to calculate a truthful reputation for PH vehicle without revealing the privacy and weight information of member vehicles. Inspired by this, we explore homomorphic encryption, data perturbation, and truth discovery techniques to design a privacy-preserving and reliable trust-aware platoon recommendation scheme. Compared with existing schemes, this article features the following advantages. The feedback reports collection process and never reveals the privacy and weight information of member vehicles. Besides, PH vehicles' reputation is iteratively calculated by fog node and cloud service provider without the involvement of member vehicles, which considers high-speed mobility and equipment limitations of vehicles. The security analysis illustrates that our scheme fulfills security requirements of IoV. Further, we deployed Network Simulator-3 (NS-3) and Simulation of Urban Mobility (SUMO) for simulations, which demonstrated that our scheme reduces the computation burden by at least <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\text{42.62{\%}}$</tex-math></inline-formula> , communication burden by at least <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\text{2{\%}}$</tex-math></inline-formula> , and storage burden by at least <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\text{45.95{\%}}$</tex-math></inline-formula> , compared with related schemes.

Adapting Geo-Indistinguishability for Privacy-Preserving Collection of Medical Microdata

In the era of the Fourth Industrial Revolution, the increasing demand for data collection and sharing for analysis purposes has raised concerns regarding privacy violations. Protecting individual privacy during the collection and dissemination of sensitive information has emerged as a critical concern. In this paper, we propose a privacy-preserving framework for collecting users’ medical microdata, utilizing geo-indistinguishability (Geo-I), a concept based on well-known differential privacy. We adapt Geo-I, originally designed for protecting location information privacy, to collect medical microdata while minimizing the reduction in data utility. To mitigate the reduction in data utility caused by the perturbation mechanism of Geo-I, we propose a novel data perturbation technique that utilizes the prior distribution information of the data being collected. The proposed framework enables the collection of perturbed microdata with a distribution similar to that of the original dataset, even in scenarios that demand high levels of privacy protection, typically requiring significant perturbations to the original data. We evaluate the performance of our proposed algorithms using real-world data and demonstrate that our approach significantly outperforms existing methods, ensuring user privacy while preserving data utility in medical data collection.

Evaluation of Key Spatiotemporal Learners for Print Track Anomaly Classification Using Melt Pool Image Streams

Recent applications of machine learning in metal additive manufacturing (MAM) have shown great potential to resolve critical barriers to MAM's widespread adoption. Recent research on the topic highlights the significance of using melt pool signatures to predict defects on-the-fly. While high-fidelity melt pool image data has the potential to enable accurate predictions, hardly any work exists on the use of state-of-the-art spatiotemporal models to leverage the information embedded in the transient and sequential nature of the additive process. This work introduces and implements some of the major deep spatiotemporal learners that can be adapted to classify melt pool image streams from different materials, systems, and applications. In this regard, two-stream networks with a spatial and temporal stream, a recurrent spatial network, and a factorized 3D convolutional neural network were investigated herein. The generalization abilities of these models to perturbations in the melt pool image data are tested using data perturbation techniques that are grounded in palpable process situations. The implemented architectures exhibit the ability to learn the spatiotemporal features of the melt pool image sequences. However, only the Kinetics400 pre-trained SlowFast network, which belongs to the two-stream category, showed strong generalisation abilities to data perturbations.

Exploring the Tradeoff between Utility and Privacy for Analyzing Stock Trend

In this study of exploring the tradeoff between utility and privacy, we have used the S &amp; P 500 data as a proof of concept, for Association Rule Data Mining using privacy-preserved data. The purpose of choosing S &amp; P 500 was that unlike synthetically generated data, S &amp; P 500 index though public, was real-world data nonetheless and hence incorporated in it all the political and social events that occurred over the period of time under consideration. Secondly, we could use stock price as a quasi-identifier. Any seasoned stockbroker would be able to identify the organization simply by looking at the stock price fluctuations over a period even if the ticker information is removed. And stock price is the most important variable used in any data mining related with stock market. Hence, we had to perturb this one variable in order to test for anonymity and utility. Finally, the data being numerical in nature helped us utilize the statistical data perturbation techniques. Association rule mining techniques were first applied to the original data. The data was then perturbed and, once again, using binary similarity algorithms, trend matching was performed on the identified stocks from the original data. The three tests of different random data generations were evaluated for data utility and privacy. The results re-emphasized on the tradeoff between the data utility and privacy where greater perturbation meant greater privacy but higher utility loss and vice versa.

Efficient big data security analysis on HDFS based on combination of clustering and data perturbation algorithm using health care database

In this manuscript proposes an efficient big data security analysis on HDFS based on the combination of Improved Deep Fuzzy K-means Clustering (IDFKM) Algorithm and Modified 3D rotation data perturbation algorithm using health care database. To compile a similar group of data, an Improved Deep Fuzzy K-means Clustering (IDFKM) Algorithm is used as partitioning the medical data. After clustering, Modified 3D rotation data perturbation technique is used to satisfy the privacy requirement of the client. Modified 3D rotation Data Perturbation technique perturbs each and every sensitive data of the cluster and all the key parameters values used for clustering have warehoused in the database file sector. The proposed approach is executed by Java program, its efficiency is assessed by Health care database. The metrics under the study of memory usage attains higher accuracy 34.765%, 23.44%, 52.74%, 18.74%, lower execution time 35.23%, 23.76%, 27.86%, 27.76%, higher Efficiency 26.85%, 38.97%, 28.97%, 35.65%. then the proposed method is compared with the existing methods such asSecurity Analysis of SDN Applications for Big Data with spoofing identity, Tampering with data, Repudiation threats, Information disclosure, Denial of service and Elevation of privileges (STRIDE), Big Data Analysis-based Secure Cluster Management for using Ant Colony Optimization (ACA) Optimized Control Plane in Software-Defined Networks, System Architecture for Secure Authentication and Data Sharing in Cloud Enabled Big Data Environment using LemperlZivMarkow Algorithm (LZMA) and Density-based Clustering of Applications with Noise (DBSCAN), Big Data Based Security Analytics using data based security analytics (BDSA) approach for Protecting Virtualized Infrastructures in Cloud Computing respectively.

Privacy protection model considering privacy-utility trade-off for data publishing of weighted social networks based on MST-clustering and sub-graph generalization

Along with the rapid development of the Internet technology, social network sites (SNSs) are increasingly emerging. As various types of datasets are easily exposed to the network, privacy security in SNS becomes the focus of the study. Anonymization techniques, such as generalization and bucketization, and data perturbation techniques are usually employed in data publishing for privacy protection. However, in the meanwhile, the privacy protection methods reduce the utility of the data. Thus, the issue of privacy-utility trade-off becomes one very real problem. By considering the privacy-utility trade-off and combining the minimum spanning tree (MST) clustering technique and the sub-graph generalization technique, a privacy protection model is proposed for data publishing in weighted social networks. The MST-based clustering method is one representative and concise graph-based clustering algorithm. Meanwhile, as one MST of an edge-weighted graph is the minimum weight spanning tree, some important properties of the graph like the shortest path are preserved in the MST. Finally, on the basis of sub-graphs (clusters) divided by the MST-clustering, three sub-graph generalization approaches are designed in one reasonable way for data publishing to guarantee the privacy-utility trade-off of the proposed method. Experiments and result analysis indicate workability of the proposed method that the proposed model offers privacy protection for the data publishing by means of the sub-graph generalization approach while it can also improve the data utility especially for occasions like data mining, machine learning, and pattern recognition.

Machine learning interpretability for a stress scenario generation in credit scoring based on counterfactuals

To boost the application of machine learning (ML) techniques for credit scoring models, the blackbox problem should be addressed. The primary aim of this paper is to propose a measure based on counterfactuals to evaluate the interpretability of a ML credit scoring technique. Counterfactuals assist with understanding the model with regard to the classification decision boundaries and evaluate model robustness. The second contribution is the development of a data perturbation technique to generate a stress scenario.We apply these two proposals to a dataset on UK unsecured personal loans to compare logistic regression and stochastic gradient boosting (SBG). We show that training a blackbox model (SGB) as conditioned on our data perturbation technique can provide insight into model performance under stressed scenarios. The empirical results show that our interpretability measure is able to capture the classification decision boundary, unlike AUC and the classification accuracy widely used in the banking sector.

Impact of Australian Bureau of Statistics data perturbation techniques on the precision of Census population counts, and the propagation of this impact in a geospatial analysis of high-risk foot hospital admissions among an Indigenous population

ABSTRACT Precise population counts facilitate reliable estimation of differences in disease risk between population sub-groups or geographic regions, which informs rational public health decision-making. In Australia, Census population counts obtainable via the Australian Bureau of Statistics’ (ABS’) TableBuilder tool, which are widely used, are routinely perturbed to protect privacy. We examine this perturbation’s impact on: the precision of TableBuilder population counts, and a geospatial analysis of high-risk foot (HRF) hospital admissions among Western Australian (WA) Indigenous people. The former is severely compromised, disproportionately so for Indigenous people and under the currently employed ‘perturbation without additivity’ (PWoA) approach. E.g., based on fine-resolution, PWoA-affected data, the Indigenous population total for ‘Inner City’ WA was >50% lower than its true total. Imprecise calculation of WA Indigenous HRF admission counts, and misrepresentation of high-rate regions of such admissions in Perth, resulted. An approach suggested to manage this - examination of coarse-resolution data - is inadequate, partly due the modifiable areal unit problem. However, a novel map overlay technique alleviates this issue. We recommend that the ABS review the PWoA approach and consider implementing different statistical disclosure control techniques or reverting to the previously employed ‘perturbation with additivity’ approach, prior to releasing data from the 2021 Australian Census.

K-Anonymization approach for privacy preservation using data perturbation techniques in data mining

Nowadays, privacy is a big issue for everyone because the internet generates and shares massive amounts of data every day. Because traditional privacy-preserving techniques do not protect sensitive data well enough, malicious attacks on sensitive data are possible. Traditional methods may have many drawbacks due to various attacks on sensitive data. Thus, data must be preserved before being shared with others. K-Anonymity protects privacy. By definition, anonymization means that a person is not identifiable, traceable, or reachable. Anonymization using k-anonymity protects data privacy and protects entire datasets. The traditional data perturbation method only works with numerical data, but the proposed method is designed to work with categorical data. The Proposed Framework perturbs data for multiple columns at once. The proposed framework's performance is evaluated in two ways. Initially, it keeps the data mining model accurate. Second, it protects the original data privacy while minimizing data loss. The main motivation for using k-anonymity is to remove or transfer personally identifiable information.

Novel hybrid geometric data perturbation technique by means of sampling data intervals

Privacy preserving data mining has much importance in the data analysis because it maintains confidential or highly sensitive data in the purpose of data analysis. Many perturbation techniques are developed, but do not perform well. There are used for different purposes in various techniques. Though, the reconstruction of original data can be done from the samples of perturbed data. It does not always preserve the distance that exists in the original data samples. During reconstruction of the original data, some of the information is loosed which is again a complex and time consuming task for traditional data mining algorithm. Therefore, a novel hybrid data perturbation technique is introduced that overcomes the above addressed issues. The proposed paper includes a novel data perturbation technique to hybridize the traditional Geometric Data Perturbation Technique along with the Euclidean distance method to preserve privacy. It obtains the distorted or perturbed data which is sampled at different intervals depending on the levels of privacy. The proposed approach attempts to reduce the tradeoffs that persists between privacy levels, accuracy and information loss. The experimental results for this approach is implemented using real world bench mark datasets downloaded from UCI Repositories using three different data mining clustering algorithms. It is shown that clustering accuracy is similar and fairly high for the perturbed data obtained from the proposed approach when matched with the original data set clustering algorithms. The analysis were shown for different privacy levels on different data samples.

Holdout-Based Empirical Assessment of Mixed-Type Synthetic Data.

AI-based data synthesis has seen rapid progress over the last several years and is increasingly recognized for its promise to enable privacy-respecting high-fidelity data sharing. This is reflected by the growing availability of both commercial and open-sourced software solutions for synthesizing private data. However, despite these recent advances, adequately evaluating the quality of generated synthetic datasets is still an open challenge. We aim to close this gap and introduce a novel holdout-based empirical assessment framework for quantifying the fidelity as well as the privacy risk of synthetic data solutions for mixed-type tabular data. Measuring fidelity is based on statistical distances of lower-dimensional marginal distributions, which provide a model-free and easy-to-communicate empirical metric for the representativeness of a synthetic dataset. Privacy risk is assessed by calculating the individual-level distances to closest record with respect to the training data. By showing that the synthetic samples are just as close to the training as to the holdout data, we yield strong evidence that the synthesizer indeed learned to generalize patterns and is independent of individual training records. We empirically demonstrate the presented framework for seven distinct synthetic data solutions across four mixed-type datasets and compare these then to traditional data perturbation techniques. Both a Python-based implementation of the proposed metrics and the demonstration study setup is made available open-source. The results highlight the need to systematically assess the fidelity just as well as the privacy of these emerging class of synthetic data generators.

Advanced extreme learning machine‐based ensemble classification scheme with enhanced data perturbation for human DNA sequences

AbstractThe dramatic growth in machine learning has brought in significant features and quantified non‐linear associations in the data derived from sensitive medical datasets. The data should be preserved without influencing the associated classifications by applying a robust, effective and reliable data perturbation technique before enforcing ensemble classification. In this paper, an Integrated Condensation Scheme imposed Privacy Preserving Rotation‐based Data Perturbation and Ensemble Classification (ICS‐PPR‐DPEC) is proposed for ensuring privacy of such sensitive data. Condensation Algorithm‐based Data Perturbation is used for constructing homogenous groups determined from the distance between tuples. It also generates a rotation matrix for conducting perturbation that ensures higher data sensitivity protection before it is sent for classification. Advanced Extreme Learning Machine‐based Ensemble Classification Scheme includes kernel, norm‐optimized and regularized Extreme Learning Machine (ELM)‐based classifiers for attaining predominant classification accuracy in identifying human DNA sequences. This approach facilitates classification by constructing ensembles which are trained through randomly resampled ELM classifiers. It includes an objective function that systematically improves the accuracy and diversity among resulting ensembles. The experimental results of the proposed ICS‐PPR‐DPEC are found to be excellent in terms of classification Accuracy, Precision, Recall, and Kappa statistic when compared to the benchmarked techniques.

An Application of Random Walk Resampling to Phylogenetic HMM Inference and Learning.

Statistical resampling methods are widely used for confidence interval placement and as a data perturbation technique for statistical inference and learning. An important assumption of popular resampling methods such as the standard bootstrap is that input observations are identically and independently distributed (i.i.d.). However, within the area of computational biology and bioinformatics, many different factors can contribute to intra-sequence dependence, such as recombination and other evolutionary processes governing sequence evolution. The SEquential RESampling ("SERES") framework was previously proposed to relax the simplifying assumption of i.i.d. input observations. SERES resampling takes the form of random walks on an input of either aligned or unaligned biomolecular sequences. This study introduces the first application of SERES random walks on aligned sequence inputs and is also the first to demonstrate the utility of SERES as a data perturbation technique to yield improved statistical estimates. We focus on the classical problem of recombination-aware local genealogical inference. We show in a simulation study that coupling SERES resampling and re-estimation with recHMM, a hidden Markov model-based method, produces local genealogical inferences with consistent and often large improvements in terms of topological accuracy. We further evaluate method performance using empirical HIV genome sequence datasets.

Measurement and control of geo-location privacy on Twitter

The widespread diffusion of Online Social Networks and Media (OSNEM) has generated a huge amount of users’ personal data. As this data is often publicly available, users’ privacy is at risk. To address this issue, users may control the release of their sensitive data on OSNEM. An example of data that users rarely publish is their location. Besides being a privacy-sensitive information, location is a business-relevant data that third parties, e.g., Location-Based Service (LBS) providers, may be interested to obtain. It is, therefore, of paramount importance to understand to what extent the secrecy of location information can be violated. In this work, we investigate how users can measure the privacy of their geo-location on OSNEM and to control the factors affecting it. We define the privacy of a target user as the geographical distance between her actual unexposed location and the location estimated by an attacker. To measure privacy, we propose a novel deep learning architecture that uncovers a target user’s position based only on the publicly-available locations shared by users on Twitter. Results show that locations can be accurately unveiled for the majority of the users, thus suggesting the need for countermeasures to improve their privacy. To control privacy, we propose data perturbation techniques that users can apply to tune the public exposure of their location, and we show the resulting privacy improvements. To shed light on the factors influencing privacy, we then propose a machine learning model that measures privacy based on several users’ features (e.g., social and behavioral characteristics). Unlike the aforementioned deep learning approach, this model also allows to quantify the impact that each feature has on privacy. We observe that features related to the history of users’ visited locations proved to be the most relevant factors affecting privacy. Finally, we explore potential side effects resulting from the application of data perturbation strategies. In particular, we examine, as a study case, the trade-off between users’ privacy and the effectiveness of a proximity marketing LBS. Results suggest that privacy can be guaranteed while not significantly lowering the effectiveness of the LBS.

Best Feature Selection for Horizontally Distributed Private Biomedical Data Based on Genetic Algorithms

Due to the growing success of machine learning in the healthcare domain, medical institutions are striving to share their patients' data in the intention to build more accurate models which will be used to make better decisions. However, due to the privacy of the data, they are reluctant. To build the best models, they have to make the best feature selection for horizontally distributed private biomedical data. The previous proposed solutions are based on data perturbation techniques with the loss of performance. In this article, the researchers propose an original solution without perturbation. This is so the data utility is preserved and therefore the performance. The proposed solution uses a genetic algorithm, a distributed Naïve Bayes classifier, and a trusted third-party. The results obtained by the proposed approach surpass those obtained by other researchers, for the same problem.

A Framework for Privacy-Preserving Multi-Party Skyline Query Based on Homomorphic Encryption

Nowadays, the management and analyses of `big data' are becoming indispensable for numerous organizations all over the world. In many cases, multiple organizations want to perform data analyses on their combined databases. Skyline query is one of the popular operations for selecting representative objects from a large database, where any other object within the database does not dominate each of the representative objects, called `skyline'. Like other data analytics operations, the multi-party skyline query can provide benefits to the participating organizations by retrieving the skyline objects from their combined databases. Such a multi-party skyline query demands the disclosure of individual parties' objects to others during the computation. But, owing to the data privacy and security concern of the present IT era, such disclosure of the individual parties' databases is strictly prohibited. Considering this issue, we are proposing a new framework for the privacy-preserving multi-party skyline query, exploiting additive homomorphic encryption along with data anonymization, perturbation, and randomization techniques. The underlying protocols within our proposed framework ensure that every participating party can identify its multi-party skyline objects without revealing the objects to others during the multi-party skyline query. The detailed privacy and security analyses show that the proposed framework can achieve the desired computation goal without privacy leakage. Besides, the performance evaluation through complexity analyses, extensive simulations, and comprehensive comparison also demonstrate the utility and the efficiency of the proposed framework.

Big Data Performance Analysis on a Hadoop Distributed File System Based on Geometric Data Perturbation Technique

Abstract In this paper, we proposed a Big Data Security Analysis based on a modified perturbation technique in a Hadoop Distributed File System which is commonly used in various real time applications. This work have utilized an improved K-means clustering algorithm, which selects the initial clustering centres based on the density parameters. The poor assurance of initial centres and in order to improve the precision and packing effect of the K-means clustering computation is to be need a new method. In order to group the data of same cluster based on their similarities clustering algorithm was used. In the various clustering methods, this paper is utilizing the geometric data perturbation technique which is clustered, data are perturbed whose values are challenging to be recognized. The performance was evaluated on a sample Health Care database and the metrics under studies are memory usage, precision, recall, accuracy, F-measure, clustering time and execution time, Perturbation time, De-perturbation time. In this paper, we focused only on Perturbation Time, De-Perturbation Time and the Accuracy. The experimental result shows that the proposed approach reduces the complexity and shown to less computation time and better accuracy than that of the existing techniques.

ClaMPP: a cloud-based multi-party privacy preserving classification scheme for distributed applications

Privacy preservation in a distributed environment is a challenging task as it requires efficient control strategies for authentication and integrity preservation of various users and applications. In a distributed environment, most of the data items are distributed among various parties which may be located across different geographical locations. In the literature, most of the techniques proposed for privacy preserving consider only two parties collaboration for data items sharing using data perturbation and homomorphic encryption technique. Since data perturbation is less accurate and homomorphic encryption incurs high computation cost, so to fill these gaps, this paper proposes a Cloud-based Multi-party Privacy Preserving Classification Scheme (ClaMPP), which uses Naive Bayesian Classifier using multi-party random masking and polynomial aggregation technique for privacy preservation among different parties. The proposed ClaMPP scheme consists of four protocols which are described as follows. Protocols I and II are used to calculate conditional probabilities between the item vectors in a secure manner by using Privacy Preserving Conditional Probability Protocol. In the protocol III, the prior probabilities between the item vectors are computed by using additive property of Paillier homomorphic encryption technique. The naive Bayesian classification model constructed from the calculated values from protocols I, II and III is used in protocol IV to generate the predictions for items based on user’s interest. Three datasets MovieLens100K, MovieLens20M and Jester are used in ClaMPP to analyse the privacy, accuracy and overhead costs incurred. It has been experimentally demonstrated that the proposed scheme is secure, and the privacy preservation does not affect the accuracy of the prediction. Comparative analysis of ClaMPP has been done with other related schemes based on off-line model computation cost. It has been demonstrated from the results obtained that computation cost decreases significantly in comparison with the other state-of-art techniques.

Effective Gene Patterned Association Rule Hiding Algorithm for Privacy Preserving Data Mining on Transactional Database

Abstract Association Rule Hiding methodology is a privacy preserving data mining technique that sanitizes the original database by hide sensitive association rules generated from the transactional database. The side effect of association rules hiding technique is to hide certain rules that are not sensitive, failing to hide certain sensitive rules and generating false rules in the resulted database. This affects the privacy of the data and the utility of data mining results. In this paper, a method called Gene Patterned Association Rule Hiding (GPARH) is proposed for preserving privacy of the data and maintaining the data utility, based on data perturbation technique. Using gene selection operation, privacy linked hidden and exposed data items are mapped to the vector data items, thereby obtaining gene based data item. The performance of proposed GPARH is evaluated in terms of metrics such as number of sensitive rules generated, true positive privacy rate and execution time for selecting the sensitive rules by using Abalone and Taxi Service Trajectory datasets.

Optimal model averaging by data perturbation for spatial noisy data

Geostatistical models play an important role in spatial data analysis, in which model selection is inevitable. Model selection methods, such as AIC and BIC, are popular for selecting appropriate models. In recent years, some model averaging methods, such as smoothed AIC and smoothed BIC, are also applied to spatial data models. However, the corresponding averaging estimators are outperformed by optimal model averaging estimators (Hansen in Econometrica 75:1175–1189, 2007) for the ordinary linear models. Therefore, this paper focuses on the optimal model averaging method for geostatistical models. We propose a weight choice criterion for the model averaging estimator on the basis of the generalized degrees of freedom and data perturbation technique. We further theoretically prove the resultant estimator is asymptotically optimal in terms of the mean squared error, and numerically demonstrate its satisfactory performance. Finally, the proposed method is applied to a mercury data set.