Random-data perturbation techniques and privacy-preserving data mining

Kargupta, Hillol; Datta, Souptik; Wang, Qi; Sivakumar, Krishnamoorthy

doi:10.1007/s10115-004-0173-6

Cited by 170 publications

(89 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…However, data perturbation techniques have the drawback that they do not yield accurate aggregation results. It is noted by Kargupta et al (Kargupta, et al (2005)) that random matrices have predictable structures in the spectral domain. This predictability develops a random matrixbased spectral-filtering technique which retrieves original data from the dataset distorted by adding random values.…”

Section: Wsn Privacymentioning

confidence: 98%

Security and Privacy in Wireless Sensor Networks

Ukil¹

2010

Smart Wireless Sensor Networks

View full text Add to dashboard Cite

Section: Wsn Privacymentioning

confidence: 98%

Security and Privacy in Wireless Sensor Networks

Ukil¹

2010

Smart Wireless Sensor Networks

View full text Add to dashboard Cite

“…They proposed a spectral filtering technique which makes use of the theory of random matrices to produce a close estimate of an original data set from the perturbed (released) version of the data set. [13] Islam et al proposed a framework for adding noise to all attributes both numerical and categorical in two steps; in the first step following a data swapping technique we add noise to sensitive class attribute values, which are also known as labels [11]. Peng peng Lin et.al in their work have explored the use of feature selection techniques for privacy preservation purpose.…”

Section: Literature Surveymentioning

confidence: 99%

Hybrid Perturbation Technique using Feature Selection Method for Privacy Preservation in Data Mining

Priyadarsini¹,

Valarmathi²,

Sivakumari³

2012

IJCA

View full text Add to dashboard Cite

Privacy-preserving in data mining refers to the area of data mining that seeks to safeguard sensitive information from unsolicited or unsanctioned disclosure and hence protecting individual data records and their privacy. Data perturbation is a privacy preservation technique which does addition / multiplication of noise to the original data. It performs anonymization based on the data type of sensitive data. Generalization is a technique were quasi identifiers data are replaced by some other more general term. In this paper privacy protection is applied to high dimensional datasets like Adult and Census. For ranking the attributes, information gain feature subset selection method is used. The high ranking attributes with sensitive information are set as quasi identifiers of the datasets. A hybrid perturbation technique is used to perturb categorical and numeric attributes of both the datasets and the utility of the datasets is measured using accuracy on data mining functionalities. The data distortion is measured using maintenance of Rank of Features (CK) between the original and perturb datasets. Experimental results show that utility of the perturbed datasets comparable with the original dataset and the Census dataset has comparable CK value than adult dataset.

show abstract

“…The random noise is then added to the original variable. There are also more complex mixture models that can be used for adding noise which achieve higher protection levels since it has been found that additive random noise can yield high reidentification risk (Kargupta, et al, 2005).…”

Section: Adding Noise To Continuous Variablesmentioning

confidence: 99%

Measurement Error and Statistical Disclosure Control

Shlomo

2010

Privacy in Statistical Databases

View full text Add to dashboard Cite

Statistical agencies release microdata to researchers after applying statistical disclosure control (SDC) methods. Noise addition is a perturbative SDC method which is carried out by adding independent random noise to a continuous variable or by misclassifying values of a categorical variable according to a probability mechanism. Because these errors are purposely introduced into the data by the statistical agency, the perturbation parameters are known and can be used by researchers to adjust statistical inference through measurement error models. However, statistical agencies rarely release perturbation parameters and therefore modifications to SDC methods are proposed that a priori ensure valid inferences on perturbed datasets. Measurement Error and Statistical Disclosure ControlNatalie Shlomo 1 1 Southampton Statistical Sciences Research Institute, University of Southampton, Highfield, Southampton, SO17 1BJ, United Kingdom N.Shlomo@Soton.ac.ukAbstract. Statistical agencies release microdata to researchers after applying statistical disclosure control (SDC) methods. Noise addition is a perturbative SDC method which is carried out by adding independent random noise to a continuous variable or by misclassifying values of a categorical variable according to a probability mechanism. Because these errors are purposely introduced into the data by the statistical agency, the perturbation parameters are known and can be used by researchers to adjust statistical inference through measurement error models. However, statistical agencies rarely release perturbation parameters and therefore modifications to SDC methods are proposed that a priori ensure valid inferences on perturbed datasets.

show abstract

Random-data perturbation techniques and privacy-preserving data mining

Cited by 170 publications

References 29 publications

Security and Privacy in Wireless Sensor Networks

Security and Privacy in Wireless Sensor Networks

Hybrid Perturbation Technique using Feature Selection Method for Privacy Preservation in Data Mining

Measurement Error and Statistical Disclosure Control

Contact Info

Product

Resources

About