A novel technique to secure the acute myocardial infarct images

Image

A novel technique to secure the acute myocardial infarct images

Privacy is an essential part in various applications of data mining, which mostly considers the areas of forensics, health, financial, behavioural, and other sort of confidential data. They may occur due to the requirement to create user profiles, build models related to social network and to detect terrorism among others. For example, mining the data of a health care system may have the necessity to dissect clinical records and other medical transactions. But the underlying problem is that the privacy laws may be broken when such different data sets of different users are combined. It is unsafe to allow the health organizations to disclose the data even though the identifiers are deleted because the original information can be identified by the building of identification attacks for connecting different data set. Thus arises the need for better techniques which pay attention for securing private information. It also preserves the statistical behaviour and characteristics which are necessary for data mining related applications. The approach, we discuss in this paper are defined in the following way: Assume there are N organizations A1; A2; …; AN, where every organization, AI contains a transaction database DBi. It is quite common that some statistics features of the union of the databases required by the data miner.

Though the organizations agree with the fact, they don’t like to outsource their actual information. It is very difficult for the third party user to analyse the data without balancing the privacy of such data. This is known as the census problem, which is illustrated by Chawla et al. At this point, the original data are generally perturbed and it is disclosed in its distorted form. Any user can access the released data. The work specifically takes into account a proposed technique to maintain privacy. This is boosted by the result furnished by Kargupta et al. in their research, which pinpoints the drawbacks of additive data perturbation. In particular, the research work well discovers the chances of applying the technique of ‘project edge’ for building a modified form of data. The distorted data is revealed to the user who is mining the data. It can also be explained and proved that the statistical properties are maintained well in the distorted form of data. The theorem of Johnson and Lindenstrauss laid the foundation for this approach which proves that a collection of points in a Euclidean space which is n-dimensional and can be mapped onto another subspace which is p-dimensional, where p=log n. Thus the pairwise length of the two points is secured by an atomic value. Hence, it is understood that the original information is susceptible to change when the data is mapped onto a lower subspace, while preserving its statistical characteristics. It is assumed that the confidential data is from the same domain and there is no sort of collusion between the parties.

For more kindly go through: Biomedical Research

Biomedical Research accepts direct submissions from authors: Attach your word file with e-mail and send it to biomedres@peerjournal.org

Media Contact:
Joel James
Managing Editor
Biomedical Research