K nearest neighbor classifier over secured perturbed data

Privacy Preserving Data Mining has gained very specific area of interest for researchers due to the impact of various security issues. With the voluminous growth of data, threat to individual's private information also grows. Developing useful data mining models without accessing private information has become a major concern. Many studies on data perturbation techniques for protecting sensitive data focus on adding noise to the original data. Manipulating Gaussian noise to the sensitive data has somehow balanced the privacy preservation and the utility of data mining. This paper deals with perturbing sensitive data using Gaussian noise and builds a secure kNN classifier model that provides secured mining. We propose an efficient approach that aims to provide better secured data mining result with minimum information loss.