An Efficient Dynamic Privacy-preserving Secure Knn Classification Using Semi-honest Model

Data mining has large applications in many real world areas. Classification is one of the frequently needed tasks in data mining frameworks. For the existing studies, due to the rise of different privacy problems, various solutions to the theoretical and practical classification issues have been suggested under different security models. In this paper proposed a protected k-NN classifier over Paillier encrypted data in the real world dataset. The proposed protocol saves the privacy of user’s input query, confidentiality of data and also hides the data access patterns. Our first work is secure k-NN classifier is to develop on the encrypted data using Paillier scheme under the privacy preserving semihonest protocol model. Also, we empirically estimate the efficiency of our proposed protocol by a real-world dataset under various parameter settings.