Attribute Partition based Advanced Homomorphic Privacy Preserving Model for Homogeneous Multiple Datasets

Abstract

With the exponential growth of data size, storage of data and computer memory, an integrated data security model on large datasets is critical. Privacy protection in machine learning is used in real-time applications to turn sensitive data toward privacy. Multiple databases are spread for pattern analysis across multiple users for multiple user applications. The numerous pattern analysis datasets are based on machine learning models such as grouping, clustering or selection function models. In huge databases, the revelation of personal data or trends has become a significant issue. The suggested vertical attribute partitioning Based Privacy Preservation in decision tree building considers rule sharing based classification with secure multiparty computation so as to generate high accuracy in mining results, without compromising privacy. In the suggested approach, a novel homomorphic encryption based partitioning decision tree model is created and implemented on multi-party datasets for privacy preserving. Experimental results revealed that the new system has excellent computational accuracy with privacy in the patterns compared to the existing models.