Random forest based classification of user data and access protection

Abstract

Nowadays Machine learning techniques based Data Privacy is considered as an important factor in efficiency measuring of user sensitive data privacy in a public cloud. At present, data owners are willing to protect their data from unauthorized trainers and users at the time of data classification based on machine learning techniques. The current classification techniques based on privacy preserving allows single owner data sources not a multi-owner data sources. Hence, the learning process depends on single owner information. When a multi-owner data is coming for classification means the present classification system will not be classified efficiently and accurately. Hence, an alternate system is required to handle a multi-owner data classification process with a privacy preserving storage system. The proposed privacy preserving random forest scheme efficiently handles the multi-owner data sources without the involvement of the trusted curator at the time data classification. Here, the random forest classification technique is used for classification and Advanced Encryption Standard (AES)Technique is used for the purpose of maintenance of the privacy during the sending and receiving of statistical information resulted from the analysis and classification process between sender and receiver. The classification process is applied to, student grade system analysis and classification done in a different manner like horizontal and vertical data partition process. When an individual student data is required means horizontal partition is applied else if particular subject details are required means vertical partition is applied. Then, AES encryption technique is applied to protect the user information. When compared to the existing classification and privacy preserving techniques, the proposed machine learning based classification technique with AES provides better privacy to user sensitive data.