Privacy Preservation of Associative Classification and its Optimization by Machine Learning

Abstract

XVII newlineAbstract newlineKeywords newlineData Mining, Classification, Associative Classification, Privacy preserving data mining, newlineGenetic Algorithm newlineBackground newlineInternet has brought a drastic change in today s world. Due to advancement in usage of newlineInternet and pattern discovery from huge amount of data flowing through internet, newlinepersonal information of an individual or organization can be traced. In today s world, newlineprivacy is the major concern and people are very much concerned about their sensitive newlineinformation which they don t want to share. Hence, to protect the private information is newlinebecoming extremely crucial, which can be achieved through privacy preserving data newlinemining. Nevertheless, data transformation takes place while trying to preserve privacy newlineand as a consequence it effects the accuracy or outcomes. Certainly optimization newlinetechniques namely genetic algorithm will aid to improve the accuracy or outcomes by newlineproviding balance between privacy and accuracy. newlineResearch Objectives newlineFollowing are the main research objectives that has been carried out: newline1) To inspect the various classification techniques for discovering interesting newlinepatterns considering the accuracy and training time parameters. newline2) To examine association functionality and carry out the optimization of association newlinerules with the proposed approach GA-AR and compare it with the existing newlinetechniques such as PSO, CSO, and AMO considering the time and accuracy newlineparameter. newline3) To implement the existing associative classification technique namely CBA on newlinemicro-data to produce class association rules and compare it with proposed newlinetechnique CBFP. newlineXVIII newline4) To apply privacy-preserving methods specifically anonymization for protecting newlinethe privacy of sensitive information from third parties or intruders which newlinegenerates privacy preserved class association rules. newline5) To utilize the genetic algorithm (proposed approach) for improving the outcome newlineproficiency and to compare the genetic algorithm with neural network for newlineoptimizing the outcome while maintaining the usefulness of data. newline6) To reduce t