A Deep Q-Learning Sanitization Approach for Privacy Preserving Data Mining

Abstract

With the establishment of the 5G network, a number of data-intensive applications will be developed. Privacy of information over the network is increasingly relevant, and require protection. The privacy of information while utilizing data is a trade-off that needs to be addressed. In this paper, we propose data privacy of 5G connected devices over heterogeneous networks (5G-Hetnets). A deep Q learning (DQL) based technique is applied to sensitize sensitive information from a given database while keeping the balance between privacy protection and knowledge discovery during the sanitization process. It takes transaction states as input and results in state and action pair. The DQL discovers the transactions dynamically, then the sanitization operation hide the sensitive information by minimizing side effects. The proposed approach shows significant improvement of performance compared to greedy and meta-heuristics and heuristics approaches.