Privacy preservation in social networks using autoencoders
Anand Vikram Singh, M. K. Singh · Computational Methods in Science and Technology · 2024
Owing to social networks’ widespread use, numerous ideas have been put forth to safeguard the networks’ privacy. These works all presuppose that the attackers make use of the same background information. However, many individuals have varying needs when it comes to privacy protection. Therefore, if assaults with identical background knowledge fail to satisfy privacy requirements, it forfeits the opportunity to capitalize on variations in users’ privacy requirements and obtain superior utility. In this work, we present a system from the most common technique to deal with today&s;s real-time problems i.e. Autoencoders i.e. based on neural networks. In this paper, we used the ARNET dataset of around 70K users and implemented Autoencoders to preserve the privacy of users. The performance of the network has been validated by evaluating the performance by the Random Forest classifier before and after anonymization. Performance parameters show the effectiveness of the proposed solution to the dataset of social network users.