Improving Social Network Inference Attacks via Deep Neural Networks
Open AccessIn modern society, social networks play an important role for online users. However, one unignorable problem behind the booming of the services is privacy issues. At the same time, neural networks have been swiftly developed in recent years, and are proven to be very effective in inference attacks. In the first part of this dissertation, a new framework for inference attacks in social networks is developed, which smartly integrates and modifies the existing state-of-the-art Convolutional Neural Network (CNN) models. As a result, the framework can fit wider applicable scenarios for inference attacks no matter whether a user has a legit profile image or not. Moreover, the framework is able to boost the existing high-accuracy CNN for sensitive information prediction. In addition to the framework, the dissertation also shows the detailed configuration of a Fully Connected Neural Network (FCNN) for inference attacks. Furthermore, traditional machine learning algorithms are implemented to compare the results from the constructed FCNN.In the second part of the dissertation, a scheme for social network de-anonymization via Recurrent Neural Network (RNN) is proposed. This scheme improves both traditional structure-based schemes, and traditional Natural Language Processing (NLP) schemes for social network de-anonymization. Specifically, the new scheme is friendlier to modern social network data than structure-based schemes. It also provides less data preprocessing time, easier implementation, and higher tolerance to noisy data than traditional NLP schemes. The results show that the new scheme inhibits many drawbacks of the existing methods, and increases de-anonymization rate.Overall, the key contribution of the dissertation lies in the integration and modification of existing deep neural networks for social network inference attacks. The major challenge is how to adapt deep learning techniques onto modern social networks to maximize both inference attack rate and de-anonymization rate.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.