Privacy-Preserving Graph Convolutional Networks for Text Classification
This addresses privacy risks for sensitive text data in domains like social networks, though it is incremental as it applies existing differential privacy techniques to GCNs.
The paper tackles the problem of privacy leaks in graph convolutional networks (GCNs) for text classification by adapting differential privacy to GCN training, achieving up to 90% performance of non-private variants with strong privacy guarantees of epsilon = 1.0.
Graph convolutional networks (GCNs) are a powerful architecture for representation learning on documents that naturally occur as graphs, e.g., citation or social networks. However, sensitive personal information, such as documents with people's profiles or relationships as edges, are prone to privacy leaks, as the trained model might reveal the original input. Although differential privacy (DP) offers a well-founded privacy-preserving framework, GCNs pose theoretical and practical challenges due to their training specifics. We address these challenges by adapting differentially-private gradient-based training to GCNs and conduct experiments using two optimizers on five NLP datasets in two languages. We propose a simple yet efficient method based on random graph splits that not only improves the baseline privacy bounds by a factor of 2.7 while retaining competitive F1 scores, but also provides strong privacy guarantees of epsilon = 1.0. We show that, under certain modeling choices, privacy-preserving GCNs perform up to 90% of their non-private variants, while formally guaranteeing strong privacy measures.