Adversarial Attacks on Node Embeddings via Graph Poisoning

Understand

The goal of network representation learning is to learn low-dimensional node embeddings that capture the graph structure and are useful for solving downstream tasks.

• However, despite the proliferation of such methods, there is currently no study of their robustness to adversarial attacks.
• We provide the first adversarial vulnerability analysis on the widely used family of methods based on random walks.
• We derive efficient adversarial perturbations that poison the network structure and have a negative effect on both the quality of the embeddings and the downstream tasks.