AI RESEARCH
Spectral Embeddings Leak Graph Topology: Theory, Benchmark, and Adaptive Reconstruction
arXiv CS.LG
•
ArXi:2604.21094v1 Announce Type: new Graph Neural Networks (GNNs) excel on relational data, but standard benchmarks unrealistically assume the graph is centrally available. In practice, settings such as Federated Graph Learning, distributed systems, and privacy-sensitive applications involve graph data that are localized, fragmented, noisy, and privacy-leaking. We present a unified framework for this setting. We