Shift Aggregate Extract Networks
This work addresses graph classification challenges for social network analysis, offering improved performance on large datasets, though it appears incremental as it builds on existing decomposition techniques.
The authors tackled the problem of learning effective representations for large graphs, particularly social networks with high variability, by introducing an architecture based on deep hierarchical decompositions. They showed that their approach outperforms state-of-the-art methods on large social network datasets and remains competitive on small chemobiological benchmarks.
We introduce an architecture based on deep hierarchical decompositions to learn effective representations of large graphs. Our framework extends classic R-decompositions used in kernel methods, enabling nested part-of-part relations. Unlike recursive neural networks, which unroll a template on input graphs directly, we unroll a neural network template over the decomposition hierarchy, allowing us to deal with the high degree variability that typically characterize social network graphs. Deep hierarchical decompositions are also amenable to domain compression, a technique that reduces both space and time complexity by exploiting symmetries. We show empirically that our approach is able to outperform current state-of-the-art graph classification methods on large social network datasets, while at the same time being competitive on small chemobiological benchmark datasets.