HNHN: Hypergraph Networks with Hyperedge Neurons
This work addresses the problem of learning from complex relational data in domains like social networks or biology, offering a novel method for hypergraph analysis.
The authors tackled hypergraph representation learning by proposing HNHN, a hypergraph convolution network with nonlinear activations on both hypernodes and hyperedges and a flexible normalization scheme, resulting in improved classification accuracy and speed on real-world datasets compared to state-of-the-art methods.
Hypergraphs provide a natural representation for many real world datasets. We propose a novel framework, HNHN, for hypergraph representation learning. HNHN is a hypergraph convolution network with nonlinear activation functions applied to both hypernodes and hyperedges, combined with a normalization scheme that can flexibly adjust the importance of high-cardinality hyperedges and high-degree vertices depending on the dataset. We demonstrate improved performance of HNHN in both classification accuracy and speed on real world datasets when compared to state of the art methods.