DPHGNN: A Dual Perspective Hypergraph Neural Networks

TL;DR

DPHGNN introduces a dual-perspective hypergraph neural network that captures both lower-order and higher-order semantics, improving hypernode classification and real-world prediction tasks through topology-aware inductive biases.

Contribution

It proposes a novel dual-perspective HGNN with equivariant operator learning, enhancing expressivity and performance over existing models.

Findings

Outperforms seven state-of-the-art baselines on eight datasets.

Achieves ~7% higher macro F1-Score in RTO prediction.

Provides a theoretical framework for hypergraph neural network expressivity.

Abstract

Message passing on hypergraphs has been a standard framework for learning higher-order correlations between hypernodes. Recently-proposed hypergraph neural networks (HGNNs) can be categorized into spatial and spectral methods based on their design choices. In this work, we analyze the impact of change in hypergraph topology on the suboptimal performance of HGNNs and propose DPHGNN, a novel dual-perspective HGNN that introduces equivariant operator learning to capture lower-order semantics by inducing topology-aware spatial and spectral inductive biases. DPHGNN employs a unified framework to dynamically fuse lower-order explicit feature representations from the underlying graph into the super-imposed hypergraph structure. We benchmark DPHGNN over eight benchmark hypergraph datasets for the semi-supervised hypernode classification task and obtain superior performance compared to seven state-of-the-art baselines. We also provide a theoretical framework and a synthetic hypergraph isomorphism test to express the power of spatial HGNNs and quantify the expressivity of DPHGNN beyond the Generalized Weisfeiler Leman (1-GWL) test. Finally, DPHGNN was deployed by our partner e-commerce company for the Return-to-Origin (RTO) prediction task, which shows ~7% higher macro F1-Score than the best baseline.