SPHINX: Structural Prediction using Hypergraph Inference Network

2024-10-04

arXiv.org

AI Summary

High-order relational structures are ubiquitous in real-world scenarios yet notoriously difficult to annotate; existing methods either ignore higher-order interactions or artificially reduce them to pairwise edges, leading to structural distortion. To address this, we propose an unsupervised implicit hypergraph inference framework that learns latent high-order topology solely from node-level observational signals—without requiring any hypergraph prior or ground-truth annotations. Our core innovation is a soft-clustering-driven, differentiable hyperedge generation mechanism, coupled with k-subset differentiable sampling to ensure stable and discrete hypergraph construction. The framework is jointly optimized end-to-end with hypergraph neural networks. Evaluated on trajectory prediction, it successfully recovers interpretable high-order collaborative patterns and achieves significant accuracy gains. Ablation studies confirm that the inferred hyperstructure delivers substantial modeling utility beyond conventional graph abstractions.

Technology Category

• Reasoning under Uncertainty: Relational Probabilistic Models
• Machine Learning: Graph-based Machine Learning
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Web Mining and Content Analysis: Bridging structured and unstructured data

Abstract

The importance of higher-order relations is widely recognized in a large number of real-world systems. However, annotating them is a tedious and sometimes impossible task. Consequently, current approaches for data modelling either ignore the higher-order interactions altogether or simplify them into pairwise connections. In order to facilitate higher-order processing, even when a hypergraph structure is not available, we introduce Structural Prediction using Hypergraph Inference Network (SPHINX), a model that learns to infer a latent hypergraph structure in an unsupervised way, solely from the final node-level signal. The model consists of a soft, differentiable clustering method used to sequentially predict, for each hyperedge, the probability distribution over the nodes and a sampling algorithm that converts them into an explicit hypergraph structure. We show that the recent advancement in k-subset sampling represents a suitable tool for producing discrete hypergraph structures, addressing some of the training instabilities exhibited by prior works. The resulting model can generate the higher-order structure necessary for any modern hypergraph neural network, facilitating the capture of higher-order interaction in domains where annotating them is difficult. Through extensive ablation studies and experiments conducted on two challenging datasets for trajectory prediction, we demonstrate that our model is capable of inferring suitable latent hypergraphs, that are interpretable and enhance the final performance.

Problem

• Infers latent hypergraph structure from node signals
• Enables higher-order processing without annotated hypergraphs
• Improves performance in trajectory prediction tasks

Innovation

• Unsupervised hypergraph structure inference
• Differentiable clustering for hyperedge prediction
• K-subset sampling for discrete hypergraphs