Anchor-guided Hypergraph Condensation with Dual-level Discrimination

The increasing prevalence of large-scale hypergraphs poses significant computational challenges for hypergraph neural network (HNN) training. To address this, hypergraph condensation (HGC) distills large real hypergraphs into compact yet informative synthetic ones, beyond graph condensation (GC) methods limited to pairwise relations. However, existing HGC methods rely on decoupled training architectures, where structure generators are pre-trained on the original hypergraph but not jointly optimized with condensed features during refinement, resulting in misaligned structures that degrade downstream utility. Moreover, trajectory-based optimization incurs substantial computational overhead in refinement, limiting condensation efficiency. To tackle these issues, we propose \textbf{A}nchor-guided \textbf{H}yper\textbf{G}raph \textbf{C}ondensation with \textbf{D}ual-level \textbf{D}iscrimination (\textbf{AHGCDD}), which consists of three key components: (1) a node initialization module based on Heat Kernel PageRank (HKPR) to encode structural knowledge into feature semantics; (2) an anchor-guided hyperedge synthesis strategy for joint optimization of condensed features and structure; (3) a theoretically grounded dual-level discrimination objective for utility-preserving condensation without redundant HNN training. Extensive experiments demonstrate the superior effectiveness and efficiency of AHGCDD.

Many real-world systems, such as social networks, online recommendation platforms, and biological interactions, involve complex group relationships rather than simple one-to-one connections. Hypergraph neural networks are powerful tools for analyzing such data, but training them on large datasets can require substantial computational resources and memory, making them difficult to use efficiently in practice. Our research focuses on compressing large hypergraph datasets into much smaller synthetic versions while still preserving most of their useful information. Existing methods often struggle because they do not coordinate different parts of the compression process well, and they repeatedly retrain HNN models during optimization, leading to slow and expensive computation. To address these problems, we develop a new framework called AHGCDD. Instead of treating data compression as separate independent steps, our method learns compact representations and relationship structures together in a unified way. It also avoids many unnecessary training procedures, greatly improving efficiency. Experiments on multiple datasets show that our approach can produce smaller yet highly informative hypergraphs, enabling faster training while maintaining strong performance on downstream learning tasks.