Attributed hypergraph generation aims to synthesize higher-order interaction structures together with node attributes, requiring the joint modeling of complex topology and structure--attribute dependencies. Existing methods incorporate attribute information into the generation process, but they remain largely topology-driven, where attributes mainly serve as auxiliary conditions for node selection. This limits their ability to capture the coupling among attribute semantics, structural roles, and member co-occurrence patterns. To address the limitations, we propose CaN, a core-aware neural generation framework for attributed hypergraphs. CaN integrates the hierarchical k-core structure as an explicit generative prior and uses deep neural encoders to model dependencies among multi-dimensional node attributes. It contains a core-aware structural feature allocation module that assigns node- and hyperedge-level structural features under global statistical and feasibility constraints, and a dynamic autoregressive member assignment module that constructs hyperedges based on hyperedge context, attribute embeddings, residual degree states, and core-level constraints. A two-stage optimization strategy further enhances generation quality. Experiments on real-world attributed hypergraphs show the effectiveness of CaN in structural fidelity and structure--attribute consistency.
Xiangfei Fang, Ran Bao, Heng Zhang· Proceedings of the 32nd ACM...· 0 citations
Sparse-dense matrix multiplication (SpMM), a fundamental computational kernel in graph analytics and scientific computing, can be substantially accelerated on modern GPUs by leveraging both dense and sparse Tensor Core units, thereby enabling high-throughput computation. However, existing methods often fail to exploit these hardware units effectively when applied to real-world sparse matrices that exhibit strong local structural heterogeneity. In particular, methods that either (i) reorganize all regions into dense-like tiles or (ii) aggressively convert them into strict 2:4 structured sparsity blocks typically incur low effective block density, substantial padding overhead, and nontrivial preprocessing costs. To address these challenges, we propose AFH-SpMM, a novel Auto-Fit Heterogeneous SpMM framework designed for adaptively parallelizing sparse-dense matrix multiplication on Tensor Core-equipped GPUs. Using a 16-row window as the basic processing granularity, AFH-SpMM adaptively maps local regions to two hardware-efficient computation paths: 16 × 16 dense tiles targeted to Dense Tensor Cores and 16 × 8 row-wise 2:4 structured-sparse tiles targeted to Sparse Tensor Cores. For the sparse computation path, AFH-SpMM further exploits local column proximity to mitigate subsequent memory-access and address-generation overheads. At runtime, the two block types are executed within a single fused kernel, while preserving distinct operand layouts and specialized MMA pipelines for each path. Experiments on 600 SuiteSparse matrices across NVIDIA RTX PRO 6000, H100, and A800 show that AFH-SpMM achieves average speedups of 1.33 × (up to 5.72 ×), and often leads cuSPARSE, ASpT, Sputnik, RoDe, Acc-SpMM, and MP-SpMM, with especially strong gains on medium and large matrices.
Zhi-Rui Chen, Heng Zhang, Kai-Fan Jia· Proceedings of the Internati...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
