dual-stream heterogeneous graph learning

Traditional GNNs on heterogeneous information networks (HINs) suffer from semantic confusion across node/edge types, limited expressive power due to heterogeneous features, and heavy reliance on manual architecture design. Method: We propose a non-recursive message-passing mechanism that fundamentally eliminates cross-hop type ambiguity; integrate a HIN-customized differentiable neural architecture search (DARTS) to automatically discover optimal heterogeneous neighborhood aggregation paths; and introduce heterogeneous feature disentanglement to enhance representation discriminability. Contribution/Results: This work pioneers the non-recursive paradigm for HIN learning, balancing architectural flexibility and search efficiency. Extensive experiments on multiple standard and large-scale real-world HIN benchmarks demonstrate consistent superiority over state-of-the-art methods in node classification and link prediction, achieving significant average performance gains.

This paper identifies and quantifies a significant performance heterogeneity phenomenon in Graph Neural Networks (GNNs) for graph-level classification and regression: identical models exhibit substantial performance variance across individual graph samples, unexplained solely by topological differences. To address this, we propose a heterogeneity measurement framework based on Tree Mover’s Distance (TMD), the first to jointly model graph topology and node feature distributions. We further design a data-aware selective rewiring strategy and a spectral-adaptive depth selection mechanism. Experiments demonstrate that our approach improves average graph classification accuracy by 2.3% and substantially reduces hyperparameter tuning overhead. We empirically validate that performance heterogeneity strongly correlates with inter-class distance ratios. Moreover, our spectral-driven depth heuristic achieves performance on par with manually optimized layer counts across multiple benchmarks.

Graph neural networks (GNNs) remain challenging to understand and apply for secondary-school students and machine learning practitioners due to conceptual abstraction and fragmented pedagogical resources. Method: We propose a unified encoder–decoder pedagogical and practical framework that systematically integrates core GNN mechanisms—including message passing, GCN, and GAT—and designs task-specific decoders for node classification, link prediction, and other downstream tasks. Grounded in engineering practice, we develop the first reproducible GNN入门 (introductory) curriculum, combining theoretical exposition with large-scale homogeneous-graph experiments. Contribution/Results: We quantitatively characterize how model performance scales with training data size and graph structural complexity—revealing novel empirical patterns. The framework provides standardized benchmarking protocols, hyperparameter tuning guidelines, and task-adaptation strategies. It significantly enhances pedagogical interpretability and industrial deployability of GNNs, lowering barriers to entry without sacrificing technical rigor.

Existing heterogeneous graph neural networks (HGNNs) suffer from parameter explosion and relation collapse—where the number of learnable parameters grows quadratically with the number of relation types, and discriminative capacity across relations degrades—hindering scalability to large-scale heterogeneous graphs with abundant relation schemas. To address this, we propose the Blend&Grind mechanism: a lightweight, scalable HGNN framework that models diverse relations efficiently using a single shared parameter set. It operates in a unified feature space via differentiable relation blending (Blend) and hierarchical relation refinement (Grind), augmented by relation-aware projection and structured regularization. Evaluated on multiple benchmarks, our method reduces model parameters to just 1/28.96 of the prior state-of-the-art, accelerates training throughput by 8.12×, and improves node classification accuracy by up to 7%, thereby substantially overcoming the scalability bottleneck of HGNNs.

This paper addresses three critical challenges in hypergraph learning: (i) the ambiguity of homophily definitions, (ii) architectural neglect of higher-order structural properties, and (iii) structural biases in prevailing benchmark datasets. To tackle these, we propose the first theoretical framework for higher-order homophily, formally defining and empirically validating it as a key determinant of hypergraph neural network (HNN) performance. We introduce a unified MultiSet message-passing paradigm and a novel architecture—MultiSetMixer—featuring hyperedge-aware node representations and joint node-hyperedge random sampling. Furthermore, we systematically expose fundamental structural distributional biases across mainstream benchmarks. Extensive experiments demonstrate that our approach achieves significant improvements over state-of-the-art methods across multiple benchmarks. The work establishes a theoretically grounded, interpretable, and scalable paradigm for hypergraph learning.

This study systematically investigates the effectiveness boundaries and practical value of Graph Neural Networks (GNNs) across twelve application domains. Building upon a unified design space, it derives both spectral and spatial formulations of GNNs from first principles, analyzes their expressive power through the lens of the Weisfeiler–Leman test, and evaluates domain-specific graph construction strategies and architectural choices. The work establishes the first cross-domain analytical framework that disentangles genuine performance gains from baseline biases, uncovering common challenges such as heterophily, scaling effects, and deployment gaps. It clarifies the applicability limits of GNNs, highlights the discrepancy between leaderboard-topping models and deployable ones, and offers constraint-aware practical guidelines to address issues including oversmoothing, over-squashing, and distributional shifts.

Graph Neural Networks (GNNs) often underperform on heterophilous graphs, where adjacent nodes exhibit significantly different labels or features. While existing approaches primarily focus on architectural modifications, they fail to fundamentally alleviate the underlying heterophily issue. This work proposes GRAPHITE, a novel framework that explicitly enhances graph homophily through direct graph transformation. Specifically, GRAPHITE introduces auxiliary feature nodes and reconstructs the graph structure based on a homophily-aware criterion, thereby optimizing message passing. Theoretical analysis and extensive experiments demonstrate that GRAPHITE substantially outperforms state-of-the-art GNN methods across multiple heterophilous graph benchmarks, while maintaining competitive performance on homophilous graphs.

Message-passing graph neural networks (GNNs), particularly low-pass graph convolutional networks (GCNs), suffer from performance degradation on heterophilic graphs. Method: This paper proposes a lightweight graph-structure fine-tuning paradigm, grounded in spectral graph theory. It establishes an analytical relationship between topological perturbations—specifically self-loops and parallel edges—and low-pass filtering behavior, enabling efficient assessment of graph filter adaptability without costly eigen-decomposition. An interpretable structural enhancement strategy is designed, with quantitative characterization of its impact on the Laplacian spectrum distribution. Results: On multiple heterophilic graph benchmarks, adding only self-loops or parallel edges significantly improves GCN accuracy (average +3.2%), while offering computational efficiency, zero retraining overhead, and generalizability to other low-pass GNNs. The work introduces a “structure-as-prior” perspective for heterophilic graph modeling.

This study systematically evaluates the intrinsic validity of heterogeneous graph neural networks (HGNNs), addressing prevalent implicit assumptions and the lack of causal validation in the field. We propose the first causal effect estimation framework for HGNNs, integrating counterfactual analysis, minimal sufficient adjustment set identification, cross-method consistency checks, and sensitivity analysis. Conducting large-scale replication experiments across 21 datasets and 20 baseline models, we find that heterogeneous information exerts a statistically significant positive causal effect on model performance—primarily by enhancing node representation homogeneity and mitigating distributional shift, thereby improving classification discriminability; in contrast, model complexity exhibits no significant causal contribution. The implementation is publicly available, establishing a causal benchmark for interpretable evaluation and architecture design of HGNNs.

Real-world graph data commonly suffer from poor scalability, dynamic evolution, directed heterophily, missing node features, and structural uncertainty—challenging the deployment of Graph Neural Networks (GNNs) in industrial applications such as social networks and recommender systems. To address these multifaceted challenges, we propose five complementary models: SIGN (for scalable static graphs), TGN (for temporal graphs), Dir-GNN (for directed heterophilous graphs), FP (for robust feature propagation under feature absence), and NuGget (for game-theoretic structural inference). Our framework systematically integrates temporal graph modeling, directed graph learning, robust feature propagation, and structure reasoning. It enables efficient training and strong generalization on graphs with millions of nodes. Extensive experiments demonstrate that each model achieves significant improvements over state-of-the-art methods on its respective task, effectively bridging the gap between academic GNN designs and industrial-scale graph requirements.