Advanced Graph Neural Networks

Discover maximal balanced quasi-clique enumeration for signed graphs. A novel NP-hard algorithm identifies cohesive subgraphs with positive and negative edges using branch-and-bound optimization.

Framework for testing conditional and unconditional independence between binary graphs using community correlations and graph encoder embeddings.

LPS-GNN enables efficient graph neural network processing on 100 billion-edge graphs using single GPUs, achieving 13.8% improvements in user acquisition tasks through novel graph partitioning.

Theoretical analysis explains why climate foundation models succeed at field prediction but fail at extreme event detection through in-context learning complexity.

Survey of graph-based techniques in retrieval-augmented generation systems, examining their roles in database construction, algorithms, and reasoning with structured knowledge.

Study proposes learnable communication graphs for multi-agent systems, enabling dynamic information sharing that adapts to task demands and reduces computational resource consumption.

GeoGraphNetworks: 110 validated spatial networks spanning US and UK transportation and hydrological systems in analysis-ready JSON and XLSX formats with complete topological and geographic data.

Systematic mapping of human-centric evaluation approaches for semantic resources like ontologies and knowledge graphs, synthesizing 15 years of research into a theoretical framework.

CORE applies Information Bottleneck principles to augment graph data for link prediction, simultaneously recovering missing edges and reducing noise to enhance model robustness.

Prompt-driven framework combining LLMs with knowledge graphs for reliable knowledge-based question answering through structured subgraph retrieval and stepwise reasoning validation.

Self-supervised heterogeneous graph neural network framework using multi-scale meta-path contrastive learning for improved local-global structural modeling.

Graph matching framework for cross-modality intravascular ultrasound and optical coherence tomography sequence registration with simultaneous temporal and rotational alignment.