Place for sharing the trend and foundation graph technology
GraphOmakase
Learning Higher-Order Structure from Incomplete Spatiotemporal Data: Multi-Scale Hypergraph Laplacians with Neural Refinement Learning Higher-Order Structure from Incomplete Spatiotemporal Data: Multi-Scale Hypergraph Laplacians with Neural RefinementSensor networks increasingly govern modern infrastructure, yet the data they lose are rarely missing in the uniform-random patterns assumed by standard imputation benchmarks. Loop detectors
GraphOmakase
Random Sets Graph Neural Networks Random-Set Graph Neural NetworksUncertainty quantification has become an important factor in understanding the data representations produced by Graph Neural Networks (GNNs). Despite their predictive capabilities being ever useful across industrial workspaces, the inherent uncertainty induced by the nature of the data is a huge mitigating
GraphOmakase
Root-to-Leaf Path Random Walks, Normalized Hodge Laplacians, and Cheeger Inequalities on Simplicial Complexes Root-to-Leaf Path Random Walks, Normalized Hodge Laplacians, and Cheeger Inequalities on Simplicial ComplexesWe introduce root-to-leaf path random walks on double covers of graded signed graphs and analyze their behavior in a general setting. Viewing simplicial complexes within
GraphOmakase
Deep GraphRAG: A Balanced Approach to Hierarchical Retrieval and Adaptive Integration Deep GraphRAG: A Balanced Approach to Hierarchical Retrieval and Adaptive IntegrationGraph-based Retrieval-Augmented Generation (GraphRAG) frameworks face a trade-off between the comprehensiveness of global search and the efficiency of local search. Existing methods are often challenged by navigating large-scale hierarchical
GraphOmakase
Topology-Aware Representation Alignment for Semi-Supervised Vision-Language Learning Topology-Aware Representation Alignment for Semi-Supervised Vision-Language LearningVision-language models have shown strong performance, but they often generalize poorly to specialized domains. While semi-supervised vision-language learning mitigates this limitation by leveraging a small set of labeled image-text pairs together with abundant unlabeled images, existing methods
GraphOmakase
Learning Laplacian Forms for Graph Signal Processing via the Deformed Laplacian Learning Laplacian Forms for Graph Signal Processing via the Deformed LaplacianLearning the graph Laplacian from observed data is one of the most investigated and fundamental tasks in Graph Signal Processing (GSP). Different variants of the Laplacian, such as the
GraphOmakase
GraphFrames, a major graph analysis library update New GraphFrames release: Improved performance, new algorithms, and documentation | Sem Sinchenko posted on the topic | LinkedInOn behalf of the GraphFrames maintainers, I am happy to announce the delivery of a new release. It is a significant improvement! It improves performance and memory management:
GraphOmakase
PolyGraph Discrepancy: a classifier-based metric for graph generation PolyGraph Discrepancy: a classifier-based metric for graph generationExisting methods for evaluating graph generative models primarily rely on Maximum Mean Discrepancy (MMD) metrics based on graph descriptors. While these metrics can rank generative models, they do not provide an absolute measure of performance.
GraphOmakase
Graphs are maximally expressive for higher-order interactions Graphs are maximally expressive for higher-order interactionsWe demonstrate that graph-based models are fully capable of representing higher-order interactions, and have a long history of being used for precisely this purpose. This stands in contrast to a common claim in the recent literature on
GraphOmakase
A Case for Hypergraphs to Model and Map SNNs on Neuromorphic Hardware paper link : https://arxiv.org/abs/2601.16118 Keywords * Spiking Neural Network (SNN) * Neuromorphic Hardware * Hypergraph * Graph partitioning * LLM 기반 거대 AI 인프라가 수천 대의 서버로 구성된 대규모 데이터센터를 필요로 하는 것과 대조적으로, 인간의 뇌는 놀라울 정도로 높은
GraphOmakase
Position: Message-passing and spectral GNNs are two sides of the same coin paper link : https://arxiv.org/abs/2602.10031 Keywords * MPNN, Spectral GNN * Graph Shift Operator * Spectral PE * Roadmap and Vision * 새로운 GNN 프로젝트를 위한 적절한 아키텍처들을 구축하고자 할 때 자연스럽게 두 가지 표준적인 접근 방식 사이에서 고민하게 됩니다.
GraphOmakase
RAG-ANYTHING: ALL-IN-ONE RAG FRAMEWORK RAG-Anything: All-in-One RAG FrameworkRetrieval-Augmented Generation (RAG) has emerged as a fundamental paradigm for expanding Large Language Models beyond their static training limitations. However, a critical misalignment exists between current RAG capabilities and real-world information environments. Modern knowledge repositories are inherently multimodal, containing rich combinations of textual