Contextual Stochastic Block Model (CSBM)
收藏资源简介:
Contextual Stochastic Block Model (CSBM)数据集是由二分Stochastic Block Model (SBM)扩展而来,每个节点除了具有社区标签外,还附加了从高斯混合模型(GMM)中采样的特征向量。该数据集用于研究半监督节点分类问题,旨在通过已知的部分节点标签来准确分类剩余节点。数据集的创建过程基于SBM和GMM的结合,生成了具有社区结构和特征向量的图数据。CSBM数据集主要应用于图神经网络(GNN)的半监督学习领域,旨在解决节点分类问题,特别是在信息理论阈值下的精确恢复问题。
The Contextual Stochastic Block Model (CSBM) dataset is an extension of the bipartite Stochastic Block Model (SBM). Each node is equipped with not only a community label but also a feature vector sampled from the Gaussian Mixture Model (GMM). This dataset is designed for researching semi-supervised node classification tasks, which aims to accurately classify the remaining nodes by leveraging the partially known node labels. The CSBM dataset is constructed by combining SBM and GMM, generating graph data that possesses both community structure and feature vectors. It is primarily applied in the semi-supervised learning field of Graph Neural Networks (GNNs) to address node classification problems, particularly the exact recovery task under the information-theoretic threshold.




