其中GNN 和 MLP 是根据无监督损失函数共同优化的,左边的损失函数鼓励类间的节点尽可能的接近,右边则是鼓励类间正交且每个类具有相同数量的节点。 可以得出,该池化不是仅对节点进行删除操作,而是对graph进行了聚类,类间的邻接矩阵和特征矩阵全都改变了。但是此模型却优于谱聚类,因为谱聚类仅仅考虑了邻接矩阵,而该框...
et al. Cell clustering for spatial transcriptomics data with graph neural networks. Nat Comput Sci 2, 399–408 (2022). https://doi.org/10.1038/s43588-022-00266-5 Download citation Received18 October 2021 Accepted19 May 2022 Published27 June 2022 Issue DateJune 2022 DOIhttps://doi.org/...
基于GNN的层次人脸聚类-Learning Hierarchical Graph Neural Networks for Image Clustering 柠濛 25 人赞同了该文章 一、简介 本次介绍的文章来自CVPR 2021,是目前图像聚类领域比较新的一篇文章,作者来自亚马逊aws。 本文提出了一种有监督的层次GNN模型,使用一种新方法融合每一层的的连接分量,从而在下一层形成新的图...
Zhang和Chen[36,36]认为仅从节点对的局部邻居计算连接似然是足够的,并提出了一个Weisfeiler-Lehman Neural Machine [35] 和一个graph neural network [36]来从局部子图学习一般图结构特征。这与我们的工作密切相关,因为聚类任务可以简化为连接预测问题,我们也能利用图神经网络来从局部图进行学习。 Graph convolutional ...
Cell clustering for spatial transcriptomics data with graph neural networks Nat Comput Sci, 2 (2022), pp. 399-408, 10.1038/s43588-022-00266-5 View in ScopusGoogle Scholar [30] Y. Zong, T. Yu, X. Wang, Y. Wang, Z. Hu, Y. Li conST: an interpretable multi-modal contrastive learning...
我们提出了一个层次图神经网络(GNN)模型,该模型学习如何使用一组图像训练集,将一组图像聚类成未知数量的身份,该训练集使用属于不相交身份集的标签进行注释。我们的分层 GNN 使用一种新颖的方法来合并在层次结构的每个级别预测的连接组件,以在下一个级别形成一个新图。与完全无监督的层次聚类不同,分组和复杂性标准的...
Cell clustering for spatial transcriptomics data with graph neural networks. Nat. Comput. Sci. 2, 399–408 (2022). Article CAS PubMed Google Scholar Yuan, Z. et al. SOTIP is a versatile method for microenvironment modeling with spatial omics data. Nat. Commun. 13, 7330 (2022). Article...
Awesome Deep Graph Clustering is a collection of SOTA, novel deep graph clustering methods (papers, codes, and datasets). machine-learningdata-miningdeep-learningclusteringsurveysrepresentation-learningdata-mining-algorithmsnetwork-embeddinggraph-convolutional-networksgcngraph-embeddinggraph-neural-networksself-su...
{yifax, htong, tianjux, yongxinw, yuanjx, wxia, daviwipf, zhaz, soattos}@amazon.comAbstractWe propose a hierarchical graph neural network (GNN)model that learns how to cluster a set of images into an un-known number of identities using a training set of imagesannotated with labels ...
using a combination of discrete variational models with graph neural networks to hierarchically discover different domains and then learning the structure within each domain. To obtain more easily interpretable structures, Sun et al. [37] propose an Edge-Enhanced Graph Auto-Encoder that inducesdeterminis...