Deep clustering method 优于只对输入数据进行降维而没有聚类优化目标的AE方法 对比AE方法和GAE与ARGA,后者因为利用到图结构,即使没有专门使用聚类目标,效果也更优 Deep graph clustering methods 由于同时利用到了图结构和深度聚类信息,效果较没有聚类优化目标的方法更优。 DGI最大化节点到整张图的互信息来学习节点...
Deep Graph Contrastive Learning: GRACE 所提出的GRACE框架主要包括两个阶段:数据扩充和对比学习。在每次迭代中,我们首先从一组所有可能的扩充中采样两个扩充函数。对于图上的数据扩充,我们在拓扑和属性级别进行混合扩充,以构建不同的节点上下文。 Graph Contrastive Learning with Adaptive Augmentation: GCA 增广是CL的...
Contrastive learning shows great potential in deep clustering. It uses constructed pairs to discover the feature distribution that is required for the clustering task. In addition to conventional augmented pairs, recent methods have introduced more methods of creating highly confident pa...
该论文的related work被分为两段式写法,首先是介绍了deep clustering在各领域的工作,并在结尾处指出:这些方法取得了良好的效果,但它们忽略了聚类分配学习和表示学习之间的联系。相比之下,我们的方法考虑了它们的联系,同时学习特征表示和聚类分配。其次引入了contrastive learning,介绍了当前流行的对比学习方法后,提出问题...
DCRN:Deep Graph Clustering via Dual Correlation Reduction_友谊路夹老师的博客-CSDN博客MCGRL:结点级别与图级别通用的自监督框架:MVGRL-CSDN博客数据增强、网络架构和目标函数这三个关键因素显着决定了对比方法的聚类性能。根据这些因素,我们在表 1 中总结了我们提出的 SCGC 和其他对比深度图聚类方法之间的差异。
论文信息论文标题:Simple Contrastive Graph Clustering论文作者:Yue Liu, Xihong Yang, Sihang Zhou, Xinwang Liu论文来源:2022,arXiv论文地址:download 论文代码:download1 Introduction 贡献:提出了一种简单的对比深度图聚类方法,称为 SCGCSCGC。SCGCSCGC 不
An official source code for paper Hard Sample Aware Network for Contrastive Deep Graph Clustering, accepted by AAAI 2023. Any communications or issues are welcomed. Please contact yueliu19990731@163.com. If you find this repository useful to your research or work, it is really appreciate to star...
An official source code for paperHard Sample Aware Network for Contrastive Deep Graph Clustering, accepted by AAAI 2023. Any communications or issues are welcomed. Please contactyueliu19990731@163.com. If you find this repository useful to your research or work, it is really appreciate to star ...
Simple Contrastive Graph Clustering Contrastive learning has recently attracted plenty of attention in deep graph clustering for its promising performance. However, complicated data augmentations and time-consuming graph convolutional operation undermine the efficiency of these methods. To solve this problem,...
而右图则是作者提出的新颖的Clustering-level的对比学习,这里和昨天那篇论文【AAAI2021】Contrastive Clustering的区别就是这里使用了图结构。 【图结构如何用于聚类】 而图结构到底是怎么用于聚类的呢,这里作者主要是想根据图结构中是否存在边来表示节点是否再一个簇中,这样两两之间有边的节点的相似度为: ...