遵循SSL的框架,GSSL方法的损失函数也可以推广到SSL的损失函数中,其中包含三个部分,如Eq. (1) \mathcal{L}(f)=\underbrace{\mathcal{L}_s\left(f, \mathcal{D}_l\right)}_{\text {supervised loss }}+\lambda \underbrace{\mathcal{L}_u\left(f, \mathcal{D}_u\right)}_{\text {unsupervised...
本文关… 传奇电焊悍匪 [Defense] Graph Structure Learning for Robust Graph Neural Networks 论文链接,代码链接Abstract图神经网络 (GNN) 是图表示学习的强大工具。然而,最近的研究表明,GNN 容易受到精心设计的扰动,称为对抗性攻击。对抗性攻击很容易欺骗 GNN 对下游任务进行预测… 临时注册...
Graph-based semi-supervised methods define a graph where the nodes are labeled and unlabeled examples in the dataset,and edges reflect the similarity of examples.These methods usually assume label smoothness over the graph.Graph methods are nonparametric,discriminative,and transductive in nature.These ...
这启发我们将这个思想应用到 gnn 中,以方便图的半监督学习。 为了对图进行数据增强:提出在 GRAND 中进行随机传播,其中每个节点的特征可以部分或全部被随机删除(dropout),然后受扰动的特征矩阵在图中传播。因此,每个节点都可以对特定的邻域不敏感,从而增加了 GRAND 的健壮性。 此外,随机传播的设计可以自然地分离特征...
在这项工作中,我们通过设计用于半监督学习的图数据增强和一致性正则化策略来解决这些问题。具体来说,我们提出了图随机神经网络(GRAND),一个简单但强大的基于图的半监督学习框架 为了有效地扩充图数据,我们提出了在GRAND中随机传播的方法,即每个节点的特征可以被部分(dropout)或全部随机删除,然后扰动特征矩阵在图上传播...
We consider an EM-like algorithm for semi-supervised learning on deep neural networks: In forward pass, the graph is constructed based on the network output, and the graph is then used for loss calculation to help update the network by back propagation in the backward pass. We demonstrate ...
论文信息 论文标题:Deeper Insights into Graph Convolutional Networks for Semi-Supervised Learning论文作者:Qimai Li, Zhichao Han, Xiao-Ming Wu论文来源:2018, AAAI论文地址:do
Semi-supervisedLearningwithGraphLearning-ConvolutionalNetworks BoJiang,ZiyanZhang,DoudouLin,JinTang ∗ andBinLuo SchoolofComputerScienceandTechnology,AnhuiUniversity,Hefei,230601,China jiangbo@ahu.edu,{zhangziyanahu,ahulindd}@163,ahhftang@gmail,luobin@ahu.edu Abstract GraphConvolutionalNeuralNetworks(graphCNN...
Graph Convolutional Neural Networks (graph CNNs) have been widely used for graph data representation and semi-supervised learning tasks. However, existing graph CNNs generally use a fixed graph which may not be optimal for semi-supervised learning tasks. In this paper, we propose a novel Graph ...
Semi-Supervised Contrastive Learning 形式上,无监督对比学习有望达到如下效果: score(f(xi),f(x+i))≫score(f(xi),f(x−i))score(f(xi),f(xi+))≫score(f(xi),f(xi−)) 也就是正样本之间的距离远远小于负样本对之间的距离,其中ff是编码器,提出的无监督对比损失如下: ...