arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2201.11349cs.LGcs.AI
收藏

置信度可能欺骗:分布偏移下图神经网络的自训练

Confidence May Cheat: Self-Training on Graph Neural Networks under Distribution Shift

Beijing University of Posts and Telecommunications(北京邮电大学);Ant Group(蚂蚁集团);Peng Cheng Laboratory(鹏城实验室)

Hongrui Liu, Binbin Hu, Xiao Wang, Chuan Shi, Zhiqiang Zhang, Jun Zhou

摘要:

图卷积网络(GCN)最近引起了人们的广泛兴趣,并在图上实现了最先进的性能,但其成功通常取决于大量昂贵且耗时的标记数据的仔细训练。为了缓解标记数据的稀缺性,自训练方法已被广泛采用,通过标记高置信度的未标记节点,然后将它们添加到训练步骤中。在这条线上,我们对现有的图形自我训练方法进行了实证研究。令人惊讶的是,我们发现高置信度的未标记节点并不总是有用的,甚至引入了分布偏移问题之间的原始标记数据集和增强数据集的自我训练,严重阻碍了自我训练的能力。为此,本文提出了一种新的分布覆盖图自训练框架(DR-GST),它可以恢复原始标记数据集的分布。具体地说,我们首先证明了在分布偏移的情况下,自训练框架中的损失函数的相等性,以及如果每个伪标记节点被适当的系数加权,则人口分布。考虑到系数的难解性,在观测到两者变化趋势一致的情况下,提出用信息增益代替系数,其中信息增益分别通过DR-GST中的dropout变分推断和dropedge变分推断进行估计.然而,这样的加权损失函数将放大不正确的伪标签的影响。因此,我们应用损失校正方法来改善伪标签的质量。我们的理论分析和五个基准数据集上的大量实验都证明了所提出的DR-GST的有效性,以及DR-GST中每个精心设计的组件。

英文摘要:

Graph Convolutional Networks (GCNs) have recently attracted vast interest and achieved state-of-the-art performance on graphs, but its success could typically hinge on careful training with amounts of expensive and time-consuming labeled data. To alleviate labeled data scarcity, self-training methods have been widely adopted on graphs by labeling high-confidence unlabeled nodes and then adding them to the training step. In this line, we empirically make a thorough study for current self-training methods on graphs. Surprisingly, we find that high-confidence unlabeled nodes are not always useful, and even introduce the distribution shift issue between the original labeled dataset and the augmented dataset by self-training, severely hindering the capability of self-training on graphs. To this end, in this paper, we propose a novel Distribution Recovered Graph Self-Training framework (DR-GST), which could recover the distribution of the original labeled dataset. Specifically, we first prove the equality of loss function in self-training framework under the distribution shift case and the population distribution if each pseudo-labeled node is weighted by a proper coefficient. Considering the intractability of the coefficient, we then propose to replace the coefficient with the information gain after observing the same changing trend between them, where information gain is respectively estimated via both dropout variational inference and dropedge variational inference in DR-GST. However, such a weighted loss function will enlarge the impact of incorrect pseudo labels. As a result, we apply the loss correction method to improve the quality of pseudo labels. Both our theoretical analysis and extensive experiments on five benchmark datasets demonstrate the effectiveness of the proposed DR-GST, as well as each well-designed component in DR-GST.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑