[2006.14278] Graph Structural-topic Neural Networkopen searchopen navigation menucontact arXivsubscribe to arXiv mailings

Graph Convolutional Networks (GCNs) achieved tremendous success by effectively gathering local features for nodes. However, commonly do GCNs focus more on node features but less on graph structures within the neighborhood, especially higher-order structural patterns. However, such local structural patterns are shown to be indicative of node properties in numerous fields. In addition, it is not just single patterns, but the distribution over all these patterns matter, because networks are complex and the neighborhood of each node consists of a mixture of various nodes and structural patterns. Correspondingly, in this paper, we propose Graph Structural-topic Neural Network, abbreviated GraphSTONE, a GCN model that utilizes topic models of graphs, such that the structural topics capture indicative graph structures broadly from a probabilistic aspect rather than merely a few structures. Specifically, we build topic models upon graphs using anonymous walks and Graph Anchor LDA, an LDA varia

Date: 2020/10/18 12:53

Related Entries

Read more [2010.07515] RNNs can generate bounded hierarchical languages with optimal memoryopen searchopen nav...
0 users, 2 mentions 2020/10/17 23:21
Read more MIT Tech Review: AI学習の常識覆す新手法、 ワンショット「未満」学習
1 users, 17 mentions 2020/10/20 15:52
Read more 「全数調査なら何でもわかる」という誤解 - 間違えがちな母集団とサンプリングそしてベイズ統計 - - ill-identified diary
299 users, 51 mentions 2020/10/21 13:40
Read more 不老におけるOptunaを利用した分散ハイパーパラメータ最適化 - 今村秀明(名古屋大学 Optuna講習会)
1 users, 4 mentions 2020/10/23 03:52
Read more Paddingが「死角」を生む!?そんな盲点を検証! | AI-SCHOLAR | AI:(人工知能)論文・技術情報メディア
2 users, 4 mentions 2020/10/26 11:58