图神经网络GNN研究进展：表达性、预训练、OGB（附71页ppt）.pdf

发布时间：2022-05-31 发布人：admin 分类：说明书资料大小：20.36M 资料格式：pdf 举报版权申诉

syp_net-12356720-4744300845407772863.pdf-第1页.png

第1页 / 共71页

syp_net-12356720-4744300845407772863.pdf-第2页.png

第2页 / 共71页

syp_net-12356720-4744300845407772863.pdf-第3页.png

第3页 / 共71页

syp_net-12356720-4744300845407772863.pdf-第4页.png

第4页 / 共71页

syp_net-12356720-4744300845407772863.pdf-第5页.png

第5页 / 共71页

syp_net-12356720-4744300845407772863.pdf-第6页.png

第6页 / 共71页

syp_net-12356720-4744300845407772863.pdf-第7页.png

第7页 / 共71页

syp_net-12356720-4744300845407772863.pdf-第8页.png

第8页 / 共71页

文本预览

Advancements in Graph Neural Networks: PGNNs, Pretraining, and OGB Jure Leskovec Includes joint work with W. Hu, J. You, M. Fey, Y. Dong, B. Liu, M. Catasta, K. Xu, S. Jegelka, M. Zitnik, P. Liang, V. Pande

Modern ML Toolbox Images Text/Speech Modern deep learning toolbox is designed for simple sequences & grids Jure Leskovec, Stanford University 2

But not everything can be represented as a sequence or a grid How can we develop neural networks that are much more broadly applicable? New frontiers beyond classic neural networks that learn on images and sequences Jure Leskovec, Stanford University 3

Representation Learning in Graphs z … Input: Network Predictions: Node labels, New links, Generated graphs and subgraphs Jure Leskovec, Stanford University 4

Networks of Interactions Social networks Knowledge graphs Biological networks Complex Systems Molecules Jure Leskovec, Stanford University Code 5

Why is it Hard? Networks are complex! § Arbitrary size and complex topological structure (i.e., no spatial locality like grids) vs. Text Networks § No fixed node ordering or reference point § Often dynamic and have multimodal features Images Jure Leskovec, Stanford University 6

Graph Neural Networks TARGET NODE B A D E INPUT GRAPH C F A B C D A C A B E F A Each node defines a computation graph § Each edge in this graph is a transformation/aggregation function Scarselli et al. 2005. The Graph Neural Network Model. IEEE Transactions on Neural Networks. Jure Leskovec, Stanford University 7

Graph Neural Networks TARGET NODE B A D E INPUT GRAPH C F A B C D A C A B E F A Neural networks Intuition: Nodes aggregate information from their neighbors using neural networks Inductive Representation Learning on Large Graphs. W. Hamilton, R. Ying, J. Leskovec. NIPS, 2017. Jure Leskovec, Stanford University 8

分享到：

赞收藏

资料库

图神经网络GNN研究进展：表达性、预训练、OGB（附71页ppt）.pdf

相关推荐

人工智能

热门标签

最新资料