- 1、本文档共67页,可阅读全部内容。
- 2、原创力文档(book118)网站文档一经付费(服务费),不意味着购买了该文档的版权,仅供个人/单位学习、研究之用,不得用于商业用途,未经授权,严禁复制、发行、汇编、翻译或者网络传播等,侵权必究。
- 3、本站所有内容均由合作方或网友上传,本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺!文档内容仅供研究参考,付费前请自行鉴别。如您付费,意味着您自己接受本站规则且自行承担风险,本站不退款、不进行额外附加服务;查看《如何避免下载的几个坑》。如果您已付费下载过本站文档,您可以点击 这里二次下载。
- 4、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等,请点击“版权申诉”(推荐),也可以打举报电话:400-050-0827(电话支持时间:9:00-18:30)。
查看更多
Heterogeneous Graph Neural Network
and Applications
Prof. Chuan Shi
shichuan@
Beijing University of Posts and Telecommunications
© 2009 BUPT TSEG
2 2
Step-by-Step Representation Learning
➢ Traditional representation learning models use hand-
crafted features and relatively simple trainable classifier.
hand-
crafted “Simple”
feature Trainable output
Classifier
extractor
➢ Has the following limitations:
Very Tedious and Costly to develop hand-crafted features
Usually highly dependents on one application, and cannot be
transferred easily to other applications
3 3
End-to-End Representation Learning
Low-
level Mid-level High-level Trainable output
features features classifier
features
➢ Hierarchy of representations with increasing level of
abstraction.
➢ Image recognition
Pixel → edge → texton → motif → part → object
➢ Text:
Character → word → word group → clause → sentence → story
4
4
Why do network embedding?
➢ Networks are a general language for describing and
modeling complex systems
Social Economic networks Biomedical networks
networks
Information I
文档评论(0)