eduzhai > Applied Sciences > Engineering >

Graph Signal Processing and Deep Learning Convolution Pooling and Topology

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 20 pages

Abstract: Deep learning, particularly convolutional neural networks (CNNs), haveyielded rapid, significant improvements in computer vision and related domains.But conventional deep learning architectures perform poorly when data have anunderlying graph structure, as in social, biological, and many other domains.This paper explores 1)how graph signal processing (GSP) can be used to extendCNN components to graphs in order to improve model performance; and 2)how todesign the graph CNN architecture based on the topology or structure of thedata graph.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...