Graph Convolution Networks
Zach explains how CNNs have limitations when applied to graphs due to varying numbers of neighbors in nodes. Kipf and Welling introduced graph convolution networks by utilizing the Fourier transform to handle different neighborhood sizes in graphs, similar to audio filtering with Fourier transform. This approach allows for the application of convolution on graphs by modifying frequency components.In this clip
From this podcast
Machine Learning Street Talk (MLST)
#71 - ZAK JOST (Graph Neural Networks + Geometric DL) [UNPLUGGED]
Related Questions