`

Timezone: »

 
Poster
Dirichlet Graph Variational Autoencoder
Jia Li · Jianwei Yu · Jiajin Li · Honglei Zhang · Kangfei Zhao · Yu Rong · Hong Cheng · Junzhou Huang

Wed Dec 09 09:00 PM -- 11:00 PM (PST) @ Poster Session 4 #1176
in Poster Session 5 »

Graph Neural Networks (GNN) and Variational Autoencoders (VAEs) have been widely used in modeling and generating graphs with latent factors. However there is no clear explanation of what these latent factors are and why they perform well. In this work, we present Dirichlet Graph Variational Autoencoder (DGVAE) with graph cluster memberships as latent factors. Our study connects VAEs based graph generation and balanced graph cut, and provides a new way to understand and improve the internal mechanism of VAEs based graph generation. Specifically, we first interpret the reconstruction term of DGVAE as balanced graph cut in a principled way. Furthermore, motivated by the low pass characteristics in balanced graph cut, we propose a new variant of GNN named Heatts to encode the input graph into cluster memberships. Heatts utilizes the Taylor series for fast computation of Heat kernels and has better low pass characteristics than Graph Convolutional Networks (GCN). Through experiments on graph generation and graph clustering, we demonstrate the effectiveness of our proposed framework.

[Paper]

Author Information

Jia Li (The Chinese University of Hong Kong)
Jianwei Yu (CUHK)
Jiajin Li (The Chinese University of Hong Kong)
Honglei Zhang (Georgia Institute of Technology)
Kangfei Zhao (The Chinese University of Hong Kong)
Yu Rong (Tencent AI Lab)
Hong Cheng (The Chinese University of Hong Kong)
Junzhou Huang (University of Texas at Arlington / Tencent AI Lab)

More from the Same Authors