SGVAE: Sequential Graph Variational Autoencoder
This work addresses scalability and expressivity issues in graph generative models for researchers in machine learning, but it appears incremental as it builds on existing variational autoencoder frameworks with a sequential approach.
The authors tackled the problem of limited scalability and expressivity in generative models of graphs by introducing a sequential graphical variational autoencoder that operates directly on graphical representations, framing graph encoding and decoding as sequential processes to learn a latent space; experiments on a cycle dataset showed promise but indicated a need for relaxing the distribution over node permutations.
Generative models of graphs are well-known, but many existing models are limited in scalability and expressivity. We present a novel sequential graphical variational autoencoder operating directly on graphical representations of data. In our model, the encoding and decoding of a graph as is framed as a sequential deconstruction and construction process, respectively, enabling the the learning of a latent space. Experiments on a cycle dataset show promise, but highlight the need for a relaxation of the distribution over node permutations.