Adversarial Regularizer (Variational) Graph Autoencoders
In this tutorial, we study how to improve GAE and VGAE by means of an adversarial regularizer. After recalling some material from the previous tutorial and discussing some motivations, we focus on the ARGA and ARGVA autoencoders: We analyze some details of their architecture and training algorithm from the original paper, and we go through their PyTorch-Geometric implementation step by step.
Download the material of the lecture here.