Conditional Variational Autoencoders with Learnable Conditional Embeddings

An approach to add conditions to CVAE models without retraining

Requirements

This article is about conditional variational autoencoders (CVAE) and requires a minimal understanding of this type of model. If you are not familiar with CVAEs, I can recommend the following articles: VAEs with PyTorch, Understanding CVAEs. Please familiarize yourself with CVAEs before reading this article. My code examples are written in Python using PyTorch and PyTorch Lightning.

Introduction

I recently came across the paper: “Population-level integration …

deep learning editors pick latent-space mnist variational-autoencoder