Variational Autoencoder with Arbitrary Conditioning

Oleg Ivanov, Michael Figurnov, Dmitry Vetrov

Abstract: We propose a single neural probabilistic model based on variational autoencoder that can be conditioned on an arbitrary subset of observed features and then sample the remaining features in "one shot". The features may be both real-valued and categorical. Training of the model is performed by stochastic variational Bayes. The experimental evaluation on synthetic data, as well as feature imputation and image inpainting problems, shows the effectiveness of the proposed approach and diversity of the generated samples.

Keywords: unsupervised learning, generative models, conditional variational autoencoder, variational autoencoder, missing features multiple imputation, inpainting

TL;DR: We propose an extension of conditional variational autoencoder that allows conditioning on an arbitrary subset of the features and sampling the remaining ones.

Code: [ tigvarts/ucm](https://github.com/tigvarts/ucm) + [ 2 community implementations](https://paperswithcode.com/paper/?openreview=SyxtJh0qYm)

Data: [CelebA](https://paperswithcode.com/dataset/celeba)

Community Implementations: [ 3 code implementations](https://www.catalyzex.com/paper/arxiv:1806.02382/code)