Variational Autoencoder with Arbitrary Conditioning

By Oleg Ivanov, Michael Figurnov, Dmitry P. Vetrov. 2018

171 citations recorded by the source corpus.

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.

Subfields: Computer Vision and Pattern Recognition, Statistical and Nonlinear Physics

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Statistical and Nonlinear Physics

Citations in the corpus

Learning Neural Causal Models from Unknown Interventions (2019, methodology): “Neural network methods equipped with learned masks, such as [46, 47, 48, 49], exist in the literature, but only a few [44] have been adapted to causal inference.”
Getting a CLUE: A Method for Explaining Uncertainty Estimates (2020, methodology): “We use the same architecture for the VAEACs used as ground truth generative models in the computationally grounded evaluation framework put forth in Section 4.”
Flow Models for Arbitrary Conditional Likelihoods (2019, methodology): “We compare our model against VAEAC by training them on 100,000 samples from the joint distribution.”
Explaining by Removing: A Unified Framework for Model Explanation (2020, methodology): “(2020) demonstrated a similar approach using a variational autoencoder-like model (Ivanov et al., 2018), and future work could leverage other conditional generative models (Douglas et al.”
GP-VAE: Deep Probabilistic Time Series Imputation (2019, background): “There are also some VAE-based imputation methods that are designed for a setting where the data is complete at training time and the missingness only occurs at test time [15, 16, 20].”
MIWAE: Deep Generative Modelling and Imputation of Incomplete Data Sets (2019, methodology): “A way to design a model with easily approximable conditionals was recently proposed by Ivanov et al. (2019). Although their primary concern was to treat cases that involve complete training sets (like image inpainting), they propose a heuristic modification of their objective to handle missing data ”