Learning Conditional Variational Autoencoders with Missing Covariates
This addresses a limitation in generative modeling for datasets with incomplete covariates, though it is incremental as it extends existing CVAE frameworks to handle missing data.
The authors tackled the problem of learning conditional variational autoencoders (CVAEs) when auxiliary covariates have missing values, by proposing a method that augments CVAEs with a prior for missing covariates and uses amortised variational inference, resulting in outperformance over previous methods on simulated and clinical trial datasets.
Conditional variational autoencoders (CVAEs) are versatile deep generative models that extend the standard VAE framework by conditioning the generative model with auxiliary covariates. The original CVAE model assumes that the data samples are independent, whereas more recent conditional VAE models, such as the Gaussian process (GP) prior VAEs, can account for complex correlation structures across all data samples. While several methods have been proposed to learn standard VAEs from partially observed datasets, these methods fall short for conditional VAEs. In this work, we propose a method to learn conditional VAEs from datasets in which auxiliary covariates can contain missing values as well. The proposed method augments the conditional VAEs with a prior distribution for the missing covariates and estimates their posterior using amortised variational inference. At training time, our method marginalises the uncertainty associated with the missing covariates while simultaneously maximising the evidence lower bound. We develop computationally efficient methods to learn CVAEs and GP prior VAEs that are compatible with mini-batching. Our experiments on simulated datasets as well as on a clinical trial study show that the proposed method outperforms previous methods in learning conditional VAEs from non-temporal, temporal, and longitudinal datasets.