Heteroscedastic Temporal Variational Autoencoder For Irregular Time Series
This addresses a significant problem in domains with irregular time series data, offering an incremental improvement over existing deep latent variable models.
The paper tackles the challenge of probabilistic interpolation for irregularly sampled time series by proposing the Heteroscedastic Temporal Variational Autoencoder (HeTVAE), which outperforms baseline and recent models in reflecting variable uncertainty due to sparse sampling.
Irregularly sampled time series commonly occur in several domains where they present a significant challenge to standard deep learning models. In this paper, we propose a new deep learning framework for probabilistic interpolation of irregularly sampled time series that we call the Heteroscedastic Temporal Variational Autoencoder (HeTVAE). HeTVAE includes a novel input layer to encode information about input observation sparsity, a temporal VAE architecture to propagate uncertainty due to input sparsity, and a heteroscedastic output layer to enable variable uncertainty in output interpolations. Our results show that the proposed architecture is better able to reflect variable uncertainty through time due to sparse and irregular sampling than a range of baseline and traditional models, as well as recently proposed deep latent variable models that use homoscedastic output layers.