Heteroscedastic Temporal Variational Autoencoder For Irregular Time Series

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Shukla, Satya Narayan, Marlin, Benjamin M.
Format: Preprint
Published: 2021
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909378333900800
author Shukla, Satya Narayan
Marlin, Benjamin M.
author_facet Shukla, Satya Narayan
Marlin, Benjamin M.
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2107_11350
institution arXiv
publishDate 2021
record_format arxiv
spellingShingle Heteroscedastic Temporal Variational Autoencoder For Irregular Time Series
Shukla, Satya Narayan
Marlin, Benjamin M.
Machine Learning
Artificial Intelligence
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.
title Heteroscedastic Temporal Variational Autoencoder For Irregular Time Series
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2107.11350