Hierarchical Deep Generative Models for Multi-Rate Multivariate Time Series

Zhengping Che, Sanjay Purushotham, Guangyu Li, Bo Jiang, Yan Liu
Proceedings of the 35th International Conference on Machine Learning, PMLR 80:784-793, 2018.

Abstract

Multi-Rate Multivariate Time Series (MR-MTS) are the multivariate time series observations which come with various sampling rates and encode multiple temporal dependencies. State-space models such as Kalman filters and deep learning models such as deep Markov models are mainly designed for time series data with the same sampling rate and cannot capture all the dependencies present in the MR-MTS data. To address this challenge, we propose the Multi-Rate Hierarchical Deep Markov Model (MR-HDMM), a novel deep generative model which uses the latent hierarchical structure with a learnable switch mechanism to capture the temporal dependencies of MR-MTS. Experimental results on two real-world datasets demonstrate that our MR-HDMM model outperforms the existing state-of-the-art deep learning and state-space models on forecasting and interpolation tasks. In addition, the latent hierarchies in our model provide a way to show and interpret the multiple temporal dependencies.

Cite this Paper


BibTeX
@InProceedings{pmlr-v80-che18a, title = {Hierarchical Deep Generative Models for Multi-Rate Multivariate Time Series}, author = {Che, Zhengping and Purushotham, Sanjay and Li, Guangyu and Jiang, Bo and Liu, Yan}, booktitle = {Proceedings of the 35th International Conference on Machine Learning}, pages = {784--793}, year = {2018}, editor = {Dy, Jennifer and Krause, Andreas}, volume = {80}, series = {Proceedings of Machine Learning Research}, month = {10--15 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v80/che18a/che18a.pdf}, url = {http://proceedings.mlr.press/v80/che18a.html}, abstract = {Multi-Rate Multivariate Time Series (MR-MTS) are the multivariate time series observations which come with various sampling rates and encode multiple temporal dependencies. State-space models such as Kalman filters and deep learning models such as deep Markov models are mainly designed for time series data with the same sampling rate and cannot capture all the dependencies present in the MR-MTS data. To address this challenge, we propose the Multi-Rate Hierarchical Deep Markov Model (MR-HDMM), a novel deep generative model which uses the latent hierarchical structure with a learnable switch mechanism to capture the temporal dependencies of MR-MTS. Experimental results on two real-world datasets demonstrate that our MR-HDMM model outperforms the existing state-of-the-art deep learning and state-space models on forecasting and interpolation tasks. In addition, the latent hierarchies in our model provide a way to show and interpret the multiple temporal dependencies.} }
Endnote
%0 Conference Paper %T Hierarchical Deep Generative Models for Multi-Rate Multivariate Time Series %A Zhengping Che %A Sanjay Purushotham %A Guangyu Li %A Bo Jiang %A Yan Liu %B Proceedings of the 35th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2018 %E Jennifer Dy %E Andreas Krause %F pmlr-v80-che18a %I PMLR %P 784--793 %U http://proceedings.mlr.press/v80/che18a.html %V 80 %X Multi-Rate Multivariate Time Series (MR-MTS) are the multivariate time series observations which come with various sampling rates and encode multiple temporal dependencies. State-space models such as Kalman filters and deep learning models such as deep Markov models are mainly designed for time series data with the same sampling rate and cannot capture all the dependencies present in the MR-MTS data. To address this challenge, we propose the Multi-Rate Hierarchical Deep Markov Model (MR-HDMM), a novel deep generative model which uses the latent hierarchical structure with a learnable switch mechanism to capture the temporal dependencies of MR-MTS. Experimental results on two real-world datasets demonstrate that our MR-HDMM model outperforms the existing state-of-the-art deep learning and state-space models on forecasting and interpolation tasks. In addition, the latent hierarchies in our model provide a way to show and interpret the multiple temporal dependencies.
APA
Che, Z., Purushotham, S., Li, G., Jiang, B. & Liu, Y.. (2018). Hierarchical Deep Generative Models for Multi-Rate Multivariate Time Series. Proceedings of the 35th International Conference on Machine Learning, in Proceedings of Machine Learning Research 80:784-793 Available from http://proceedings.mlr.press/v80/che18a.html.

Related Material