Please use this identifier to cite or link to this item: https://scholarbank.nus.edu.sg/handle/10635/171912
Title: Factorized Inference in Deep Markov Models for Incomplete Multimodal Time Series
Authors: Tan, Zhi-Xuan
Harold Soh Soon Hong 
Ong, Desmond C 
Keywords: cs.LG
cs.LG
cs.AI
cs.NE
stat.ML
Issue Date: 7-Feb-2020
Publisher: AAAI
Citation: Tan, Zhi-Xuan, Harold Soh Soon Hong, Ong, Desmond C (2020-02-07). Factorized Inference in Deep Markov Models for Incomplete Multimodal Time Series. Thirty-Fourth AAAI Conference on Artificial Intelligence (AAAI). ScholarBank@NUS Repository.
Abstract: Integrating deep learning with latent state space models has the potential to yield temporal models that are powerful, yet tractable and interpretable. Unfortunately, current models are not designed to handle missing data or multiple data modalities, which are both prevalent in real-world data. In this work, we introduce a factorized inference method for Multimodal Deep Markov Models (MDMMs), allowing us to filter and smooth in the presence of missing data, while also performing uncertainty-aware multimodal fusion. We derive this method by factorizing the posterior p(z|x) for non-linear state space models, and develop a variational backward-forward algorithm for inference. Because our method handles incompleteness over both time and modalities, it is capable of interpolation, extrapolation, conditional generation, label prediction, and weakly supervised learning of multimodal time series. We demonstrate these capabilities on both synthetic and real-world multimodal data under high levels of data deletion. Our method performs well even with more than 50% missing data, and outperforms existing deep approaches to inference in latent time series.
Source Title: Thirty-Fourth AAAI Conference on Artificial Intelligence (AAAI)
URI: https://scholarbank.nus.edu.sg/handle/10635/171912
Appears in Collections:Elements
Staff Publications

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
1905.13570v3.pdf2.69 MBAdobe PDF

OPEN

PublishedView/Download

Page view(s)

100
checked on Oct 14, 2021

Download(s)

2
checked on Oct 14, 2021

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.