Probabilistic time series forecasting with deep non-linear state space models

被引:59
作者
Du, Heming [1 ]
Du, Shouguo [2 ]
Li, Wen [3 ]
机构
[1] Australian Natl Univ, Sch Comp Sci, Coll Engn & Comp Sci, Canberra, ACT, Australia
[2] Shanghai Municipal Big Data Ctr, Shanghai, Peoples R China
[3] Shanghai Univ Int Business & Econ, Sch Stat & Informat, Dept Econ Stat, Shanghai, Peoples R China
基金
中国国家自然科学基金;
关键词
Artificial Intelligence; machine learning; time series;
D O I
10.1049/cit2.12085
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Probabilistic time series forecasting aims at estimating future probabilistic distributions based on given time series observations. It is a widespread challenge in various tasks, such as risk management and decision making. To investigate temporal patterns in time series data and predict subsequent probabilities, the state space model (SSM) provides a general framework. Variants of SSM achieve considerable success in many fields, such as engineering and statistics. However, since underlying processes in real-world scenarios are usually unknown and complicated, actual time series observations are always irregular and noisy. Therefore, it is very difficult to determinate an SSM for classical statistical approaches. In this paper, a general time series forecasting framework, called Deep Non-linear State Space Model (DNLSSM), is proposed to predict the probabilistic distribution based on estimated underlying unknown processes from historical time series data. We fuse deep neural networks and statistical methods to iteratively estimate states and network parameters and thus exploit intricate temporal patterns of time series data. In particular, the unscented Kalman filter (UKF) is adopted to calculate marginal likelihoods and update distributions recursively for non-linear functions. After that, a non-linear Joseph form covariance update is developed to ensure that calculated covariance matrices in UKF updates are symmetric and positive definitive. Therefore, the authors enhance the tolerance of UKF to round-off errors and manage to combine UKF and deep neural networks. In this manner, the DNLSSM effectively models non-linear correlations between observed time series data and underlying dynamic processes. Experiments in both synthetic and real-world datasets demonstrate that the DNLSSM consistently improves the accuracy of probability forecasts compared to the baseline methods.
引用
收藏
页码:3 / 13
页数:11
相关论文
共 44 条
[1]  
[Anonymous], 2013, Advances in Neural Information Processing Systems
[2]  
Bar-Shalom Y., 2004, Tracking and Navigation: Theory Algorithms and Software
[3]  
Borovykh A., 2017, Conditional time series forecasting with convolutional neural networks
[4]  
BOX GEP, 1968, ROY STAT SOC C-APP, V17, P91
[5]  
Chatfield C., 2000, TIME SERIES FORECAST, DOI DOI 10.1201/9781420036206/TIME-SERIES-FORECASTING-CHRIS-CHATFIELD
[6]  
Du H., 2021, ARXIV PREPRINT ARXIV
[7]   Learning Object Relation Graph and Tentative Policy for Visual Navigation [J].
Du, Heming ;
Yu, Xin ;
Zheng, Liang .
COMPUTER VISION - ECCV 2020, PT VII, 2020, 12352 :19-34
[8]  
Durbin J., 2012, Time series analysis by state space methods, V38
[9]  
Dymirkovsky G., 2012, 2012 7th IEEE Conference on Industrial Electronics and Applications (ICIEA 2012). Proceedings, P1543, DOI 10.1109/ICIEA.2012.6360969
[10]  
Eleftheriadis S, 2017, ADV NEUR IN, V30