Abstract:In real-world scenarios like traffic and energy, massive time-series data with missing values and noises are widely observed, even sampled irregularly. While many imputation methods have been proposed, most of them work with a local horizon, which means models are trained by splitting the long sequence into batches of fit-sized patches. This local horizon can make models ignore global trends or periodic patterns. More importantly, almost all methods assume the observations are sampled at regular time stamps, and fail to handle complex irregular sampled time series arising from different applications. Thirdly, most existing methods are learned in an offline manner. Thus, it is not suitable for many applications with fast-arriving streaming data. To overcome these limitations, we propose BayOTIDE: Bayesian Online Multivariate Time series Imputation with functional decomposition. We treat the multivariate time series as the weighted combination of groups of low-rank temporal factors with different patterns. We apply a group of Gaussian Processes (GPs) with different kernels as functional priors to fit the factors. For computational efficiency, we further convert the GPs into a state-space prior by constructing an equivalent stochastic differential equation (SDE), and developing a scalable algorithm for online inference. The proposed method can not only handle imputation over arbitrary time stamps, but also offer uncertainty quantification and interpretability for the downstream application. We evaluate our method on both synthetic and real-world datasets.We release the code at {<a class="link-external link-https" href="https://github.com/xuangu-fang/BayOTIDE" rel="external noopener nofollow">this https URL</a>}

Factorization of posteriors and partial imputation algorithm for graphical models with missing data

Mixed Graphical Models with Missing Data and the Partial Imputation EM Algorithm

Factor Graph Model Based User Profile Matching Across Social Networks

Maximum a Posteriori Inference for Factor Graphs via Benders' Decomposition

Gaussian Graphical Model Estimation and Selection for High-Dimensional Incomplete Data Using Multiple Imputation and Horseshoe Estimators

Integrating Approximate Single Factor Graphical Models.

Online Missing Value Imputation for High-Dimensional Mixed-Type Data via Generalized Factor Models

A Factor Tree Inference Algorithm for Bayesian Networks and Its Application

Incorporating graph information in Bayesian factor analysis with robust and adaptive shrinkage priors

Online Inference Based on Approximate Factors for Probabilistic Knowledge Bases

Coupled Compound Poisson Factorization

BAYESIAN SEQUENTIAL LEARNING FROM INCOMPLETE DATA ON DECOMPOSABLE GRAPHICAL MODELS

Scalable Probabilistic Matrix Factorization with Graph-Based Priors

Computing Bayes Factors from Data with Missing Values

Partial Gaussian Graphical Model Estimation

Measuring Multi-Source Redundancy in Factor Graphs

Factor Graph Neural Networks

Efficient Learning of Discrete Graphical Models

Joint estimation of multiple dependent Gaussian graphical models with applications to mouse genomics

BayOTIDE: Bayesian Online Multivariate Time series Imputation with functional decomposition

Simultaneous Inference For High-Dimensional Approximate Factor Model