---
title: "Multi-version Tensor Completion for Time-delayed Spatio-temporal Data"
authors:
  - "Cheng Qian"
  - "Nikos Kargas"
  - "Cao Xiao"
  - "Lucas Glass"
  - "Nicholas Sidiropoulos"
  - "Jimeng Sun"
doi: "10.24963/ijcai.2021/400"
url: "https://ijcai.org/proceedings/2021/400"
pdf: "https://ijcai.org/proceedings/2021/0400.pdf"
venue: "Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence"
year: 2021
pages: "2906-2912"
track: "Main Track"
keywords:
  - "Machine Learning Applications: Applications of Unsupervised Learning"
  - "Machine Learning: Time-series; Data Streams"
  - "Machine Learning: Unsupervised Learning"
---

# Multi-version Tensor Completion for Time-delayed Spatio-temporal Data

Cheng Qian, Nikos Kargas, Cao Xiao, Lucas Glass, Nicholas Sidiropoulos, Jimeng Sun

*Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence*, pages 2906-2912. DOI: [10.24963/ijcai.2021/400](https://doi.org/10.24963/ijcai.2021/400)

## Abstract

Real-world spatio-temporal data is often incomplete or inaccurate due to various data loading delays. For example, a location-disease-time tensor of case counts can have multiple delayed updates of recent temporal slices for some locations or diseases. Recovering such missing or noisy (under-reported) elements of the input tensor can be viewed as a generalized tensor completion problem. Existing tensor completion methods usually assume that i) missing elements are randomly distributed and ii) noise for each tensor element is i.i.d. zero-mean. Both assumptions can be violated for spatio-temporal tensor data. We often observe multiple versions of the input tensor with different under-reporting noise levels. The amount of noise can be time- or location-dependent as more updates are progressively introduced to the tensor. We model such dynamic data as a multi-version tensor with an extra tensor mode capturing the data updates. We propose a low-rank tensor model to predict the updates over time. We demonstrate that our method can accurately predict the ground-truth values of many real-world tensors. We obtain up to  27.2%  lower root mean-squared-error compared to the best baseline method. Finally, we extend our method to track the tensor data over time, leading to significant computational savings.