Time series cluster kernels to exploit informative missingness and incomplete label information

PATTERN RECOGNITION

卷:115

Article

Mikalsen, Karl Oyvind^1,2  Soguero-Ruiz, Cristina³  Bianchi, Filippo Maria⁴  Revhaug, Arthur²  Jenssen, Robert¹ 

[1] UiT Arctic Univ Norway, Dept Phys & Technol, Tromso, Norway

[2] Univ Hosp North Norway UNN, Dept Gastrointestinal Surg, Tromso, Norway

[3] Univ Rey Juan Carlos, Dept Signal Theory & Comm Telemat & Comp, Fuenlabrada, Spain

[4] UiT, Dept Math & Stat, Tromso, Norway

关键词: Multivariate time series;    Kernel methods;    Missing data;    Informative missingness;    Semi-supervised learning;   

DOI  :  10.1016/j.patcog.2021.107896

来源: Elsevier

【 摘 要 】

The time series cluster kernel (TCK) provides a powerful tool for analysing multivariate time series sub-ject to missing data. TCK is designed using an ensemble learning approach in which Bayesian mixture models form the base models. Because of the Bayesian approach, TCK can naturally deal with missing values without resorting to imputation and the ensemble strategy ensures robustness to hyperparame-ters, making it particularly well suited for unsupervised learning. However, TCK assumes missing at random and that the underlying missingness mechanism is ignor-able, i.e. uninformative, an assumption that does not hold in many real-world applications, such as e.g. medicine. To overcome this limitation, we present a kernel capable of exploiting the potentially rich in-formation in the missing values and patterns, as well as the information from the observed data. In our approach, we create a representation of the missing pattern, which is incorporated into mixed mode mix -ture models in such a way that the information provided by the missing patterns is effectively exploited. Moreover, we also propose a semi-supervised kernel, capable of taking advantage of incomplete label information to learn more accurate similarities. Experiments on benchmark data, as well as a real-world case study of patients described by longitudinal electronic health record data who potentially suffer from hospital-acquired infections, demonstrate the effectiveness of the proposed methods. (c) 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ )

【 授权许可】

Free