Multivariate time series imputation with transformers

Simple item page
buir.contributor.authorYıldız, A. Yarkın
buir.contributor.authorKoç, Emirhan
buir.contributor.authorKoç, Aykut
buir.contributor.orcidYıldız, A. Yarkın|0000-0003-4100-9653
buir.contributor.orcidKoç, Emirhan|0000-0002-7275-1570
buir.contributor.orcidKoç, Aykut|0000-0002-6348-2663
dc.citation.epage2521en_US
dc.citation.spage2517en_US
dc.citation.volumeNumber29en_US
dc.contributor.authorYıldız, A. Yarkın
dc.contributor.authorKoç, Emirhan
dc.contributor.authorKoç, Aykut
dc.date.accessioned2023-02-13T09:10:17Z
dc.date.available2023-02-13T09:10:17Z
dc.date.issued2022-11-25
dc.departmentDepartment of Electrical and Electronics Engineeringen_US
dc.departmentNational Magnetic Resonance Research Center (UMRAM)en_US
dc.description.abstractProcessing time series with missing segments is a fundamental challenge that puts obstacles to advanced analysis in various disciplines such as engineering, medicine, and economics. One of the remedies is imputation to fill the missing values based on observed values properly without undermining performance. We propose the Multivariate Time-Series Imputation with Transformers (MTSIT), a novel method that uses transformer architecture in an unsupervised manner for missing value imputation. Unlike the existing transformer architectures, this model only uses the encoder part of the transformer due to computational benefits. Crucially, MTSIT trains the autoencoder by jointly reconstructing and imputing stochastically-masked inputs via an objective designed for multivariate time-series data. The trained autoencoder is then evaluated for imputing both simulated and real missing values. Experiments show that MTSIT outperforms state-of-the-art imputation methods over benchmark datasets.en_US
dc.description.provenanceSubmitted by Betül Özen (ozen@bilkent.edu.tr) on 2023-02-13T09:10:17Z No. of bitstreams: 1 Multivariate_Time_Series_Imputation_With_Transformers.pdf: 874874 bytes, checksum: e31bae558e888079e8d07f13ed68aef1 (MD5)en
dc.description.provenanceMade available in DSpace on 2023-02-13T09:10:17Z (GMT). No. of bitstreams: 1 Multivariate_Time_Series_Imputation_With_Transformers.pdf: 874874 bytes, checksum: e31bae558e888079e8d07f13ed68aef1 (MD5) Previous issue date: 2022-11-25en
dc.identifier.doi10.1109/LSP.2022.3224880en_US
dc.identifier.eissn1558-2361
dc.identifier.issn1070-9908
dc.identifier.urihttp://hdl.handle.net/11693/111208
dc.language.isoEnglishen_US
dc.publisherIEEEen_US
dc.relation.isversionofhttps://www.doi.org/10.1109/LSP.2022.3224880en_US
dc.source.titleIEEE Signal Processing Lettersen_US
dc.subjectTransformersen_US
dc.subjectTime series analysisen_US
dc.subjectTrainingen_US
dc.subjectDecodingen_US
dc.subjectData modelsen_US
dc.subjectMedical servicesen_US
dc.subjectComputational modelingen_US
dc.titleMultivariate time series imputation with transformersen_US
dc.typeArticleen_US

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
Multivariate_Time_Series_Imputation_With_Transformers.pdf
Size:
854.37 KB
Format:
Adobe Portable Document Format
Description:
Download
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
1.69 KB
Format:
Item-specific license agreed upon to submission
Description:
Download

Collections

Scholarly Publications - Electrical and Electronics Engineering
Scholarly Publications - UMRAM