Multivariate Time Series Open-Set Recognition Using Multi-Feature Extraction and Reconstruction

Hyeryeong Oh; Seoung Bum Kim

doi:10.1109/ACCESS.2022.3222310

Multivariate Time Series Open-Set Recognition Using Multi-Feature Extraction and Reconstruction

Hyeryeong Oh, Seoung Bum Kim

Research output: Contribution to journal › Article › peer-review

2 Citations (Scopus)

Abstract

In real-world classification tasks, deep neural networks show innovative performance in various fields. However, traditional classification methods are constructed based on a set of predefined classes and force unknown classes that determine their categorization into one of the predefined classes. This problem is addressed by the research field known as open-set recognition. Existing open-set recognition methods claim that the unique features of unknowns cannot be maintained by using only the final features. In other words, the various feature extraction methods should be considered to effectively reflect the characteristics of unknowns. In this study, we propose an open-set recognition model equipped with multi-feature extraction for multivariate time series data. The results of experiments with various multivariate time series datasets indicate that the proposed method shows improved capability to detect unknown classes while maintaining good predictive performance.

Original language	English
Pages (from-to)	120063-120073
Number of pages	11
Journal	IEEE Access
Volume	10
DOIs	https://doi.org/10.1109/ACCESS.2022.3222310
Publication status	Published - 2022

Bibliographical note

Funding Information:
This work was supported in part by the Brain Korea 21 FOUR, the Ministry of Science and ICT (MSIT) in Korea under the ITRC Support Program supervised by the Institute for Information Communication Technology Planning and Evaluation, under Grant IITP-2020-0-01749; and in part by the National Research Foundation of Korea Grant through the Korea Government under Grant RS-2022-00144190.

Publisher Copyright:
© 2013 IEEE.

Keywords

multichannel signal
multisensor signal
multivariate time series
Open-set recognition
reconstruction error

ASJC Scopus subject areas

General Computer Science
General Materials Science
General Engineering
Electrical and Electronic Engineering

Access to Document

10.1109/ACCESS.2022.3222310

Cite this

@article{18c75948e287424184ac725bf5c6c0e1,

title = "Multivariate Time Series Open-Set Recognition Using Multi-Feature Extraction and Reconstruction",

abstract = "In real-world classification tasks, deep neural networks show innovative performance in various fields. However, traditional classification methods are constructed based on a set of predefined classes and force unknown classes that determine their categorization into one of the predefined classes. This problem is addressed by the research field known as open-set recognition. Existing open-set recognition methods claim that the unique features of unknowns cannot be maintained by using only the final features. In other words, the various feature extraction methods should be considered to effectively reflect the characteristics of unknowns. In this study, we propose an open-set recognition model equipped with multi-feature extraction for multivariate time series data. The results of experiments with various multivariate time series datasets indicate that the proposed method shows improved capability to detect unknown classes while maintaining good predictive performance.",

keywords = "multichannel signal, multisensor signal, multivariate time series, Open-set recognition, reconstruction error",

author = "Hyeryeong Oh and Kim, {Seoung Bum}",

note = "Funding Information: This work was supported in part by the Brain Korea 21 FOUR, the Ministry of Science and ICT (MSIT) in Korea under the ITRC Support Program supervised by the Institute for Information Communication Technology Planning and Evaluation, under Grant IITP-2020-0-01749; and in part by the National Research Foundation of Korea Grant through the Korea Government under Grant RS-2022-00144190. Publisher Copyright: {\textcopyright} 2013 IEEE.",

year = "2022",

doi = "10.1109/ACCESS.2022.3222310",

language = "English",

volume = "10",

pages = "120063--120073",

journal = "IEEE Access",

issn = "2169-3536",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - JOUR

T1 - Multivariate Time Series Open-Set Recognition Using Multi-Feature Extraction and Reconstruction

AU - Oh, Hyeryeong

AU - Kim, Seoung Bum

N1 - Funding Information: This work was supported in part by the Brain Korea 21 FOUR, the Ministry of Science and ICT (MSIT) in Korea under the ITRC Support Program supervised by the Institute for Information Communication Technology Planning and Evaluation, under Grant IITP-2020-0-01749; and in part by the National Research Foundation of Korea Grant through the Korea Government under Grant RS-2022-00144190. Publisher Copyright: © 2013 IEEE.

PY - 2022

Y1 - 2022

N2 - In real-world classification tasks, deep neural networks show innovative performance in various fields. However, traditional classification methods are constructed based on a set of predefined classes and force unknown classes that determine their categorization into one of the predefined classes. This problem is addressed by the research field known as open-set recognition. Existing open-set recognition methods claim that the unique features of unknowns cannot be maintained by using only the final features. In other words, the various feature extraction methods should be considered to effectively reflect the characteristics of unknowns. In this study, we propose an open-set recognition model equipped with multi-feature extraction for multivariate time series data. The results of experiments with various multivariate time series datasets indicate that the proposed method shows improved capability to detect unknown classes while maintaining good predictive performance.

AB - In real-world classification tasks, deep neural networks show innovative performance in various fields. However, traditional classification methods are constructed based on a set of predefined classes and force unknown classes that determine their categorization into one of the predefined classes. This problem is addressed by the research field known as open-set recognition. Existing open-set recognition methods claim that the unique features of unknowns cannot be maintained by using only the final features. In other words, the various feature extraction methods should be considered to effectively reflect the characteristics of unknowns. In this study, we propose an open-set recognition model equipped with multi-feature extraction for multivariate time series data. The results of experiments with various multivariate time series datasets indicate that the proposed method shows improved capability to detect unknown classes while maintaining good predictive performance.

KW - multichannel signal

KW - multisensor signal

KW - multivariate time series

KW - Open-set recognition

KW - reconstruction error

UR - http://www.scopus.com/inward/record.url?scp=85142833911&partnerID=8YFLogxK

U2 - 10.1109/ACCESS.2022.3222310

DO - 10.1109/ACCESS.2022.3222310

M3 - Article

AN - SCOPUS:85142833911

SN - 2169-3536

VL - 10

SP - 120063

EP - 120073

JO - IEEE Access

JF - IEEE Access

ER -

Multivariate Time Series Open-Set Recognition Using Multi-Feature Extraction and Reconstruction

Abstract

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this