State Discovery and Prediction from Multivariate Sensor Data

Olli Pekka Rinta-Koski; Miki Sirola; Le Ngu Nguyen; Jaakko Hollmén

doi:10.1007/978-3-030-91445-5_10

State Discovery and Prediction from Multivariate Sensor Data

Olli Pekka Rinta-Koski, Miki Sirola, Le Ngu Nguyen, Jaakko Hollmén^*

^*Corresponding author for this work

Stockholm University

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

Abstract

The advent of cloud computing and autonomous data centers operating fully without human supervision has highlighted the need for fault-tolerant architectures and intelligent software tools for system parameter optimization. Demands on computational throughput have to be balanced with environmental concerns, such as energy consumption and waste heat. Using multivariate time series data collected from an experimental data center, we build a state model using clustering, then estimate the states represented by the clusters using both a hidden Markov model and a long-short term memory neural net. Knowledge of future states of the system can be used to solve tasks such as reduced energy consumption and optimized resource allocation in the data center.

Original language	English
Title of host publication	Advanced Analytics and Learning on Temporal Data - 6th ECML PKDD Workshop, AALTD 2021, Revised Selected Papers
Editors	Vincent Lemaire, Simon Malinowski, Anthony Bagnall, Thomas Guyet, Romain Tavenard, Georgiana Ifrim
Publisher	Springer
Pages	155-169
Number of pages	15
ISBN (Print)	9783030914448
DOIs	https://doi.org/10.1007/978-3-030-91445-5_10
Publication status	Published - 2021
MoE publication type	A4 Conference publication
Event	International Workshop on Advanced Analytics and Learning on Temporal Data - Virtual, Online Duration: 13 Sept 2021 → 17 Sept 2021 Conference number: 6

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Publisher	Springer
Volume	13114 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Workshop

Workshop	International Workshop on Advanced Analytics and Learning on Temporal Data
Abbreviated title	AALTD
City	Virtual, Online
Period	13/09/2021 → 17/09/2021

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1007/978-3-030-91445-5_10

Cite this

Rinta-Koski, O. P., Sirola, M., Nguyen, L. N., & Hollmén, J. (2021). State Discovery and Prediction from Multivariate Sensor Data. In V. Lemaire, S. Malinowski, A. Bagnall, T. Guyet, R. Tavenard, & G. Ifrim (Eds.), Advanced Analytics and Learning on Temporal Data - 6th ECML PKDD Workshop, AALTD 2021, Revised Selected Papers (pp. 155-169). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13114 LNAI). Springer. https://doi.org/10.1007/978-3-030-91445-5_10

Rinta-Koski, Olli Pekka ; Sirola, Miki ; Nguyen, Le Ngu et al. / State Discovery and Prediction from Multivariate Sensor Data. Advanced Analytics and Learning on Temporal Data - 6th ECML PKDD Workshop, AALTD 2021, Revised Selected Papers. editor / Vincent Lemaire ; Simon Malinowski ; Anthony Bagnall ; Thomas Guyet ; Romain Tavenard ; Georgiana Ifrim. Springer, 2021. pp. 155-169 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{c9c1e79cc6f748dfb7e5127b658e7679,

title = "State Discovery and Prediction from Multivariate Sensor Data",

abstract = "The advent of cloud computing and autonomous data centers operating fully without human supervision has highlighted the need for fault-tolerant architectures and intelligent software tools for system parameter optimization. Demands on computational throughput have to be balanced with environmental concerns, such as energy consumption and waste heat. Using multivariate time series data collected from an experimental data center, we build a state model using clustering, then estimate the states represented by the clusters using both a hidden Markov model and a long-short term memory neural net. Knowledge of future states of the system can be used to solve tasks such as reduced energy consumption and optimized resource allocation in the data center.",

author = "Rinta-Koski, {Olli Pekka} and Miki Sirola and Nguyen, {Le Ngu} and Jaakko Hollm{\'e}n",

note = "Publisher Copyright: {\textcopyright} 2021, Springer Nature Switzerland AG.; International Workshop on Advanced Analytics and Learning on Temporal Data, AALTD ; Conference date: 13-09-2021 Through 17-09-2021",

year = "2021",

doi = "10.1007/978-3-030-91445-5_10",

language = "English",

isbn = "9783030914448",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer",

pages = "155--169",

editor = "Vincent Lemaire and Simon Malinowski and Anthony Bagnall and Thomas Guyet and Romain Tavenard and Georgiana Ifrim",

booktitle = "Advanced Analytics and Learning on Temporal Data - 6th ECML PKDD Workshop, AALTD 2021, Revised Selected Papers",

address = "Germany",

}

Rinta-Koski, OP, Sirola, M, Nguyen, LN & Hollmén, J 2021, State Discovery and Prediction from Multivariate Sensor Data. in V Lemaire, S Malinowski, A Bagnall, T Guyet, R Tavenard & G Ifrim (eds), Advanced Analytics and Learning on Temporal Data - 6th ECML PKDD Workshop, AALTD 2021, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 13114 LNAI, Springer, pp. 155-169, International Workshop on Advanced Analytics and Learning on Temporal Data, Virtual, Online, 13/09/2021. https://doi.org/10.1007/978-3-030-91445-5_10

State Discovery and Prediction from Multivariate Sensor Data. / Rinta-Koski, Olli Pekka; Sirola, Miki; Nguyen, Le Ngu et al.
Advanced Analytics and Learning on Temporal Data - 6th ECML PKDD Workshop, AALTD 2021, Revised Selected Papers. ed. / Vincent Lemaire; Simon Malinowski; Anthony Bagnall; Thomas Guyet; Romain Tavenard; Georgiana Ifrim. Springer, 2021. p. 155-169 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13114 LNAI).

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

TY - GEN

T1 - State Discovery and Prediction from Multivariate Sensor Data

AU - Rinta-Koski, Olli Pekka

AU - Sirola, Miki

AU - Nguyen, Le Ngu

AU - Hollmén, Jaakko

N1 - Conference code: 6

PY - 2021

Y1 - 2021

N2 - The advent of cloud computing and autonomous data centers operating fully without human supervision has highlighted the need for fault-tolerant architectures and intelligent software tools for system parameter optimization. Demands on computational throughput have to be balanced with environmental concerns, such as energy consumption and waste heat. Using multivariate time series data collected from an experimental data center, we build a state model using clustering, then estimate the states represented by the clusters using both a hidden Markov model and a long-short term memory neural net. Knowledge of future states of the system can be used to solve tasks such as reduced energy consumption and optimized resource allocation in the data center.

AB - The advent of cloud computing and autonomous data centers operating fully without human supervision has highlighted the need for fault-tolerant architectures and intelligent software tools for system parameter optimization. Demands on computational throughput have to be balanced with environmental concerns, such as energy consumption and waste heat. Using multivariate time series data collected from an experimental data center, we build a state model using clustering, then estimate the states represented by the clusters using both a hidden Markov model and a long-short term memory neural net. Knowledge of future states of the system can be used to solve tasks such as reduced energy consumption and optimized resource allocation in the data center.

UR - http://www.scopus.com/inward/record.url?scp=85121914417&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-91445-5_10

DO - 10.1007/978-3-030-91445-5_10

M3 - Conference article in proceedings

AN - SCOPUS:85121914417

SN - 9783030914448

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 155

EP - 169

BT - Advanced Analytics and Learning on Temporal Data - 6th ECML PKDD Workshop, AALTD 2021, Revised Selected Papers

A2 - Lemaire, Vincent

A2 - Malinowski, Simon

A2 - Bagnall, Anthony

A2 - Guyet, Thomas

A2 - Tavenard, Romain

A2 - Ifrim, Georgiana

PB - Springer

T2 - International Workshop on Advanced Analytics and Learning on Temporal Data

Y2 - 13 September 2021 through 17 September 2021

ER -

Rinta-Koski OP, Sirola M, Nguyen LN, Hollmén J. State Discovery and Prediction from Multivariate Sensor Data. In Lemaire V, Malinowski S, Bagnall A, Guyet T, Tavenard R, Ifrim G, editors, Advanced Analytics and Learning on Temporal Data - 6th ECML PKDD Workshop, AALTD 2021, Revised Selected Papers. Springer. 2021. p. 155-169. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-91445-5_10

State Discovery and Prediction from Multivariate Sensor Data

Abstract

Publication series

Workshop

UN SDGs

Access to Document

Other files and links

Fingerprint

Science-IT

Cite this

State Discovery and Prediction from Multivariate Sensor Data

Abstract

Publication series

Workshop

UN SDGs

Access to Document

Other files and links

Fingerprint

Equipment

Science-IT

Cite this