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 | |
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 |