MiniRocket: a very fast (almost) deterministic transform for time series classification

Angus Dempster; Daniel F. Schmidt; Geoffrey I. Webb

doi:10.1145/3447548.3467231

MiniRocket: a very fast (almost) deterministic transform for time series classification

Angus Dempster, Daniel F. Schmidt, Geoffrey I. Webb

Department of Data Science & AI

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

400 Citations (Scopus)

Abstract

Rocket achieves state-of-the-art accuracy for time series classification with a fraction of the computational expense of most existing methods by transforming input time series using random convolutional kernels, and using the transformed features to train a linear classifier. We reformulate Rocket into a new method, MiniRocket. MiniRocket is up to 75 times faster than Rocket on larger datasets, and almost deterministic (and optionally, fully deterministic), while maintaining essentially the same accuracy. Using this method, it is possible to train and test a classifier on all of 109 datasets from the UCR archive to state-of-the-art accuracy in under 10 minutes. MiniRocket is significantly faster than any other method of comparable accuracy (including Rocket), and significantly more accurate than any other method of remotely similar computational expense.

Original language	English
Title of host publication	KDD'21 - Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining
Editors	Jiliang Tang, Tyler Derr
Place of Publication	New York NY USA
Publisher	Association for Computing Machinery (ACM)
Pages	248-257
Number of pages	10
ISBN (Electronic)	9781450383325
DOIs	https://doi.org/10.1145/3447548.3467231
Publication status	Published - 2021
Event	ACM International Conference on Knowledge Discovery and Data Mining 2021 - Virtual, Online, Singapore Duration: 14 Aug 2021 → 18 Aug 2021 Conference number: 27th https://dl.acm.org/doi/proceedings/10.1145/3447548 (Proceedings)

Conference

Conference	ACM International Conference on Knowledge Discovery and Data Mining 2021
Abbreviated title	KDD 2021
Country/Territory	Singapore
City	Virtual, Online
Period	14/08/21 → 18/08/21
Internet address	https://dl.acm.org/doi/proceedings/10.1145/3447548 (Proceedings)

Keywords

convolution
scalable
time series classification
transform

Access to Document

10.1145/3447548.3467231

Cite this

@inproceedings{69705b6b06df421da8a533f5414c490f,

title = "MiniRocket: a very fast (almost) deterministic transform for time series classification",

abstract = "Rocket achieves state-of-the-art accuracy for time series classification with a fraction of the computational expense of most existing methods by transforming input time series using random convolutional kernels, and using the transformed features to train a linear classifier. We reformulate Rocket into a new method, MiniRocket. MiniRocket is up to 75 times faster than Rocket on larger datasets, and almost deterministic (and optionally, fully deterministic), while maintaining essentially the same accuracy. Using this method, it is possible to train and test a classifier on all of 109 datasets from the UCR archive to state-of-the-art accuracy in under 10 minutes. MiniRocket is significantly faster than any other method of comparable accuracy (including Rocket), and significantly more accurate than any other method of remotely similar computational expense.",

keywords = "convolution, scalable, time series classification, transform",

author = "Angus Dempster and Schmidt, \{Daniel F.\} and Webb, \{Geoffrey I.\}",

note = "Funding Information: This material is based on work supported by an Australian Government Research Training Program Scholarship, and the Australian Research Council under award DP190100017. The authors would like to thank Professor Eamonn Keogh and all the people who have contributed to the UCR time series classification archive. Figures showing mean ranks were produced using code from [15]. Publisher Copyright: {\textcopyright} 2021 ACM. Copyright: Copyright 2021 Elsevier B.V., All rights reserved.; ACM International Conference on Knowledge Discovery and Data Mining 2021, KDD 2021 ; Conference date: 14-08-2021 Through 18-08-2021",

year = "2021",

doi = "10.1145/3447548.3467231",

language = "English",

pages = "248--257",

editor = "Jiliang Tang and Tyler Derr",

booktitle = "KDD'21 - Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining",

publisher = "Association for Computing Machinery (ACM)",

address = "United States of America",

url = "https://dl.acm.org/doi/proceedings/10.1145/3447548",

}

Dempster, A , Schmidt, DF & Webb, GI 2021, MiniRocket: a very fast (almost) deterministic transform for time series classification. in J Tang & T Derr (eds), KDD'21 - Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. Association for Computing Machinery (ACM), New York NY USA, pp. 248-257, ACM International Conference on Knowledge Discovery and Data Mining 2021, Virtual, Online, Singapore, 14/08/21. https://doi.org/10.1145/3447548.3467231

MiniRocket: a very fast (almost) deterministic transform for time series classification. / Dempster, Angus ; Schmidt, Daniel F.; Webb, Geoffrey I.
KDD'21 - Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. ed. / Jiliang Tang; Tyler Derr. New York NY USA: Association for Computing Machinery (ACM), 2021. p. 248-257.

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

TY - GEN

T1 - MiniRocket

T2 - ACM International Conference on Knowledge Discovery and Data Mining 2021

AU - Dempster, Angus

AU - Schmidt, Daniel F.

AU - Webb, Geoffrey I.

N1 - Conference code: 27th

PY - 2021

Y1 - 2021

N2 - Rocket achieves state-of-the-art accuracy for time series classification with a fraction of the computational expense of most existing methods by transforming input time series using random convolutional kernels, and using the transformed features to train a linear classifier. We reformulate Rocket into a new method, MiniRocket. MiniRocket is up to 75 times faster than Rocket on larger datasets, and almost deterministic (and optionally, fully deterministic), while maintaining essentially the same accuracy. Using this method, it is possible to train and test a classifier on all of 109 datasets from the UCR archive to state-of-the-art accuracy in under 10 minutes. MiniRocket is significantly faster than any other method of comparable accuracy (including Rocket), and significantly more accurate than any other method of remotely similar computational expense.

AB - Rocket achieves state-of-the-art accuracy for time series classification with a fraction of the computational expense of most existing methods by transforming input time series using random convolutional kernels, and using the transformed features to train a linear classifier. We reformulate Rocket into a new method, MiniRocket. MiniRocket is up to 75 times faster than Rocket on larger datasets, and almost deterministic (and optionally, fully deterministic), while maintaining essentially the same accuracy. Using this method, it is possible to train and test a classifier on all of 109 datasets from the UCR archive to state-of-the-art accuracy in under 10 minutes. MiniRocket is significantly faster than any other method of comparable accuracy (including Rocket), and significantly more accurate than any other method of remotely similar computational expense.

KW - convolution

KW - scalable

KW - time series classification

KW - transform

UR - https://www.scopus.com/pages/publications/85114944594

U2 - 10.1145/3447548.3467231

DO - 10.1145/3447548.3467231

M3 - Conference Paper

AN - SCOPUS:85114944594

SP - 248

EP - 257

BT - KDD'21 - Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining

A2 - Tang, Jiliang

A2 - Derr, Tyler

PB - Association for Computing Machinery (ACM)

CY - New York NY USA

Y2 - 14 August 2021 through 18 August 2021

ER -

MiniRocket: a very fast (almost) deterministic transform for time series classification

Abstract

Conference

Keywords

Access to Document

Other files and links

Target-agnostic analytics: Building agile predictive models for big data

Cite this

MiniRocket: a very fast (almost) deterministic transform for time series classification

Abstract

Conference

Keywords

Access to Document

Other files and links

Projects

Target-agnostic analytics: Building agile predictive models for big data

Cite this