Proceedings of the 2017 SIAM International Conference on Data Mining 2017
DOI: 10.1137/1.9781611974973.32
|View full text |Cite
|
Sign up to set email alerts
|

Indexing and classifying gigabytes of time series under time warping

Abstract: Time series classification maps time series to labels. The nearest neighbour algorithm (NN) using the Dynamic Time Warping (DTW) similarity measure is a leading algorithm for this task. NN compares each time series to be classified to every time series in the training database. With a training database of N time series of lengths L, each classification requires ϑ(N · L 2 ) computations. The databases used in almost all prior research have been relatively small (with less than 10, 000 samples) and much of the r… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
4
1

Citation Types

0
32
0

Year Published

2017
2017
2023
2023

Publication Types

Select...
5
2
1

Relationship

2
6

Authors

Journals

citations
Cited by 45 publications
(39 citation statements)
references
References 19 publications
0
32
0
Order By: Relevance
“…First, we assessed the scalability of TS-CHIEF w.r.t training set size. We used a Satellite Image Time Series (SITS) dataset [45] composed of 1 million time series of length 46, with 24 classes. We evaluated the accuracy and the total runtime as a function of the number of training time series, starting from a subsample of 58, and logarithmically increasing up to 131,879 (a sufficient quantity to clearly define the trend).…”
Section: Increasing Training Set Sizementioning
confidence: 99%
See 2 more Smart Citations
“…First, we assessed the scalability of TS-CHIEF w.r.t training set size. We used a Satellite Image Time Series (SITS) dataset [45] composed of 1 million time series of length 46, with 24 classes. We evaluated the accuracy and the total runtime as a function of the number of training time series, starting from a subsample of 58, and logarithmically increasing up to 131,879 (a sufficient quantity to clearly define the trend).…”
Section: Increasing Training Set Sizementioning
confidence: 99%
“…It is orders of magnitude faster than both FLAT-COTE and HIVE-COTE while attaining accuracy that ranks at least as well as them when assessed on the benchmark UCR archive, as illustrated in Figure 1. Figure 2 shows an experiment that demonstrates the scalability of TS-CHIEF using the Satellite Image Time Series (SITS) dataset [45]. It is 900x faster than HIVE-COTE for 1,500 time series (13 min versus 8 days).…”
Section: Introductionmentioning
confidence: 99%
See 1 more Smart Citation
“…Time Series Classification (TSC) problems are encountered in various real world data mining tasks ranging from health care [1]- [3] and security [4], [5] to food safety [6], [7] and power consumption monitoring [8], [9]. As deep learning models have revolutionized many machine learning fields such as computer vision [10] and natural language processing [11], [12], researchers recently started to adopt these models for TSC tasks [13].…”
Section: Introductionmentioning
confidence: 99%
“…Dynamic Time Warping (DTW) lower bounds play a key role in speeding up many forms of time series analytics [6,10,11,17]. Several lower bounds have been proposed [6-8, 15, 19].…”
Section: Introductionmentioning
confidence: 99%