Fast Shapelets: A Scalable Algorithm for Discovering Time Series Shapelets

By Eamonn J. Keogh, Thanawin Rakthanmanon. SDM, 2013

431 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Economics and Econometrics, Signal Processing

View Paper: https://www.semanticscholar.org/paper/c50510c8ccf2a24e98db1ca3b1d34e41a759d286

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Anthony J. Bagnall
2. Jason Lines
3. Nick S. Jones
4. Ben D. Fulcher
5. Tim Oates
6. Zhiguang Wang
7. Jessica Lin
8. Yixin Chen

In Economics and Econometrics

1. Anthony J. Bagnall
2. Jason Lines
3. Rohit J. Kate
4. Jon Hills
5. Eamonn J. Keogh
6. Jessica Lin
7. Zhicheng Cui
8. Yixin Chen

In Signal Processing

1. Anthony J. Bagnall
2. Jason Lines
3. Rohit J. Kate
4. Tim Oates
5. Zhiguang Wang
6. Ben D. Fulcher
7. Nick S. Jones
8. Eamonn J. Keogh

Citations in the corpus

• Matrix Profile I: All Pairs Similarity Joins for Time Series: A Unifying View That Includes Motifs, Discords and Shapelets (2016, methodology): “Shapelets can be used to classify time series (essentially, the nearest shapelet algorithm), offering the benefits of speed, intuitiveness and at least on some domains, significantly improved accuracy [23].”
• Classification of time series by shapelet transformation (2013, methodology): “The set W has |W| =∑maxl =min n(m− l+1) candidate shapelets.”
• Highly Comparative Feature-Based Time-Series Classification (2014, methodology): “g a large database of time-series features. We note that not all methods fit neatly into these two categories of instance-based and feature-based classification. For example, time-series shapelets [8], [9] classify new time series according to the minimum distance of particular time-series subsequence”
• TapNet: Multivariate Time Series Classification with Attentional Prototypical Network (2020, methodology): “For time series shapelet-based approaches (Rakthanmanon and Keogh 2013)(Wang et al. 2016)(Ye and Keogh 2009)(Fang, Wang, and Wang 2018), the models identify subsequences that are the most discriminating of the two classes.”
• Multi-Scale Convolutional Neural Networks for Time Series Classification (2016, methodology): “ existing methods with state-of-the-art results published within the recent three years, including: DTW with a warping window constraint set through cross validation (DTW CV) [23], Fast Shapelet (FS) [24] , SAX with vector space model (SV) [29], Bag-of-SFA-Symbols (BOSS) [26], Shotgun Classier (SC) ”
• Encoding Time Series as Images for Visual Inspection and Classification Using Tiled Convolutional Neural Networks (2014, methodology): “…competing methods: two state-of-the-art 1NN classifiers based on Euclidean distance and DTW, the recently proposed Fast-Shapelets based classifier (Rakthanmanon and Keogh 2013), the classifier based on Bag-of-Patterns (BoP) (Lin, Khade, and Li 2012; Oates et al. 2012) and the most recent SAX-VSM…”
• Imaging Time-Series to Improve Classification and Imputation (2015, result): “We compare the classification error rate of our GASF-GADF-MTF approach with previously published results of 3 competing methods and 6 best approaches proposed recently: early state-of-the-art 1NN classifiers based on Euclidean distance and DTW (Best Warping Window and no Warping Window), Fast-Shapel”
• catch22: CAnonical Time-series CHaracteristics (2019, methodology): “%) than by all but two existing methods (BOSS [25] and Fast Shapelets [24]) with a mean and standard deviation over all other classifiers of 69.0 ± 18.7% (DTW-1NN 65%, Euc-1NN 53.9%).”