Making Time-Series Classification More Accurate Using Learned Constraints

By Chotirat (Ann) Ratanamahatana, Eamonn J. Keogh. SDM, 2004

473 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Signal Processing

View Paper: https://www.semanticscholar.org/paper/8e2ecadc4f7dd691e89ea2c245cc0b83b043f9f1

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

In Signal Processing

Citations in the corpus

Toward accurate dynamic time warping in linear time and space (2007, unknown): “In fact, for many applications, using constraints leads to better classification accuracy than unconstrained DTW [17][18].”
Robust and fast similarity search for moving object trajectories (2005, background): “In order to make a fair comparison with DTW [29], we also test DTW with different warping lengths and report the best results.”
A time series forest for classification and feature extraction (2013, methodology): “Table 2: The error rates of TSF using the splitting criterion: Entrance gain (TSF) or entropy gain (TSF-entropy), random forest with 500 trees applied to the interval features with sizes power of two (interRF), 1-NN with Euclidean distance (NNEuclidean), 1-NN with the best warping window DTW (DTWBes”
k-Shape: Efficient and Accurate Clustering of Time Series (2016, background): “The ubiquity of time series has generated a substantial interest in querying [2, 38, 39, 41, 52, 61, 65], indexing [8, 11, 34, 35, 37, 63], classification [30, 47, 58, 70], clustering [36, 45, 54, 69, 71], and modeling [3, 31, 68] of such data.”
A review on time series data mining (2011, background): “Ratanamahatana and Keogh (2004) propose to learn arbitrary constraints on the warping path.”
Using dynamic time warping distances as features for improved time series classification (2016, background): “This window constraint not only speeds up computation of DTW, but somewhat non-intuitively, also improves the accuracy of time series classification (Ratanamahatana and Keogh 2004a).”
A Bag-of-Features Framework to Classify Time Series (2013, methodology): “For time series, one nearest neighbor (NN) classifiers with euclidean (NNEuclidean), or a dynamic time warping distance (NNDTW) have been widely and successfully used [1], [2], [3], [4], [5].”
Three Myths about Dynamic Time Warping Data Mining (2005, background): “For example, while there have been dozens of papers on speeding up DTW in the last decade, there has only been one on making it more accurate [ 15 ].”
Highly Comparative Feature-Based Time-Series Classification (2014, methodology): “Synthetic Control [19] 6 300 300 60 Gun point [20] 2 50 150 150 CBF [21] 3 30 900 128 Face (all) [22] 14 560 1690 131 OSU Leaf [23] 6 200 242 427”
Word spotting for historical documents (2007, background): “Recent work [23] shows that the size and shape of the global path constraint can be adapted, leading to faster DTW computations and better matching performance.”
A brief survey on sequence classification (2010, methodology): “Therefore, distance based methods are widely adopted to classify time series [61; 26; 59; 48].”
Weighted dynamic time warping for time series classification (2011, methodology): “For the detailed descriptions of the data sets, please see Ratanamahatana and Keogh [20].”
Time Warp Edit Distance with Stiffness Adjustment for Time Series Matching (2007, methodology): “For ERP, OTWED and OTWED1 we used the L1-norm, while the L2-norm has been implemented in DTW and ODTW [9].”
Rotation-invariant similarity in time series using bag-of-patterns representation (2012, background): “Such a problem can generally be handled by elastic distance measures such as Dynamic Time Warping (DTW) (Keogh 2002; Ratanamahatana and Keogh 2004; Sart et al. 2010; Ko et al. 2005).”
Towards parameter-free data mining (2004, methodology): “Here, the Euclidean distance requires that the dimensionality of two instances being compared is exactly the same, and Dynamic Time Warping (DTW) is not defined if a single data point is missing [30].”
Time-series data mining (2012, methodology): “It has been shown that these improve not only the speed but also the level of accuracy as it avoids the pathological matching introduced by extended warping [Ratanamahatana and Keogh 2004b].”
FastDTW: Toward Accurate Dynamic Time Warping in Linear Time and Space (2004, background): “Space complexity of storing the warp path = N 2 [12]”
k-Shape: Efficient and Accurate Clustering of Time Series (2015, unknown): “est in querying [2, 19, 45, 46, 48, 62, 74, 79], indexing [9, 13, 41, 42, 44, 77], classification [37, 56, 68, 88], clustering [43, 54, 64, 87, 89], and modeling [4, 38, 86] of such data.”