DTWNet: a Dynamic Time Warping Network

Dynamic Time Warping (DTW) is widely used as a similarity measure in various domains. Due to its invariance against warping in the time axis, DTW provides more meaningful discrepancy measurements between two signals than other dis- tance measures. In this paper, we propose a novel component in an artificial neural network. In contrast to the previous successful usage of DTW as a loss function, the proposed framework leverages DTW to obtain a better feature extraction. For the first time, the DTW loss is theoretically analyzed, and a stochastic backpropogation scheme is proposed to improve the accuracy and efficiency of the DTW learning. We also demonstrate that the proposed framework can be used as a data analysis tool to perform data decomposition.

[1]  Eamonn J. Keogh,et al.  Exact indexing of dynamic time warping , 2002, Knowledge and Information Systems.

[2]  Luca Antiga,et al.  Automatic differentiation in PyTorch , 2017 .

[3]  Juan Carlos Niebles,et al.  D3TW: Discriminative Differentiable Dynamic Time Warping for Weakly Supervised Action Alignment and Segmentation , 2019, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[4]  Venu Govindaraju,et al.  Direct Image Matching by Dynamic Warping , 2004, 2004 Conference on Computer Vision and Pattern Recognition Workshop.

[5]  Hongxia Jin,et al.  Accelerating Time Series Searching with Large Uniform Scaling , 2018, SDM.

[6]  Christos Faloutsos,et al.  Stream Monitoring under the Time Warping Distance , 2007, 2007 IEEE 23rd International Conference on Data Engineering.

[7]  Daniel Lemire,et al.  Faster retrieval with a two-pass dynamic-time-warping lower bound , 2008, Pattern Recognit..

[8]  Pierre Gançarski,et al.  Summarizing a set of time series by averaging: From Steiner sequence to compact multiple alignment , 2012, Theor. Comput. Sci..

[9]  M. Reinders,et al.  Multi-Dimensional Dynamic Time Warping for Gesture Recognition , 2007 .

[10]  Jie Zhang,et al.  PDP: parallel dynamic programming , 2017, IEEE CAA J. Autom. Sinica.

[11]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[12]  Micha Sharir,et al.  Dynamic Time Warping and Geometric Edit Distance: Breaking the Quadratic Barrier , 2016, ICALP.

[13]  Rohit J. Kate Using dynamic time warping distances as features for improved time series classification , 2016, Data Mining and Knowledge Discovery.

[14]  Jun Wang,et al.  Generalizing DTW to the multi-dimensional case requires an adaptive approach , 2016, Data Mining and Knowledge Discovery.

[15]  Marco Cuturi,et al.  Soft-DTW: a Differentiable Loss Function for Time-Series , 2017, ICML.

[16]  Robert Giegerich,et al.  GPU Parallelization of Algebraic Dynamic Programming , 2009, PPAM.

[17]  I. Elamvazuthi,et al.  Voice Recognition Algorithms using Mel Frequency Cepstral Coefficient (MFCC) and Dynamic Time Warping (DTW) Techniques , 2010, ArXiv.

[18]  Rafael Ramírez,et al.  Automatic assessment of violin performance using dynamic time warping classification , 2018, 2018 26th Signal Processing and Communications Applications Conference (SIU).

[19]  Liwei Wang,et al.  Gradient Descent Finds Global Minima of Deep Neural Networks , 2018, ICML.

[20]  Hossein Hamooni,et al.  Speeding up dynamic time warping distance for sparse time series data , 2017, Knowledge and Information Systems.

[21]  Donald J. Berndt,et al.  Using Dynamic Time Warping to Find Patterns in Time Series , 1994, KDD Workshop.

[22]  Gunasekaran Manogaran,et al.  Wearable sensor devices for early detection of Alzheimer disease using dynamic time warping algorithm , 2018, Cluster Computing.

[23]  Lars Schmidt-Thieme,et al.  Learning DTW-Shapelets for Time-Series Classification , 2016, CODS.

[24]  Yuan-Fang Wang,et al.  Learning a Mahalanobis Distance-Based Dynamic Time Warping Measure for Multivariate Time Series Classification , 2016, IEEE Transactions on Cybernetics.