A Hybrid Time Series Matching Algorithm Based on Feature-Points and DTW
作者： Xi Wang , Mingxing Jiang , Sheng Chen , Chao Yang , Wei Jing
关键词: Mathematics 、 Pattern recognition 、 Overhead (computing) 、 Approximation algorithm 、 Matching (graph theory) 、 Euclidean distance 、 Blossom algorithm 、 Dynamic time warping 、 Series (mathematics) 、 Artificial intelligence 、 Similarity (geometry)
摘要: Feature-points based time series approximation representation utilizes the tendency information, but lacks consideration of details. DTW (Dynamic Time Warping) similarity measurement eliminates line warp, computation complexity is high. Based on feature-points and DTW, this paper proposed a hybrid matching algorithm. Firstly, we extracted as coarse-grained representation, calculated distance between feature-points, then applied uniform sampling segmentations fine-grained Euclidean corresponding segmentations, at last, summed two distances final distance. The algorithm achieved high accuracy while lowered overhead. This used several data sets from UCR to do experiments, verified effectiveness
sci-hub.se 参考文章(13)
Michael J. Pazzani, Eamonn J. Keogh, Derivative Dynamic Time Warping. siam international conference on data mining. pp. 1- 11 ,(2001)
Jędrzej S. Bojanowski, Anton Vrieling, Andrew K. Skidmore, A comparison of data sources for creating a long-term time series of daily gridded solar radiation for Europe Solar Energy. ,vol. 99, pp. 152- 171 ,(2014) , 10.1016/J.SOLENER.2013.11.007
Gautam Mantena, Sivanand Achanta, Kishore Prahallad, Query-by-example spoken term detection using frequency domain linear prediction and non-segmental dynamic time warping IEEE Transactions on Audio, Speech, and Language Processing. ,vol. 22, pp. 946- 955 ,(2014) , 10.1109/TASLP.2014.2311322
Jian Pei, State of the Journal Editorial IEEE Transactions on Knowledge and Data Engineering. ,vol. 28, pp. 1- 1 ,(2016) , 10.1109/TKDE.2015.2500298
Mustafa Gokce Baydogan, George Runger, Learning a symbolic representation for multivariate time series classification Data Mining and Knowledge Discovery. ,vol. 29, pp. 400- 422 ,(2015) , 10.1007/S10618-014-0349-Y
David Reid, Abir Jaafar Hussain, Hissam Tawfik, Financial Time Series Prediction Using Spiking Neural Networks PLoS ONE. ,vol. 9, pp. e103656- ,(2014) , 10.1371/JOURNAL.PONE.0103656
Philippe Esling, Carlos Agon, Time-series data mining ACM Computing Surveys. ,vol. 45, pp. 1- 34 ,(2012) , 10.1145/2379776.2379788
Garima, Sangeeta Rani, Review on time series databases and recent research trends in Time Series Mining 2014 5th International Conference - Confluence The Next Generation Information Technology Summit (Confluence). pp. 109- 115 ,(2014) , 10.1109/CONFLUENCE.2014.6949290
Qing Xie, Chaoyi Pang, Xiaofang Zhou, Xiangliang Zhang, Ke Deng, Maximum error-bounded Piecewise Linear Representation for online stream approximation very large data bases. ,vol. 23, pp. 915- 937 ,(2014) , 10.1007/S00778-014-0355-0
Jiaping Zhao, Laurent Itti, Classifying Time Series Using Local Descriptors with Hybrid Sampling IEEE Transactions on Knowledge and Data Engineering. ,vol. 28, pp. 623- 637 ,(2016) , 10.1109/TKDE.2015.2492558