• DocumentCode
    1705024
  • Title

    Effectiveness of similarity measures in classification of time series data with intrinsic and extrinsic variability

  • Author

    Sengupta, Sabyasachi ; Ojha, Piyush ; Hui Wang ; Blackburn, William

  • Author_Institution
    Sch. of Comput. & Math., Univ. of Ulster, Newtownabbey, UK
  • fYear
    2012
  • Firstpage
    166
  • Lastpage
    171
  • Abstract
    Time series are hard to analyse because of their intrinsic variability which arises from the stochastic nature of the underlying process. Analysis is harder still if the underlying process is non-stationary. Further extrinsic variation may be imposed by the variability of the sampling process, e.g. by sampling at different or non-uniform time intervals. We explore the efficacy of some common distance/similarity measures - Euclidean (EUC), Neighbourhood Counting Metric (NCM), Dynamic Time Warping (DTW), Longest Common Subsequence (LCS) and All Common Subsequences (ACS) - in a nearest neighbour classifier for classifying time series data with and without extrinsic variability. An artificial dataset containing trajectories of a 2-dimensional dynamical system and a real dataset, the Australian Sign Language Dataset (AUSLAN), are explored.
  • Keywords
    image classification; sign language recognition; time series; 2-dimensional dynamical system; ACS; AUSLAN; Australian sign language dataset; DTW; EUC; Euclidean measure; LCS; NCM; all common subsequences; dynamic time warping; extrinsic variability; intrinsic variability; longest common subsequence; nearest neighbour classifier; neighbourhood counting metric; sampling process variability; similarity measures; time series data classification; Euclidean distance; Noise; Speech recognition; Spirals; Time measurement; Time series analysis; Trajectory; extrinsic variations; intrinsic variations; similarity measures;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Cybernetic Intelligent Systems (CIS), 2012 IEEE 11th International Conference on
  • Conference_Location
    Limerick
  • Type

    conf

  • DOI
    10.1109/CIS.2013.6782171
  • Filename
    6782171