Metric time series classification using dynamic warping relative to centroids of classes

The paper discusses the problem of time series classification in the case of several classes. The proposed classification model uses the matrix of distance between time series. This distance measure is defined by the dynamic time warping method. The dimension of the distance matrix is very high. The paper introduces centroids of each class as reference objects used to decrease this dimension. The distance matrix with lower dimension describes the distance between all objects and reference objects. This method is used for human activity recognition. The quality of classification on data from a mobile accelerometer is investigated. This metric algorithm of classification is compared with the separating classification algorithm.