Time Series Classification Using Locality Preserving Projections
Xiaoqing Weng, Junyi Shen · 2007
The time series is generally of high dimensionality and classifying in such a high dimensional space is often infeasible due to the curse of dimensionality. We propose a new time series classifying method, which aims to classify the time series into different classes. By using locality preserving projections (LPP), the time series can be projected into a lower-dimensional space in which the time series related to the same class are close to each other, the time series in testing set can be identified by one-nearest-neighbor classifier in the lower-dimensional space. Extensive experimental evaluations are performed on 20 time series datasets, which come from diverse fields, including medicine, biometrics, astronomy and industry. The experiment results demonstrate the effectiveness of our approach.