White Noise Windows: Data Augmentation for Time Series

Data Augmentation can be used to enlarge small datasets with synthetic samples to improve the performance of classifiers. In this paper, we present a new data augmentation method for time series that we name White Noise Windows. White Noise Windows works by multiplying each time series in a dataset with a predefined number of identical time series and thereafter replacing randomly chosen parts of the resulting time series with windows of white noise. We demonstrate the effectiveness of the proposed method by conducting experiments using two different classifiers evaluated on nine well-known datasets from the UCR Time Series Classification Archive. We show that our method can improve the performance of the classification methods and helps to avoid overfitting.