Using Variational Autoencoder to augment Sparse Time series Datasets

In machine learning, data augmentation is called the process of generating synthetic samples in order to augment sparse training datasets. Reducing the error-rate of classifiers is the main motivation. In this paper, we generate synthetic training samples of time series data using a simple implementation of the Variational Autoencoder, to test whether classification performance increases when augmenting the original training sets with manifolds of generated samples. We demonstrate the effectiveness of data augmentation using the Variational Autoencoder as a generative model, by conducting experiments with different standard classifiers evaluated on nine datasets from the UCR Time Series Classification Archive. We show that our method is beneficial in most cases, as we observed an increase of accuracy and F1-Score on all datasets.