The project aims to develop methods that can handle time series in unsupervised and supervised scenarios. Time series are often treated as vectorial data by data analysts and scientists. Such a view discards the potentially interesting and latent dynamics that a time series might possess. Our aim is to explicitly capture these important characteristics and incorporate them in machine learning algorithms. In particular, we are interest in devising new methods for dimensionality reduction and classification of time series.
People involved in this project:
Nikos Gianniotis, Nikos.Gianniotis@h-its.org
Dennis Kügler, Dennis.Kuegler@h-its.org