Time Series Anomaly

Time Series Anomaly

Time series data is ubiquitous in areas involving dynamic systems, i.e. anything whose behaviour can be described as a function of time. Such systems can include everyday machines like cars, aeroplanes or trains, but also more abstract systems like chemical processes, the stock market, or the human cardiovascular system. In a world increasingly reliant on data-driven decision-making, it is imperative to ensure any of the aforementioned systems are running as expected, which is where time series anomaly detection comes into play. It is a research area concerned with automatically detecting anomalous behaviour in time series data, ideally without the need of examples of anomalies and considering correlations between different system properties.

This research project is supported by Mercedes-Benz AG, where resulting advances are primarily applied to powertrain test benches.

People involved:


Lucas Correia
Lucas Correia
Prof. Thomas Bäck
Professor of Natural Computing
Anna V. Kononova
Assistant Professor of Efficient Heuristic Optimization