Join us for this special Tree-Ring talk by Dr. Valerie Livina on Data science techniques for real-world complex systems in Bannister 110 or via zoom.
Abstract:
In the recent years, data science techniques of tipping point analysis became popular in diverse applications, from ecology to structure health monitoring. These techniques help anticipate, detect and forecast critical transitions in dynamical systems. The methodology combines monitoring short- and long-term memory in time series with potential analysis that analyses and extrapolates the system states. For anticipating tipping points, early warning signal (EWS) indicators utilise dynamically derived lag-1 autocorrelation, power-law scaling exponent of Detrended Fluctuation Analysis, and the power-spectrum-based EWS indicator. Applications of these techniques in engineering broaden the range of the tipping point analysis in real-world complex systems and provide a methodology for data-driven failure analysis and prevention.
Data science techniques also help understand data and identify the artefacts of data processing. This is particularly relevant to measurements with large uncertainties, poor statistics, and/or imprecise time allocation, which may be the case in dendrochronology. Recently, the technique of Dynamic Time Warping was proposed for assessment and comparison of various chronologies using the global radiocarbon events [Panyushkina et al 2022]. Machine learning and AI are promising in analysing dendrochronologies, and the ongoing collaboration between LTTR and NPL will explore various methods in this area.
Email talk organizers at talks@ltrr.arizona.edu for Zoom link (or visit Bannister Building Room 110 for live presentation).