News

The project principal investigator Žarko Kovač participated in the Workshop on Machine Learning for Time Series Data which was held in New York from 28^th till 30^th April 2026 at the Simons Foundation. The workshop was organized as an activity of the The Simons Collaboration on Computational Biogeochemical Modeling of Marine Ecosystems.

This three day workshop introduced machine learning approaches for analyzing ecological and biogeochemical time series data from complex marine systems. The meeting bridged traditional statistical time series methods and mechanistic process based models, highlighting how modern tools such as vector autoregressive models, artificial neural networks, physics guided autoregressive models, and physics informed neural networks could be used together. Through a mix of lectures, hands-on tutorials, and collaborative hackathon sessions, participants learned how to select, implement, and interpret appropriate methods using real-world datasets, while building reproducible workflows and fostering cross-disciplinary collaboration. In the framework of the PHOTOCLIM project such methods are used to gap fill in situ time series data, and are a usefull tool for remote sensing data as well.

More infromation on the workshop can be found here.