Progressive visual analytics of collective behaviour data

Finding interesting patterns in large data sets remains a challenge for collective behaviour. The machine learning (ML) methods commonly used to explore such data are often complicated by numerous input parameters, which makes for long runtimes. Computer scientists in this project will develop ML methods that enable biologists and psychologists to interactively explore their large datasets, enabling a deeper understanding of their results.

The project focuses on the combination of automated analysis methods and ML with interactive visualization methods to support the progressive analysis of spatio-temporal (movement) and dynamic graph data. The main goal is to complement automated algorithms progressively with human background knowledge to tune and verify machine learning models. The project team is currently developing methods for the visual analysis of dynamic networks, which are an important data type for the study of collectives.