Christian Schroth’s master thesis titled “Robust Bayesian Cluster Analysis” has been awarded the 2020 Study Prize by the SEW-Eurodrive-Foundation. The master thesis was supervised by emergenCITY-Dr. Michael Muma.
Christian Schroth’s work makes an important contribution to robust cluster analysis. In his work, he developed highly innovative statistical learning techniques to identify homogeneous subsets (clusters) in a data set. While there are a number of established methods for this purpose, they quickly lose their reliability or break down altogether if the data are outlier-ridden, or if the data contain non-symmetric scatter around the cluster centers. In his work, Christian Schroth now demonstrates a new and coherent mathematical formulation for robust cluster analysis, which he derived using a Bayesian approach to determine the a posteriori most likely statistical cluster model for the data. He supported his theoretical work with numerous simulations and real data examples, such as the analysis of a radar data set measured at the Department of Signal Processing of the TU Darmstadt as part of a study on radar-based gait analysis.
The SEW-Eurodrive award winners will be recognized during the virtual Ernst Blickle Lecture and Study Award Ceremony. Congratulations to the winner!